Department of Energy Engineering, YearBook 2014 by Laboratory for Heat and Power

4 1 K OO

nt e rtm ing a p De ineer Eng

rgy e n of E

B R A YE

Department of Energy Engineering Laboratory for Heat and Power Laboratory for Hydraulic Turbomachinery Laboratory for Internal Combustion Engines and E-mobility YEARBOOK14

Title of publication Yearbook 2014, Department of Energy Engineering, Faculty of Mechanical Engineering, University of Ljubljana Title page & Design arnoldvuga+ Publisher ©University of Ljubljana, Faculty of Mechanical Engineering, Department of Energy Engineering Aškerčeva cesta 6, SI-1000 Ljubljana, Slovenia ISBN 978-961-6536-95-0 Editor Prof. Dr. Mihael Sekavčnik Print naknadno Edition 150 copies Price free copy Ljubljana, 2015

CIP zapis

Preface

The Department of Energy Engineering (DEE) is responsible for educational and R&D activities in the field of energy-conversion technologies and the transfer of leadingedge knowledge to a variety of study programs. Over the years the DEE has established strong links with industry in Slovenia and abroad, where it is widely recognized as a partner that provides vital information, fresh ideas, the latest research equipment and an understanding of the business environment to enable knowledge transfer and the development of industry. As a result of this, we have decided to promote an awareness of the scientific and industrial achievements of the DEEâ&#x20AC;&#x2122;s most recent successes by launching an active dissemination program. The Yearbook 2014 is part of this program. With it, we intend to highlight the current research projects being performed by senior researchers and PhD students. Each of these projects is briefly described, with the emphasis being on the expected or alreadyachieved results. The aim of this report is to attract your attention, and we are looking forward to sharing the details of particular projects with you and working with you on new ideas, proposals and projects. The Yearbook 2014 divides the project descriptions into three sections, one for each laboratory: the Laboratory for Heat and Power, the Laboratory of Hydraulic Machines and the Laboratory for Internal Combustion Engines and Electromobility. Many of the descriptions relate to industrial PhD projects, reflecting the strong industrial connections with the laboratories. In addition, we are committed to taking a pro-active role in new international research programs.

We hope that browsing through the Yearbook 2014 will inspire you with new ideas for your future research in collaboration with us. Head of Department of Energy Engineering Prof.Dr. Mihael SekavÄ?nik

Contents Laboratory for Heat and Power

Prof.Dr. Mihael Sekavčnik Assoc.Prof.Dr. Andrej Senegačnik

Life-cycle assessment of a hydrogen-based uninterruptible power supply system using renewable energy

Mitja Mori

FluMaBack - Improving the operation of Balance of Plant components

Mitja Mori

Hydrogen technologies in power supply from RES

Boštjan Drobnič

Intrusive Electrostatic Sensors Used On Lignite Fired Boiler

Igor Kuštrin

The integration of hydrogen technologies for utilisation of distributed renewable energy sources

Rok Lacko

CHP system based on HT PEMFC stack

Andrej Lotrič

Catalytic Biomass Gasification In Fluidized Bed With Superheated Steam

Jernej Mele, Ph.D.

Active Energy Network with Combined Heat and Power Systems and Hydrogen Technologies

Andrej Pirc

Laboratory for Hydraulic Turbomachinery

Prof.Dr. Branko Širok

Quantitative risk assessment on transmission network for natural gas

Tom Bajcar

Mineral wool melt fiberization on the spinner wheels

Benjamin Bizjan

Study of erosive cavitation detection in pump mode of pump- storage hydropower plant prototype

Tine Cencič

Direct measurements of thermal delay in cavitating flow for the European Space Agency (ESA)

Matevž Dular

Precision spraying in orchards and vineyards

Marko Hočevar

Investigation of vortex shedding from an airfoil by CFD simulation and computeraided flow visualization

Lovrenc Novak

Aerodynamics analysis of close range precision spraying

Aleš Malneršič

Advanced orchard spraying

Aljaž Osterman

Rotation generator of hydrodynamic cavitation

Martin Petkovšek

Hydrodynamic cavitation for wastewater treatment

Tadej Stepišnik Perdih

Kinematics of mineral wool fibers in air flow

Marko Peternelj

Laboratory for Internal Combustion Engines and e-Mobility

Assoc.Prof.Dr. Tomaž Katrašnik

Real World Emissions and Vehicle Parameters of Different Bus Powertrain Technologies

Samuel Rodman Oprešnik

Use of innovative lignocellulosic biofuels in gas turbines

Tine Seljak

Hybrid analytic-numerical 3D approach to modelling PEM fuel cells

Gregor Tavčar

Performance of a 6-cylinder turbo charged diesel engine running on tire pyrolysis oil

Rok Vihar

Optimization of the combustion process with innovative fuels and closed-loop combustion control

Urban Žvar Baškovič

100

Students’ Expert Activities

Ime Priimek

102

Short step into the history

Matija Tuma

104

Prof.Dr. Mihael Sekavčnik Assoc.Prof.Dr. Andrej Senegačnik

Laboratory for Heat and Power The Laboratory for Heat and Power’s activities cover two large, partially related, fields of study and research, while the latest developments in technology also drive its activities into other multidisciplinary fields. One of the basic research areas is energy-conversion systems, while the other has a closer focus on fuels and combustion. Recently, the research has spread to alternative, environmentally friendly, energy-conversion technologies as well as studies of the environmental impacts of energy engineering.

The primary activity of the Laboratory is education, both undergraduate and postgraduate. The Laboratory boasts modern equipment for its education and research purposes. Various power- and process-engineering-related models enable students to become familiar with many processes, machines and devices. In addition, state-ofthe-art software packages are available to the students for undergraduate studies as well as for postgraduate research. The lectures cover issues concerned with power engineering in utility and industrial power and thermal plants. They include gas and steam turbines, heat generators, fuel and combustion processes, utility and industrial energy systems, planning, operation, maintenance and the reliability of various power and thermal plants, etc. The theoretical lectures are upgraded with experimental work, especially during master’s studies.

All the members of the Laboratory for Heat and Power are also involved in research activities. These are normally closely connected with expert activities. Therefore, most of the problems being studied arise from practice. State-ofthe-art measuring equipment as well as professional and custom-developed software enable high-quality theoretical and experiential research activities. The experimental work takes place in the laboratory as well as in various utility and industrial plants. Both postgraduate and undergraduate students are also encouraged to cooperate in these research activities. Our research results have been published in many important international journals or were presented at conferences and seminars all over the world. Furthermore, many results find their way back to the application where they originated, and in this way lead to performance improvements. While the Laboratory for Heat and Power finds its direct cooperation with industrial partners very valuable, it is its role in larger national and international consortia working on various research projects that provides the opportunity for a further augmentation of our knowledge and experience.

Life-cycle assessment of a hydrogenbased uninterruptible power supply system using renewable energy ABstRACt → Purpose: The paper provides an empirical assessment of an uninterruptible power supply (UPS) system based on hydrogen technologies (HT-UPS) using renewable energy sources (RES) with regard to its environmental impacts and a comparison to a UPS system based on the internal combustion engine (ICE-UPS). Methods: For the assessment and comparison of the environmental impacts, the life-cycle assessment method (LCA) was applied, while numerical models for individual components of the UPS systems (electrolyser, storage tank, fuel cell, and ICE) were developed using GaBi software. The scope of analysis was cradle-to-end of utilization with functional unit 1 kWh of uninterrupted electricity produced. For the life-cycle inventory analysis, quantitative data was collected with on-site measurements on an experimental system, project documentation, GaBi software generic databases, and literature data. The CML 2001 method was applied to evaluate the system’s environmental impacts. Energy consumption of the manufacturing phase was estimated from gross value added (GVA) and the energy intensity of the industry sector in the manufacturer’s country. results and discussion: In terms of global warming (GW), acidification (A), abiotic depletion (AD) and eutrophication (E), manufacturing phase of HT-UPS accounts for more than 97% of environmental impacts.

Electrolyser in all its life cycle phases contributes above 50% of environmental impacts to the system’s GW, A and AD. Energy return on investment (EROI) for the HT-UPS has been calculated to be 0.143 with distinction between renewable (roughly 60%) and non-renewable energy resources inputs. HT-UPS’s life cycle GW emissions have been calculated to be 375 g of CO2 eq. per 1 kWh of uninterruptible electric energy supplied. All these values have also been calculated for the ICEUPS, and show that in terms of GW, A and AD, the ICEUPS has bigger environmental impacts and emits 1190 g of CO2 eq. per 1kWh of uninterruptible electric energy supplied. Both systems have similar operation phase energy efficiency. The ICE-UPS has a higher EROI, but uses almost none RES inputs. Conclusions: The comparison of two different technologies for providing uninterruptible power supply has shown that in all environmental impacts categories, except eutrophication, the HT-UPS is the sounder system. Most of HT-UPS’s environmental impacts result from the manufacturing phase. On the contrary, ICE-UPS system’s environmental impacts mainly result from operational phase. Efficiency of energy conversion from electricity to hydrogen to electricity again is rather low, as is energy return on investment (EROI), but these will likely improve as the technology matures.

IntroDUCtIon Mitja Mori Laboratory Laboratory for Heat and Power E-mail mitja.mori@fs.uni-lj.si Room S-I/60 Phone +386-1-4771-715 Status AsistantProfessor Research area Hydrogen technologies – Life cycle assessment

UPS systems based on hydrogen technologies (HT) are considered feasible to be integrated into advanced energy supply systems on large scale (Anderson and Leach 2004; Sherif et al. 2005; Cicconardi et al. 1997). The three subsystems: electrolyser, hydrogen storage tank and fuel cell have the ability to convert electric energy into hydrogen based chemical energy, store it, and convert it back into electricity when needed. If hydrogen is produced from renewable energy sources (RES), such a system emits very small amounts of CO2 in its operational phase. However, in other stages of its life cycle: raw materials production, manufacturing, transport and decomposition, it is expected to have certain environmental impacts. The observed system includes an electrolyser, H2 storage tank and a

LAborAtorY For HEAt AnD PowEr YEARBOOK 2014

The International Journal of Life Cycle assessment Received: 3 July 2013 /Accepted: 12 August 2014

fuel cell stack with relevant support and control equipment which enables simulations and the study of hydrogen technologies in advanced energy systems. The system is comprised of a process, support and control level (Mori et al. 2013). The scope of this study is limited to the operational level since the other two levels’ only task is the simulation of RES and enduser intermittency. The electrolyser and fuel cell are housed in two separate standard 20foot containers, along with the ex-

perimental equipment, while the storage tank is located outside. Only the container housing of the electrolyser is deemed necessary for the UPS’s function and therefore included in the life cycle inventory analysis (LCI). The electrical power for the electrolyser is supplied from the grid, but controlled in a manner that imitates the RES.

from manufacturing site to operation site, and its operation for 10 years. The same approach in LCA modelling was applied to a commercial ICE-UPS with a comparable power output. All the required quantities of used materials for the LCA numerical model (Fig. 2) were obtained from the manufacturer. It should be noted that the ICE-UPS can only cover the shortage in RES energy supply, while the HTUPS can also store excess RES energy in form of hydrogen.

Figure 1: Analysed system

LIFE CYCLE ASSESSMEnt

Figure 2: LCA model of HT-UPS system

The goal and scope define the exact questions that the analysis will try to answer, and set its spatial, temporal, and technological boundaries. The product’s function is defined and quantified by the functional unit – which in the case of this study is 1 kWh of produced uninterrupted electricity. The goal of the study was to determine environmental impacts of the HT-UPS and compare these impacts to those of the internal combustion engine ICE-UPS. Global warming (GW) and abiotic depletion (AD) as global, eutrophication (E) as local, and acidification (A) as regional environmental indicators were calculated and compared.

The electrolyser’s lifetime load factor (10 years) was estimated to be 0.5, in which availability of RES, hydrogen losses and maintenance time is considered. This load factor has been used in at least one previous scientific publication (Patyk et al. 2013), based solely on RES availability. The technical characteristics (Table 1) and the load factor result in 57,980 kg of H2 produced in 10 years. With H2 electrolysis efficiency of 0.56, 4.08∙106 kWh of electrical energy is needed. Hydrogen losses from electrolyser to the atmosphere because of system conditioning amount to 11,960 kg of H2 and contribute to environmental impacts in the UPS’s operational phase (Mori et al. 2013). Almost 106 kg of water is used for electrolysis in 10 years (Hydrogenics 2011).

Table 1: HT-UPS technical characteristics Type

Alkaline

Electrolyte

Water + 30 %m KOH

FUEL CELL

TANK

ELECTROLYSER

maximum power consumption 63 kW hydrogen production rate

15 Nm3/h / 1.34 kg/h

hydrogen output pressure

6 ÷ 25 bar

hydrogen purity (stated)

> 99.99%

specific energy consumption

6.5 kWh/Nm3

efficiency – overall

0.56

oxygen production rate (vol.)

50% H2 production

specific water consumption

< 1 l/Nm3 H2

housing

20-foot container

capacity

20 m3

maximum operation pressure

25 bar

type

PEM

nominal output power

6 kW

maximum output power

7.5 kW

required hydrogen purity

> 99.9%

hydrogen consumption rate (stated)

0.66 Nm3/kWh

specific water production

< 0.66 l/kWh

housing

20-foot container

The type of analysis was cradle-to-end of utilization, covering raw materials production, energy requirements for manufacture and assembly of parts, the system’s transport

CML 2001 impact category

Figure 3: Global warming for HT-UPS system

Fig. 3 shows the technology’s life cycle global warming (GW) impact in all its relevant phases. Although GW amounts to 368,852 kg CO2 eq. in total, or 0.38 kg CO2 eq./kWh, it is evident that most of this impact, 97%, results from components’ manufacturing, just 2% from transport and just 1% from operation. A look at the system’s components’ contribution to the total GW reveals that the electrolyser’s impact is prevalent and accounts for almost 76% (279,565 kg CO2 eq., or 0.28 CO2 eq./kWh) of the entire GW (368,852 kg CO2 eq.). YEARBOOK 2014 LAborAtorY For HEAt AnD PowEr

(1) If all energy needed in the manufacturing phase of the UPS system is included in the energy balance, the efficiency drops to 14.3% and can be named Energy return on investment (EROI) (Barnhart et al. 2013). (2)

Figure 4: Acidification for HT-UPS system

Abiotic depletion impact (AD) and acidification impact (A, Fig. 4) show the same pattern despite the fact that AD is a global and A is a regional impact criterion. The manufacturing part of the HT-UPS system contributes the most environmental impact in all observed categories (Fig. 5).

EnErGY EFFICIEnCY oF Ht-UPS SYStEM As an alternative to the HT-UPS system, the ICE based UPS system was chosen with equal output power as the HT-UPS (Table 2). For the production of the same amount of energy as the HT-UPS in 10 years, the ICE-UPS system consumes 303,072 kg of fuel. Table 2: HT-UPS technical characteristics

Figure 5: The contribution of phases in observed CML2001 impact categories for HT-UPS system

EnErGY EFFICIEnCY oF Ht-UPS SYStEM The total energy input for manufacturing, transport and 10-years operation of the HT-UPS system amounts to 6.89∙106 kWh, of which 60.3% (4.16∙106 kWh) is produced from renewable energy sources (Table 5). 98.0% (4,078,845 kWh) of RES energy is used by the electrolyser for hydrogen production i.e. electrolyser operation phase. The majority of non-renewable grid-mix energy input (2.74∙106 kWh or 39.7% of total energy input) results from manufacturing phase of the electrolyser, storage tank and fuel cell (98.5% or 2,694,895 kWh of non-RES energy). In the manufacturing phase of the system, the biggest energy consumption is the electrolyser’s manufacturing phase (86.0%). The operation phase efficiency of the HT-UPS system is calculated to be 24.1%, Eq. (1), where is electrical energy produced by HT-UPS system i.e. fuel cell and electrical energy required for hydrogen production in electrolyser.

LAborAtorY For HEAt AnD PowEr YEARBOOK 2014

motor type

gasoline (4 stroke)

volume

389 cm3

output power (DC)

6 kW

nominal motor power

9.70 kW (13 HP)

nominal output voltage

230 V / 400 V

frequency

50 Hz

fuel

unleaded gasoline

specific consumption

0.308 kg/kWhel

Fig. 6 shows environmental impacts of both systems. Ratios of values calculated for ICE-UPS in comparison to HT-UPS in GW, AD, E, and A impact categories are 3.18, 4.15, 0.54 and 1.24, respectively. In three out of the four categories, the HT-UPS shows lower environmental impacts. HT-UPS is a relatively young technology where most of the impacts result from the manufacture and can be expected to be considerably reduced by technical progress in the future. In eutrophication impact category, HT-UPS system has a higher impact due to high electricity and material consumption (steel) in the manufacturing phase (Fig. 6).

Figure 6: Environmental impacts of HT-UPS and ICE- UPS per 1 kWh of produced uninterrupted electricity

EROI comparison shows that ICE-UPS uses less total primary energy input per kWh of uninterruptible electricity delivered 4.72 kWh/kWhel compared to 7.00 kWh/kWhel with HT-UPS (ratio 1.48). But a look at the energy inputs structure (RES vs. non-RES) reveals that a maximum of 0.2% only is provided from RES in the case of ICE-UPS

compared to roughly 60% in the case of HT-UPS system (Fig. 7). Operation phase primary-energy-to-electricity conversions are similar for both systems and amount to 4.70 kWh/kWhel for ICE-UPS and 4.17 kWh/kWhel for HT-UPS system.

Figure 7: Input energy values of both UPS systems per 1 kWh of produced uninterrupted electricity produced

Conclusions A hydrogen-technologies based UPS system’s environmental impacts were analysed with the life cycle assessment method. Life cycle energy balance was calculated, and comparison to an ICE-based UPS was made. The study shows the hydrogen system’s potentials to introduce renewable energy sources in an advanced electric grid, which could reduce environmental impacts of electricity production and distribution. The results show that: • Due to electrolyser’s relatively higher power conversion capacity and subsequently its larger size compared to the fuel cell, electrolyser has the highest environmental impacts in terms of GW, AD and A when compared to fuel cell and storage tank in the HT-UPS system. • Most environmental impacts of the HT-UPS system result from materials and parts manufacture and assembly, i.e. occur in the manufacturing phase of the HT-UPS’s life cycle. The HT-UPS system is nearly environmentally neutral in its operation phase when powered by RES. • 0.375 kg CO2 eq. per 1 kWh electricity is produced with HT-UPS compared to 1.190 kg CO2 eq. in the case of ICE-UPS, which makes the HT-UPS system environmentally sounder in terms of GW. The HTUPS also has lower environmental impacts in terms of AD and A, but higher impact in terms of E, mostly due to large quantities of steel used to manufacture the storage tank. The analysis of life cycle energy balance shows that both systems’ manufacturing phase has a considerable impact on their EROI. Considerably higher energy inputs in HTUPS’s manufacturing phase make its EROI lower than that of the ICE-UPS even though the former has slightly better energy conversion efficiency in its operation phase.

References [1] Anderson D, Leach M (2004) Harvesting and Redistributing Renewable Energy: On the Role of Gas and Electricity Grids to Overcome Intermittency Through the Generation and Storage of Hydrogen. Energy Policy, 32:1603-1614 [2] Baratto F, Diwekar UM (2005) Life cycle assessment of fuel cell-based APUs. J Power Sources, 139:188-196 [3] Cetinkaya E, Dincer I, Naterer G (2012) Life cycle assessment of various hydrogen production methods. Int J Hydrog Energy, 37:2071-2080 [4] El-Shatter TF, Eskander MN, El-Hagry MT (2006) Energy Flow and Management of a Hybrid Wind/PV/ Fuel Cell Generation System. Energy Convers Manag, 47:1264-1280 [5] Hart D (2000) Sustainable energy conversion - fuel cells - the competitive option. J Power Sources, 86:23-27 [6] Hussain MM, Dincer I, Li X (2007) A preliminary life cycle assessment of PEM fuel cell powered automobiles. Appl Therm Eng, 27:2294-2299 [7] Hwang J-J et al (2013) Life cycle performance assessment of fuel cell/battery electric vehicles. Int J Hydrog Energy, 38:3433-3446 [8] Koroneus C, Dompros A, Roumbas G, Moussiopoulos N (2004) Life cycle assessment of hydrogen fuel production processes. Int J Hydrog Energy, 29:1443-1450 [9] Lacko R, Drobnič B, Leban M (2013) Hydrogen Technologies in Self-sufficient Energy System with Renewables. J Energy Technol, 6(3):11-24 [10] Lassaux S, Grégoire F, Germain A (2002) Life Cycle Assessment of Two Renewable Electricity Buffer Systems, 10th LCA Case Studies Symposium, Barcelona, Spain [11] Milewski J, Miller A, Badyda K (2010) The Control Strategy for High Temperature Fuel Cell Hybrid Systems. Online J Electron Electr Eng, 2(4):331-335 [12] Mori, M, Mržljak, T, Drobnič, B, Sekavčnik, M (2013) Integral characteristics of hydrogen production in alkaline electrolysers. J Mech Eng, ISSN 0039-2480, 59(10): 585-594 [13] Morton O (2006) Solar energy: A new day dawning?: Silicon Valley sunrise. Nature, 443:19-22 [14] Ogden J (1999) Prospects for Building a Hydrogen Energy Infrastructure. Annu Rev Energy Environ, 24:227-279 [15] Patyk A, Bachmann T A, Brisse A (2013) Life cycle assessment of H2 generation with high temperature electrolysis, Int J Hydrog Energy, 38:3865 - 3880 [16] Pehnt M (2001) Life-cycle assessment of fuel cell stacks. Int J Hydrog Energy, 10:91-101 [17] Pereira SR, Coelho MC (2013) Life cycle analysis of hydrogen – A well-to-wheels analysis for Portugal. Int J Hydrog Energy, 38:2029-2038 [18] Sherif S, Barbir F, Veziroglu TN (2005) Wind Energy and the Hydrogen Economy - Review of the Technology. Sol Energy, 78:647-660 [19] Smitkova M, Janı´cek F, Riccardi J (2011) Life cycle analysis of processes for hydrogen production. Int J Hydrogen Energy, 36:7844-51 [20] Staffel I, Ingram A (2010) Life cycle assessment of an alkaline fuel cell CHP system. Int J Hydrog Energy, 35:2491-2505

Yearbook 2014 Laboratory for Heat and Power

FluMaBack - Improving the operation of Balance of Plant components Abstract → There is significant scope to improve the durability, efficiency and cost-effectiveness of fuel cell systems. FluMaBack project is about work to improve the design and operation of balance of plant

(BoP) components in back up fuel cell systems,which enable stable, secure and efficient operation of the core components, the fuel cells.

Introduction Mitja Mori Laboratory Laboratory for Heat and Power E-mail mitja.mori@fs.uni-lj.si Room S-I/60 Phone +386-1-4771-715 Status Asistant Professor Research area balance of plant components research for hydrogen based UPS http://www.flumaback.eu/

The development of sustainable, environmentally friendly sources of energy is a global research priority. However, while a great deal of attention has been focused on developing new technologies, there is also significant potential to improve BoP components and achieve the optimal operation of fuel cell systems, an area being addressed by the FluMaBack project. The general aim of the project is to improve the performance, life time and cost of balance of plant (BoP) components for backup fuel cell systems. The project brings together partners from across the EU to develop various BoP components, including the air blower, humidifier, heat exchanger and hydrogen recirculation pump. The project partners had some previous versions of BoP components available. But they were either too robust, too expensive or not the most appropriate with respect to the operation of the overall fuel cell system. The focus now is on improving existing BoP components and developing new

Full Project Title Fluid Management component improvement for Back up fuel cell systems (FluMaBack). http://www.flumaback.eu/ Project Objectives The FluMaBack project aims to improve the performance and life time of balance of plant (BoP) components for back up fuel cell systems, and also reduce their cost. Project Funding Project budget: €4.44 million; Project funding: €2.77 million. Main goals

• Improving BoP components performance, in terms of reliability; • Improving the lifetime of BoP components; • Reducing cost in a mass production perspective; • Simplifying the manufacturing/assembly process of the entire fuel cell system

Figure 1: Balance of plant components developed in the FluMaBack project 16

Laboratory for Heat and Power Yearbook 2014

ones. At the end, we will integrate them into a backup fuel cell system and aim to prove that these BoP components – and the overall system – are more efficient, less expensive, last longer and meet all required performance criteria. Researchers in the project aim to improve the reliability of these BoP components, and also to reduce production costs, simplify manufacturing and extend their lifetime. This has involved both modifying existing components and developing new ones. The hydrogen blower and air humidifier were developed from scratch, so both are completely new products. The air-blower was modified because the fuel cell is a very specific component – this was essential to meeting the requirements of the fuel cell – while the heat exchangers were also modified and one of them was removed from the system.

operational phase of the system, where the hydrogen that also has to be produced is used as a fuel. LIFE CYCLE ASSESSMEnt Part of the work is life cycle assessment of the system in the manufacturing and operational phases. It is important to show that hydrogen technologies in backup systems are environmentally sounder and have a lower environmental impact than conventional backup systems. Limiting the environmental impact of production and addressing sustainability issues are key priorities for most companies, a trend that is likely to grow more pronounced in future. The operational stage of a backup fuel cell system is when the greatest environmental impact occurs, so the hydrogen production technology and performance characteristics of BoP components, and of the overall system, are the most important considerations. ProDUCtIon ProCESS

Figure 2: Air blower (right), Hydrogen recirculation pump (left)

The goal is to achieve an operating time of 10,000 hours over the component’s whole lifetime – equivalent to an operational life of 10 years – without any need for additional maintenance. Life cycle assessment is an important element of this work. We look at the production or manufacturing process, and calculate the environmental impact of both a single BoP component and of the system overall. On the one hand we can get clear data on the environmental impacts of single components in comparison to previous versions, and on the other we are able to compare the manufacturing process with the

This work holds real importance to the operation of fuel cells, potentially improving their efficiency, reliability and durability. However, the BoP components must meet the technical requirements of fuel cells if they are to be operational. “You cannot just plug any BoP component into the fuel cell system and expect that it will work perfectly”. The project is developing and manufacturing BoP components specifically for 3 kilowatt and 6 kilowatt backup fuel cell systems; these components need to meet some exacting requirements. Fuel cells are very sensitive to impurities. The polymer exchange membrane fuel cell is the heart of this system, and the hydrogen used to power it has to be very clean, 99.999 per cent pure. Then we also need to consider properties of the hydrogen and air provided for the operation of the fuel cell. The air has to be humidified, because dry air can cause the membrane to dry out. This leads to membrane degradation, which results in reduced power and a shorter lifetime of both the fuel cell and entire system.

Figure 3: LCA model from cradle to the end of operation YEARBOOK 2014 LAborAtorY For HEAt AnD PowEr

both the 3 kilowatt and 6 kilowatt systems, which is very positive from the cost point of view. However, developing BoP components which will fit in both systems is quite a challenge, as the second system has twice the power and thus requires twice as much air, hydrogen etc.”

Figure 4: Air humidifier with hollow semi-permeable fibres

The technology is available today to address these issues and ensure that fuel cells can work more efficiently. A sound understanding of the fluid dynamics phenomena in BoP components is essential in these terms. “If you have a blower that supplies the air or hydrogen to the fuel cell, it should have certain characteristics. So the volume flow, the efficiency of the blower, should meet specific goals.” The volume flow needs to be within specific boundaries depending on the fuel cell requirements, and many other issues should be considered. “To improve durability, components are usually more robust and thus more material is needed. If in addition more energy is needed, the component will be more expensive. Most criteria in design process preclude each other, which makes this task an even bigger challenge.” Delicate balance A delicate balance needs to be struck between these criteria, and this remains a key priority in development. The performance priorities of the components depends to a degree on the intended application and the likely market; while the project initially focused on emerging markets, now the project is looking more towards Europe and North Africa. “The air blowers and hydrogen blowers are also very useful in various different applications. If the automotive industry decides to use fuel cells very intensively, then the results of our project will be very useful in this area,” he says. Researchers are currently testing the components and assessing their performance as part of the whole fuel cell system. “A few components will be used in

Figure 5: Pressure stream lines in air blower

The project is testing these components to identify such issues and further improve their performance, with the hydrogen blower proving a particular challenge. A test rig is being built for the hydrogen blower to test its durability. “These blowers are new components. When you are dealing with hydrogen you are always confronted with challenges, because hydrogen is prone to leaking and is not compatible with some common materials used for component body, sealing, lubrication, etc.” Durability is a research priority for all of the components being developed within the project. “The components should last for 10,000 hours, and that’s without maintenance. But not just the components – the overall system itself should also last for 10,000 hours without maintenance. That is the goal and that’s what we’re working towards.”

Figure 6: Contribution of phases in global warming (GW) impact criteria for 3 kW UPS FluMaBack system in - top: manufacturing phase bottom: manufacturing and operation phase (10.000 h) with hydrogen production with electrolysis technology mix in Norway

Laboratory for Heat and Power Yearbook 2014

Figure 6: Fuel cell stack

StrUCtUrE oF tHE ProJECt Project is divided in seven (7) work packages (WP) including partners from four (4) European countries. Coordinator is Electro power systems from Turin, Italy.

[4] Presentation of Domel’s BoP component (air blower) by The Fuel Cells and Hydrogen Joint Undertaking, Hannover Messe, 7. April – 11. April 2014, Germany [5] Lange Nacht der Forschung, Alpen-Adria, Universität Klagenfurt Lakeside Science & Technology Park, http://www.langenachtderforschung.at/, Poster: Flumaback activities and LCA results, Klagenfurt, 4th of April, Austria [6] European Fuel Cell Technology & Applications Piero Lunghi Conference, Poster, Side Event Dissemination of European Projects, Poster Session | Project Cocktail 12th December, 11.-13. December 2013 [7] Development and economic challenges of hydrogen technologies breakthrough into practice, Chemical Institute, Ljubljana, 15. November 2013 [8] Programme Review Days by FCH JU, 11. November 2013, Brussels [9] FP7 Fuel Cells and Hydrogen Joint Undertaking, MIUR, Piazza Kennedy 20, Roma: Link to web, 21. January 2013 [10] Energia e idrogeno, L’esperienza italiana nel programma europeo Idrogeno e celle a combustibile [Energy and hydrogen, the Italian experience in the European Hydrogen and Fuel Cells], Organizzato da H2IT, Milano, 13. December 2012, Italy

Project Partners • Electro Power Systems: Project Coordinator, Torino, Italy • tubiflex SPA, Italy • Parco Scientifico e tecnologico per L’Ambiente Environment Park Spa, Italy • Institut Jozef Stefan, Slovenia • Fundacion Para el Desarrollo de las nuevas tecnologias del Hidrogeno en Aragon, Spain • nedstack Fuel Cell technology bV, Netherlands • onda Spa, Italy • University of Ljubljana, Slovenia • JrC - Joint research Centre - European Commission, Belgium • Domel d.o.o., Slovenia Figure 6: Partners involved in FluMaBack

Main BoP and system research/development phase is included in WP 3 and WP 4, WP 5 is mainly for testing of components and the system. In WP 6 RCS (regulation, codes and standards) and market preparation is included, and a big part plays LCA analysis of system manufacturing, operation and end-of-life.

Contact Details Dissemination Coordinator, Dr. Mitja Mori University of Ljubljana Faculty of Mechanical Engineering Aškerčeva 6, SI-1000 Ljubljana, Slovenia t: +386 1 4771 715 E: mitja.mori@fs.uni-lj.si w: http://www.flumaback.eu/ GSM: +386 41 505 003

references [1] A. Debenjak, P. Boškoski, B. Musizza, J. Petrovčič and Đ. Juričić: Fast measurement of PEM fuel cell impedance based on PRBS perturbation signals and Continuous Wavelet Transform, Journal of Power Sources, 254 (2014) 112 - 118 [2] Sinenergija: Slovenian academia and industry in development of hydrogen technology equipment, 2014 [3] Čemažar, Domen. Design and development of hydrogen recirculation blower for fuel cell application: master thesis. Ljubljana: [D. Čemažar], 2013. XIV f., 90 pg., illustrated.

YEARBOOK 2014 LAborAtorY For HEAt AnD PowEr

Hydrogen technologies in power supply from RES Abstract → With increased contribution of renewable energy sources to power supply systems their inherent disadvantage of random and irregular availability is becoming more evident. In order to enable widespread use of renewable energy sources, particularly solar and wind power, an appropriate energy storage technology is needed to provide a buffer between power production and power consumption. One of several applicable tecnologies is hydrogen based energy storage system consisting of an

electrolyzer, hydrogen tank and fuel cell system. To analyze performance and operating characteristics of hydrogen based energy storage numerical models of both electrolyzer and fuel cell system was developed. The models are based on experimental results from an actual small scale hydrogen system. The model will provide a tool for virtual testing of power supply and consumption systems with variouos combinations of power suppliers and consumers.

Introduction Boštjan Drobnič Laboratory Laboratory for Heat and Power E-mail bostjan.drobnic@fs.uni-lj.si Room S-I/67 Phone +386-1-4771-715 Status researcher Research area Hydrogen technologies, Energy systems

Sustainable power supply for the future will be based on renewable energy sources (RES) among which solar, wind and water energy are generally considered to be the main contributors. While the availability of solar, wind and in general also water energy does follow some daily and seasonal patterns it is still essentially random and irregular. Power production is therefore not synchronized with power consumption, which follows its own patterns and has also considerable random component. Unlike conventional power systems, where power production follows power consumption, RES based power supply sysems will require an appropriate buffer or energy storage system.

Figure 1: Concept of hydrogen technologies in a RES based power supply system

Hydrogen technologies are proving to be a feasible solution, [1]. During intervals when power production exceeds consumption, excess power is converted into hydrogen through 20

Laboratory for Heat and Power Yearbook 2014

electrolisys of water, while when the consumption exceeds available resources hydrogen is converted back to power in hydrogen fuel cells, Fig. 1. Electrolyzer The purpose of the electrolyzer in the observed power supply and consumption system is to transform excess electric energy which cannot be efficiently stored for longer periods of time into chemical energy of molecular hydrogen. The proces of electrolysis of water can be written as, [2] (1) In Eq. (1) F is Faraday constant and represents the electric charge needed to split one mole of water into molecular hydrogen and oxygen. The electric charge delivered to the elecrolytic cell depends on electric current and voltage. The latter also determines the efficiency of the cell since certain overpotential is required to cover internal losses of the electrolytic cell. Actual voltage required for electrolysis consists of: • theoretical (reversible) potential for electrolysis • overpotential that covers resistance of electrolyte • overpotential that covers resistance of electrodes • overpotential for time degradation of materials Taking into consideration the entire electrolyzer system including all the

ballance of plant components the overal efficiency depends on: • power consumption for the electrolytic proces including the abovementioned internal losses of electrolytic cell • hydrogen loses due to purging of hydrogen filters • supplying fresh water for the process • hydrogen accumulation in oxygen • power consumption of auxilliary components of the electrolyzer system

Again, in Eq. (2) Faraday constant (F) represents the electric charge that is released during the reaction. Similar to the electrolyzer, there are several types of energy loss at fuel cell operation as well, [3]: • internal resistance of the fuel cells • hydrogen loss due to regular purging of fuel cell • power conversion loss for charging and discharging a battery that is a part of the fuel cell system • power consumption of balance of plant components The performance of the fuel cell stack is still being researched, particularly the role and behaviour of the additional battery in the fuel cell system. Based on the results of the previous research a Simulink model of the fuel cell stack has been developed and included in the model of the hydrogen based energy storage system shown in Fig. 2. ConCLUSIonS Electric energy as a convenient form of energy has a notable disadvantage since it cannot be stored. Hence, the production must follow the consumption, which is not possible when using renewable energy sources. Energy storage is therefore an essential part of a RES based energy supply system. Hydrogen technologies offer a possibility to store electric energy in the form of chemical energy of gaseous hydrogen. As with any energy conversion, energy losses are inevitable. To study performance of hydrogen based energy storage system numerical model of its components were developed taking in consideration various energy losses. The model enables dynamic simulation of the system‘s performance within RES based power supply system with different configurations of power producers and consumers as well as the system‘s environmental impact.

Figure 2: Numerical model of hydrogen based energy storage

system

All these losses have been examnied experimentaly and appropriate empirical models have been developed. Using the Mathworks Simulink software a model of electrolyzer was set up and used in a wider model of hydrogen based energy storage system shown in Fig. 2. FUEL CELL The fuel cell subsystem is needed to convert stored energy in the form of chemical energy of molecular hydrogen back into electric energy. Basic reaction is the opposite of the electrolysis reaction, therefore the reaction is reversible, [2]. (2)

references [1] Santarelli, M., Cali, M.: Design and Analysis of Standalone Hydrogen Energy Sistems With Different Renewable Sources, International Journal of Hydrogen Energy, Vol. 29, No. 15, pp. 1571 – 1586, 2004. [2] Barbaro, P., Bianchini, C.: Catalysis for Sustainable Energy Production; WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim, 2009. [3] J. Larminie, A. Dicks. Fuel Cell Systems Explained – 2nd ed. . Anglija : John Wiley & Sons Ltd. , 2003. List of publications [1] Mori, M., Mržljak, T., Drobnič, B., Sekavčnik, M.: Integral characteristics of hydrogen production in alkaline electrolysers. Strojniški vestnik, ISSN 00392480, Oct. 2013, vol. 59, no. 10, pp. 585-594. [2] Lacko, R., Drobnič, B., Leban, M.: Hydrogen technologies in a self-sufficient energy system with renewables. Journal of energy technology, ISSN 1855-5748., Avg. 2013, vol. 6, iss. 3, pp. 11-24. [3] Mori, M., Jensterle, M., Mržljak, T., Drobnič, B.: Lifecycle assessment of a hydrogen-based uninterruptible power supply system using renewable energy. The international journal of life cycle assessment, ISSN 0948-3349, Aug. 2014, pp. 1-13

YEARBOOK 2014 LAborAtorY For HEAt AnD PowEr

Intrusive Electrostatic Sensors Used On Lignite Fired Boiler Abstract â&#x2020;&#x2019; Study is dealing with industrial application of measurement system consisting of a grid of intrusive electrostatic sensors. The system is used for monitoring the distribution and velocity of pulverized coal inside the ducts connecting coal mills and burner nozzles. The main purpose of the system is to adjust fresh-air distribution according to actual coal-particles distribution and thus optimize the combustion i.e.

reduce emission of harmful gases like NOx and CO. Besides this coal-particlesâ&#x20AC;&#x2122; velocity is measured in each duct using cross-correlation of signals of two electrostatic sensors positioned sequentially in the direction of flow. The velocity data is used for early detection of mill overload and for identification of mill wear.

Introduction Igor KuĹĄtrin Laboratory Laboratory for Heat and Power E-mail igor.kustrin@fs.uni-lj.si Room 304 Phone +386-1-4771-304 Status Research assistant Research area Coal Fired Boilers for Thermal Power Plants

Firing with solid fuel, compared to firing with liquid or gaseous fuel, is more difficult to optimize due to being unable to accurately and consistently measure fuel flow and to maintain optimum air ratio in the combustion zone. Pulverized coal combustion is somewhat similar to the combustion of liquid fuels like, for example, residual fuel oil or light fuel oil, but greatly depends on the type of coal and the manner in which it is pulverized and transported into the furnace. Lignite and brown coal, the type used in the Slovenian power plants, are pulverized and transported to the burners by means of mills, which crush coal particles with impact-plates and the mutual friction. Once coal particles are light enough gas flow carries them away from the mill to the burner through the separator. Only the smallest particles can continue to the burner through the separator. In addition, due to the stochastic turbulent process in the fan mill pulverized coal is rather unevenly distributed over the exit cross-section of the mill. Distribution of pulverized coal varies with the mill load, wear of main components of the mill and the type and quality of coal. Wear of vital parts of the mill is progressing slowly and continuously, coal properties vary more often while the mill load varies most frequently and has the greatest influence on the

Laboratory for Heat and Power Yearbook 2014

distribution of pulverized coal. Since the mill normally supplies pulverized coal to two or more burner nozzles, uneven distribution of coal means that not all burner nozzles receive the same mass flow of coal. This makes it difficult to ensure proper air ratio for each nozzle. Many experts are investing significant effort to remedy this problem and one of the methods that can be helpful is presented here.

Figure 1: Schematic arrangement of fan/ impact mill and burner nozzles with ESDsensors location

Background Sensors operating on the principle of electrostatic discharge (ESD) used for measuring the distribution, velocity and grain size of particles in real time have been already reported and discussed in the literature [1], [2], [3], [4], [5], [6]. The principle of operation is shown on Figure 1. Figure 3: Cross-correlation method

Insulating material

Resistor

Despite the fact that the primary purpose of ESD sensors is fitting the combustion-air distribution to pulverized coal distribution it is useful to know the actual velocity of coal particles in the channel for example to optimize the separator adjustment, to detect eventual mill overload, to detect excessive wear of vital mill components or to correct the level of the measured ESD signal with respect to actual velocity of coal particles. According to [7] the current I flowing into the probe depends on many variables. The general relation is expressed as:

Amplifier

(3)

ESD sensor

Duct wall

Figure 2: Principle of operation

Rods made of conductive material (ESD sensors) are inserted into the pulverized-coal channel and insulated from the walls of the channel. Walls of the duct are also made of electrically conductive material and grounded. It can be assumed that the interior of the channel is free from any external electric fields. Electric current is flowing from ESD sensor towards/ from the ground due to electrostatic charge being deposited on ESD sensor. ESD sensor is connected to the data acquisition system through a resistor of known internal resistance. This way electric current is converted to DC voltage which is further processed by the data acquisition system: filtered, averaged, amplified, converted, displayed, etc. ESD signals can be used to calculate the coal-particles distribution, velocity profile and in some cases even the quality of grinding. Velocity of Coal-Particles The velocity of coal particles can be calculated by cross correlating signals of two ESD sensors, placed sequentially in the flow direction and separated by L.

(1)

Where L is the distance between two sensors and τ is the particles transition time from one sensor to the other. Analog signals from ESD sensors are digitized and sent to a computer. For transition time determination the following equation is used:

Where: ε - the electrical permittivity of particle K - the electrical conductivity of particle d - the particle diameter V - the flow velocity in the region of the probe ρ - the density of air M ˙ - the mass flow rate or flux of particle flow Ap - the cross-sectional area of probe According to Pi Theorem [8] these variables may be expressed as four dimensionless groups such that: (4) Then if only V and M ˙ vary and other parameters are assumed to be constant, the above function can be simplified as:

(5)

Knowing the velocity mass low of coal particles may be determined. Nevertheless there is no need to determine exact mass flow of coal entering the fan/impact mill because mass flow of coal to each mill is controlled by feeder load. The distribution of coal-particles mass flow inside the duct is more important. If an assumption of approximately flat velocity profile inside the duct is made than mass flow of coal-particles is proportional to:

(6)

t=0, ±1, ±2, ±3,… (2)

Yearbook 2014 Laboratory for Heat and Power

PoSItIon oF ESD SEnSorS To determine the relative distribution of the pulverizedcoal mass among burner nozzles the sensors must be positioned so that the average mass flow of coal dust to each burner nozzle is most accurately detected. It is therefore necessary to insert two or more electrostatic sensors in each duct in the way shown in Figure 4. In the case of fan/impact mills the most suitable location for sensors is the straight vertical section of coal-dust duct between a mill and a burner where coal-dust duct already contains barriers between ducts leading to respective burner nozzles (Figure 1).

Figure 4: Position of ESD sensors

VErIFICAtIon oF MEASUrEMEntS Sensitivity For the verification of sensitivity ESD sensors the following test was carried out. Figure 5 shows the most representative results.

and forward it to the burner before next feeder bar delivers the next portion of coal. Fluctuation is very obvious. It must emphasized that 25 % feeder load is well below the normal operating range which is between 45 % and 65 %. Therefore during normal operation such coal-flow fluctuation is not present. To verify whether signal fluctuation in fact mirrors coal-dust flow fluctuation the frequency of coal-feeder bars at different loads was measured by means of a stop watch. Figure 6 shows frequency of coal-feeder bars and frequency of signal fluctuation. Frequencies of feeder bars and signal are identical which proves that signal fluctuation represents coal-flow fluctuation in this case caused by portions of coal fed to the mill by each feeder-bar.

Figure 6: Frequency of coal-feeder bars and signal-fluctuations frequency

Repeatability Repeatability was verified by repeatedly measuring the average signal levels at same coal-feeder loads. Signal levels were measured during increasing of the feeder load and during decreasing of the load. Repeatability was very satisfactory taking into account the hysteresis of the instrumentation for feeder-load control. Accuracy and usefilness Since the main purpose of ESD sensors is to determine coal-particles distribution among burner nozzles it is essential to verify if signals in fact represent coal-dust distribution with sufficient accuracy.

Figure 5: Signal of ESD sensor at 25% coal-feeder load

Two sensors per burner nozzle were electrically coupled together. Sampling frequency was 500 readings per second. Averages of 500 readings (one second) and averages of total period of five minutes are drawn. Intentionally very low feeder load was chosen to force coal-particles flow to fluctuate. Fluctuation is caused by the fact that the feeder speed is so low that mill manages to pulverize the portion of coal delivered by each feeder bar 24

LAborAtorY For HEAt AnD PowEr YEARBOOK 2014

Figure 7: Accuracy of method

Laser measuring equipment was used to provide data on coal-particles mass flow for comparison with ESD signals. Measuring with laser equipment took place in six measuring points above each ESD sensor. Figure 7 shows an example recorded at 63 % feeder load. Values of signal and mass flows are presented in a relative manner in relation to the mean value. The comparison shows that the relative values of ESD signals are in good agreement with the relative values of mass flow and represent accurately the distribution of coal dust in the coal-dust channel. Conclusions Sensitivity, repeatability and accuracy of electrostatic discharge sensors are sufficient for continuous and accurate monitoring of coal-dust distribution in coal-dust ducts between fan/impact mills and burner nozzles. Method that uses sensors for coal-dust-distribution determination can be of great help in the process of combustion optimization. Existing or upgraded control of combustion air flow can be used in connection with ESD sensors to maintain optimal air ratio at every burner nozzle regardless of coal-feeder load, wear of main mill parts and coal properties.

List of publications [1] KUŠTRIN, Igor, LENART, Jože, KOKOVNIK, Alojz, GOSTINČAR, Primož, HOČEVAR, Marjan. Combustion optimization with respect to coal-dust distribution between burner nozzles. V: 21. Mednarodno posvetovanje Komunalna energetika, 15. do 17. maj 2012, Maribor, Slovenija. VORŠIČ, Jože (ur.). Zbornik = Proceedings. Maribor: Fakulteta za elektrotehniko, računalništvo in informatiko, 2012, str. 1-10. [2] KUŠTRIN, Igor, KOKOVNIK, Alojz. Uporaba elektrostatičnih zaznaval na kotlih s premogovo prašno kurjavo za optimizacijo zgorevanja in detekcijo motenj delovanja sistema za pripravo goriva. V: Mednarodna konferenca daljinske energetike 2014. Energetska učinkovitost od proizvodnje do uporabe = Energy efficiency : from production to consumption. Ljubljana: Slovensko društvo za daljinsko energetiko: = Slovenian District Energy Association, 2014, str. 47.

References [1] Mohd. Fua'ad Bin Hj Rahmat, Dan Yaw Wee Lee, Electrostatic Sensor For Real Time Mass Flow Rate Measurement Of Particle Conveying In Pneumatic Pipeline, Jurnal Teknologi, 41(D), Dis. 2004, pp. 91-104. [2] Woodhead S. R., Denham J. C., Armour-Chelu D. I., Electrostatic Sensors Applied To The Measurement Of Electric Charge Transfer In Gas Solids Pipelines, Jurnal of Physics: Conference Series 15, 2005, pp. 108-112. [3] Lihui Peng, Yan Zhang, Yong Yan, Characterization Of Electrostatic Sensors For Flow Measurement Of Particulate Solids In Sqare-Shaped Pneumatic Conveying Pipelines, Sensors and Actuators A 141 (2008), pp. 59-67. [4] Laux S., Grusha J., McCharty K., Real Time Coal Flow Particle Size Masurement for Improved Boiler Operation, Foster Wheeler Energy Corporation Document (13 pages), (http://fwci.com/publications/tech_papers/files/TP_FIRSYS_00_03.pdf). [5] Wallash Al, Continous Voltage Monitoring Techniques for Improved ESD Auditing, EOS/ESD Symposium 03-394. [6] Woodhead S. R., Amadi-Echendu J. E., Solid Phase Velocity Measurements Utilising Electrostatic Sensors And Cross Correlation Signal Processing, 0-7803-261 5-6/95/$4.00 0 1995 IEEE. [7] Jianyong Zhang (2012). Air-Solids Flow Measurement Using Electrostatic Techniques, Electrostatics, Dr.Huseyin Canbolat (Ed.), ISBN: 978-953-51-02397, InTech, Available from: http://www.intechopen. com/books/electrostatics/air-solids-flow-measurement-using-electrostatic-techniques. [8] Curtis, W.D., Logan, J.D., Parker, W.A., Dimensional analysis and pi theorem, Linear Algebra and Its Applications, 1982, 47 ©, pp. 117-126.

Yearbook 2014 Laboratory for Heat and Power

The integration of hydrogen technologies for utilisation of distributed renewable energy sources Abstract → A hybrid energy system, based on renewable energy sources and with hydrogen storage, can become an alternative for stand-alone electricity and heat supply. The objective of my work is a numerical and experimental evaluation of the feasibility of a completely renewable supply of power and heat for

an isolated household, and a comparison to reference and alternative energy supply scenarios, including a sensitivity analysis of the effects of size (number of households included in the energy system) and geographical location on the energy system design.

Introduction Rok Lacko Laboratory Laboratory for Heat and Power E-mail rok.lacko@inea.si Room S-I/60 Phone +386-41-279-041 Status PhD student (started: October 2010, to be completed: January 2015) Research area Hydrogen technologies Mentor prof. dr. Mihael Sekavčnik

Climate change mitigation, natural environment preservation, increasing needs for energy and uncertain costs of fossil fuel supply in the future are the reasons for the increasing interest in local and renewable energy sources (RES), as opposed to fossil energy use. While fully energy self-sufficient dwellings are still rare, solar and wind power units have been widely adopted for private family homes. The introduction of RES into the energy supply, however, raises certain issues in load balancing due to their intermittent and non-storable nature. This especially applies to wind and solar energy, and to a lesser respect to other RES. Furthermore, a single self-sufficient user presents the most challenging case of RES integration, due to its inability to import and export surplus energy. Electricity supply based on renewable energy sources, coupled with hydrogen storage, has already been proposed as a technically viable solution [1]. Several technical and economic analyses of stand-alone RES energy systems with hydrogen storage, focusing on electricity supply only, have already been discussed [2]. Such hybrid energy systems have been experimentally demonstrated [3]. Furthermore, studies on REShydrogen energy systems considering combined heat and power (CHP) have been reporting on energy and exergy analysis [4], system performance assessment methodology [5] and a mathematical model of a wind-

hydrogen CHP system with metal hydride storage [6]. RES-hydrogen CHP systems have also been successfully demonstrated [7]. In my research an optimal selfsufficient renewable hydrogen energy system has been numerically determined, based on an actual geographical location (Slovenia’s coastal area), the availability of energy sources, electric load dynamics and components’ technical and economical characteristics [8]. Numerical results were experimentally proven using a demonstration system with hydrogen technologies [9]. The results of both studies have shown that the required relatively large production capacity causes excess electricity generation, which is not used by the household. Although stand-alone REShydrogen energy systems have been proposed and studied, common shortcomings of those analyses include (a) a pre-determined system design, (b) optimization of system’s performance without preliminary optimising its configuration or design, (c) overgeneralised input parameters, such as using typical daily consumption data sets only, (d) short term simulation only (day, week) and finally (e) numerical models and simulation results not being experimentally evaluated and verified. Studies also show that results are highly dependent on numerous local factors, e.g. meteorological conditions; therefore, a site-specific analysis is needed for credible results [5]. Additionally, the influence of energy system size to its optimal design has not been analysed.

In my research I aim to overcome the detected shortcomings by considering the topics of optimal hydrogenRES system design based on relevant, site specific and actual measured input data, hourly system’s dynamics analysis for a period of one year, including experimental evaluation of numerical results. In the analysed energy system hydrogen is produced (and stored in a tank) by an electrolyser, which is powered by the surplus electricity from renewable energy sources, using solar and wind technologies (specifically at summer daytime). When RES are scarce, or demand is high, additional power is needed; therefore, the fuel cell converts the chemical energy of the stored hydrogen gas directly into electricity (usually at night and in winter).

Table 2: Optimal system configuration electrical production and consumption. Production PV array

18025

3609

Fuel cell

2397

24030

100

Total Consumption

kWh/year

AC primary load (household)

4159

Electrolyser

9287

13446

100

Other

Numerical simulation results A feasible system is defined as a hybrid system configuration that is capable of meeting the required load. Under given conditions, one system configuration was found to be feasible. The optimal combination, with the lowest total net present cost (€136,063), is presented in Table 1. The cost of energy for optimal system combination is €2.8/kWh. The combined nominal primary (RES) and secondary (fuel cell) power source capacity is 33 kW, while peak demand is 3.8 kW. The optimal electrolyser and converter size is 4 kW, each, with tank capacity being 30 kg of hydrogen. Figure 1-3 show the principal components’ operating characteristics: electrolyser input power, fuel cell output power and their duration curves and hydrogen storage level, respectively. Hydrogen energy storage shows the ability to store inter-seasonal fluctuations of RES availability. Summers’ higher RES energy density is stored to be used during the colder half of the year (Figure 3). Unlike the fuel cell (Figure 2), the electrolyser (Figure 1) operates during 74% of the operation time at nominal power (Figure 3). In Figure 3, the difference between electrolyser electricity consumption and fuel cell electricity production, which represents the overall efficiency of hydrogen energy storage (averaging 26%) is also shown.

Wind turbines

Total

Results and discussion

kWh/year

10122

42.1a

462

11.1b

Unmet electric load

Capacity shortage

Renewable fraction

100

Excess electricity DC/AC conversion loss

a share of total production b share of AC primary load

Figure 1: The result of annual simulation of electrolyser’s input power in steps of one hour.

Table 1: Optimal system configuration, based on lowest net present cost. Component

Size

PV array

17 kW

Wind turbine

12 kW

Electrolyser

4 kW

Fuel cell

4 kW

Power converter

4 kW

Hydrogen tank

30 kg

Table 2: presents an optimal system’s electrical production and consumption values. The PV arrays produce 75%, while the secondary power source (fuel cell) produces 10% of the electricity. The electrolyser’s electricity consumption rate is 69%, while the household consumes 31%. All electric load was met throughout the year, and the excess electricity is 42% of overall production.

Figure 2: The result of a simulation of fuel cell’s output power: full year scale

Experimental results Figure 4 presents the chosen one-week period with both numerical and experimental results of the fuel cell and electrolyser power. Both typically start once per day. The electrolyser operates in periods of excess electricity production during the daylight. In contrast, the fuel cell mostly operates during the night, acting as a substitute for the missing solar irradiation. A comparison of integral 27

parameters, both simulation and experiment, presented in Table 4, shows remarkably good matches of both results, with differences of less than 3%.

household consumes 663 litres of oil annually; at an oil price of 1 €/litre, annual heating costs are €663. ALTERNATIVE ENERGY SYSTEM (FUEL CELL CHP + DUMP LOAD) The alternative scenario aims to offset fossil fuel consumption by using surplus energy to supply the thermal load as much as possible. HOMER enables excess electricity conversion into heat via dump load (1,512 kWh) and fuel cell waste heat use (1,513 kWh in CHP mode). Nevertheless, the renewable energy available for heating is not fully utilised; therefore, some (2,523 kWh or 302 litres of oil equivalent) fossil fuel has to be used. The alternative system utilises 15% of excess electricity and 79% of the fuel cell’s available heat. Even though there is a sufficient amount of excess energy available for heating, not all could be utilised due to the low availability of RES and a mismatch between production and demand. Thermal operation diagrams with different time scopes (day – upper, week – middle, year – lower) are shown on Figure 6. Annual operation (fuel) savings, compared to the reference scenario, account for €361.

Figure 3: Electrolyser and fuel cell duration curves (top) and hourly averaged hydrogen storage content (bottom).

Figure 5: Alternative system thermal operation

Figure 4: Simulation and experimental power distribution of a 1-week period in summer.

Parameter

EL model

EL Index experiment

FC FC Index model experiment

energy

4.20 kWh

4.17 kWh

0.99

0.74 kWh

0.72 kWh

0.97

H2 amount

0.077 kg

0.075 kg

0.97

0.046 kg

0.045 kg

0.98

70.9%

0.98

40.9%

40.6%

0.99

el. 72.2% efficiency

RENEWABLE ENERGY SYSTEM Since the alternative scenario’s amount of unused energy for heating remains sufficient to fully offset remaining fossil fuel consumption (2,523 kWh), a thermal storage was introduced to the renewable scenario. A custom numerical model was set to analyse the thermal storage operation. The results show that a 1,000 kWh storage capacity with an 8 kW electric heater (dump load) enables 100 % renewable heat and electricity supply of a reference household. The annual thermal storage heat content is shown in Figure 7. The fossil fuel consumption is replaced by better utilisation of the excess electricity (39%), of the fuel cell waste heat (87%) and of the electrolyser waste heat (77%). Since renewable scenario requires no fuel, annual operational savings are €663, compared to the reference, and €302 compared to the alternative scenario.

Table 3: Simulation and experiment results comparison.

CHP hydrogen energy system for household application REFERENCE ENERGY SYSTEM In the reference scenario, heat is supplied independently of the renewable system, by burning oil in a boiler. The

Figure 6: Storing renewable heat during the year requires a seasonal thermal storage.

Sensitivity analysis ENERGY SYSTEM SIZE The sensitivity of the optimal energy system design to its size (with respect to the number of households included) has been performed by based on the energy consumption patterns of 80 different households. Several different sizes energy systems that consist of 1, 2, 5, 10, 20, 30, 40, 60 and 80 households have been numerically analysed and compared. Results show that the (relative) dynamics of electricity consumption decrease with size (from 1 to 80 consumers interconnected), since • the (proportional) power range of electricity consumption decreases – maximal power decreases from 5.59 to 1.02 kW (- 82%) while simultaneously minimal power increases from 0.00 to 0.20 kW, • the standard deviation of consumption decreases from 0.43 to 0.15 kW (+ 65%), • the load factor increases from 0.12 to 0.45 kW (+ 270%), • the area under the (normalised) load curve increases. The described changes in the dynamics of electricity consumption result in a slight decrease in the normalised production capacity of the entire production, electrolyser capacity and hydrogen tank by 15, 22 and 25%, respectively. More notably, the normalised inverter and fuel cell capacities decrease by 82%. Furthermore, the share of fuel cell capacity within the total production capacity is decreased from 0.17 to 0.04 (- 78%). The reduction of the share of hydrogen technologies within the energy system lowers the economical parameters, such as the initial capital (- 29%), operating costs (- 66%), net present cost (- 38%) and levelised cost of energy (- 38%). The results of simulation of the physical behaviour of the energy system show that the fuel cell’s load factor increases with size by 391 %. The other operating characteristics (i.e. average power, electricity production and consumption, hydrogen production and consumption, operating hours, excess electricity) show less dependence to the size of the system (18 to 27% difference). Additional results show sufficient heat supply with increasing the energy system size, which is utilised from available sources. Proportional heat storage size varies between 1,000 and 1,525 kWh per household, while rated power of the electric heater (dump load) decreases from 8.0 to 3.4 kW (58%). With the increasing number of consumers all available heat from hydrogen technologies (fuel cell and electrolyser) is utilised. In contrast, the utilisation rate of the available excess electricity for heating does not significantly increase and varies between 39 and 67%. Installing the thermal storages dispersedly increases its total heat capacity by 33%. GEOGRAPHICAL LOCATION The optimal system design calculations have been performed for 5 different locations in Slovenia. The required total electrical production capacity varies by less than 12%, while the size of a thermal storage varies by 114%, with its maximum size in Ljubljana. The economically most efficient location to install the energy system is Portorož where NPC is 136,000 EUR. The difference to other locations is less than 32%.

References [1] Mathiesen BV, Lund H. Fuel-efficiency of hydrogen and heat storage technologies for integration of fluctuating renewable energy sources. 2005 IEEE St. Peterbg. powertech Conf. Proc., St. Peterburg: Faculty of Mechanical Engineering and Naval Architecture; 2005, p. 1–7. [2] Vosen SR, Keller JO. Hybrid energy storage systems for stand-alone electric power systems : optimization of system performance and cost through control strategies. Int J Hydrogen Energy 1999;24:1139–56. [3] Chaparro a. M, Soler J, Escudero MJ, de Ceballos EML, Wittstadt U, Daza L. Data results and operational experience with a solar hydrogen system. J Power Sources 2005;144:165–9. [4] Xie D, Wang Z, Jin L, Zhang Y. Energy and exergy analysis of a fuel cell based micro combined heat and power cogeneration system. Energy Build 2012;50:266–72. [5] Dorer V, Weber R, Weber a. Performance assessment of fuel cell micro-cogeneration systems for residential buildings. Energy Build 2005;37:1132–46. [6] Pedrazzi S, Zini G, Tartarini P. Modelling and simulation of a wind-hydrogen CHP system with metal hydride storage. Renew Energy 2012;46:14–22. [7] Voss K, Goetzberger A, Bopp G, Haeberle A, Heinzel A, Lehmberg H. The self-sufficient solar house in Freiburg results of 3 years of operation. Sol Energy 1996;58:17–23. [8] Lacko R, Drobnič B, Mori M, Sekavčnik M. Planning of a self-sufficient energy system with 100 % renewable energy sources and hydrogen storage. 22nd Int. Expert Meet. Power Eng., Maribor: 2013, p. 1–13. [9] Lacko R, Drobnič B, Leban M. Hydrogen technologies in a self-sufficient energy system with renewables. In: Avsec J, Konovšek D, Špeh N, Štumberger B, Žerdin F, editors. 3rd Int. Conf. Energy Technol. Clim. Chang., Velenje: 2013, p. 1–12.

List of publications [1] Lacko R, Drobnič B, Mori M, Sekavčnik M, Vidmar M. Stand-alone renewable combined heat and power system with hydrogen technologies for household application. Energy 2014:1–7. [2] Lacko R, Drobnič B, Leban M. Hydrogen technologies in a self-sufficient energy system with renewables. In: Avsec J, Konovšek D, Špeh N, Štumberger B, Žerdin F, editors. 3rd Int. Conf. Energy Technol. Clim. Chang., Velenje: 2013, p. 1–12. [3] Lacko R, Drobnič B, Mori M, Sekavčnik M. Planning of a self-sufficient energy system with 100 % renewable energy sources and hydrogen storage. 22nd Int. Expert Meet. Power Eng., Maribor: 2013, p. 1–13. [4] Lacko R, Drobnič B, Mori M, Sekavčnik M, Vidmar M. Stand-alone renewable CHP system with hydrogen technologies. 6th Int. Conf. Sustain. Energy Environ. Prot., Maribor: 2013, p. 588–93. [5] Lacko R, Drobnič B, Sekavčnik M, Mori M. Hydrogen energy system with renewables for isolated households: The optimal system design, numerical analysis and experimental evaluation. Energy Build 2014;80:106–13. [6] Lacko R, Leban M. Hydrogen technologies in a self-sufficient energy system with renewables. J Energy Technol 2013;6:11–24. [7] Lacko R, Pirc A, Mori M, Drobnič B. Step By Step Numerical Approach to a Self-Sufficient Micro Energy System Design. Int J Eng Res Appl 2014;4:1–10.

CHP system based on HT PEMFC stack Abstract → Efficiently combining proton exchange membrane fuel cell (PEMFC) stack with methanol steam reformer (MSR) into a small portable system is still quite a topical issue. Using methanol as a fuel in PEMFC stack includes a series of chemical processes where each proceeds at a unique temperature. In a combined MSR-PEMFC-stack system with integrated auxiliary fuel processors (vaporizer, catalytic combustor, etc.) the processes are both endothermic and exothermic hence proper thermal integration can help raising the system efficiency. First task is to identify possible configurations and prove the concept of such system with a numerical model. For that reason, three systems are designed based on different PEMFC stacks and MSR. Low-temperature

(LT) and conventional high-temperature (cHT) PEMFC stack characteristics are based on available data from commercial suppliers. Also, a novel high-temperature (nHT) PEMFC stack is proposed because its operating temperature coincides with that of MSR. A comparative study of modelled systems is performed using a mass and energy balances zero-dimensional model, which is interdependently coupled to a physical model based on finite element method (FEM). Second task is to construct a prototype of such integrated system and measure its characteristics on the experimental setup. Also, methods of construction and suitability of materials are to be identified in relation to the operating conditions.

Introduction Andrej Lotrič Laboratory Laboratory for Heat and Power E-mail andrej.lotric@fs.uni-lj.si Room 306 Phone +386-1-4771-306 Status PhD student (started: October 2011, to be completed: September 2015) Research area Hydrogen technologies, PEMFC Mentor Prof.Dr. Mihael Sekavčnik

One of the major constraints for the PEMFCs is that they use hydrogen as a fuel which has a very low energy density per volume at conditions of standard ambient temperature and pressure (SATP). Hydrogen is actually just an energy carrier and has to be produced from other sources using various processes (e.g. electrolysis, gasification or reforming). As energy carrier methanol is also showing promise because at SATP, compared to hydrogen at 800 bar, it still has more than 2.6 times higher energy density per volume and is in liquid form which greatly facilitates its storage and transportation. Methanol can be used in a direct methanol fuel cell (DMFC) or it can be transformed into hydrogen rich gas using an on-site reformer. Although systems using DMFCs are simpler comparing to combined systems with methanol reformer and PEMFCs, far greater energy densities are archived with combined systems. Methanol steam reforming is more attractive than methanol auto-thermal reforming because of its higher attainable hydrogen concentration and smaller amount of carbon monoxide (CO) in the reformate stream. Unfortunately, at operating temperatures of LT PEMFCs (around 80 °C) CO tends to strongly adsorb onto the active sites of platinum (Pt). Cleaning processes like

Laboratory for Heat and Power Yearbook 2014

water-gas shift (WGS) and/or preferential oxidation (PROX) are needed to reduce the concentration of CO. With Pt-Ru catalyst around 10 ppm of CO can be tolerated in the stream or with air-bleeding technique even up to 50 ppm [1]. The phenomenon of CO adsorption on Pt is far less pronounced at higher temperatures. That is why cHT PEMFC based on polybenzimidazole (PBI) polymer doped with phosphoric acid (H3PO4) operate reasonably unaffected at temperatures above 160 °C and tolerate up to several percent of CO [2]. Other advantages of cHT over LT PEMFCs are simplified water management (no humidification, only water vapor is present) and higher temperature level of produced heat which can be used to preheat incoming gases or vaporize water-methanol mixture. Despite higher tolerance of cHT PEMFC to CO a major drawback is the adsorption of acid anions (H2PO4-) on active sites of Pt catalyst. This slows down the oxygen reduction reaction (ORR) kinetics more than in the case of LT PEMFC, hence, lower power density is achieved. Since H3PO4 is highly soluble in water and can leach out from the membrane, the lower temperature limit of cHT PEMFC is about 130 °C. Conversely, at temperatures higher than 180 °C, the vapor pressure of H3PO4 becomes prohibitively high, which also leads to losses of acid from the fuel cell.

The main challenge of every energy system is to recuperate heat available within the system to maximize its efficiency. Since some unit reactions in combined PEMFC systems are exothermic (catalytic combustion, PROX) and others endothermic (vaporization, MSR) the goal is to direct heat from sources to sinks within the system. The most obvious step that applies such action is direct thermal coupling of the MSR and the HT PEMFC stack which can be done in two ways. One is to use the conventional catalyst in the MSR which enables to attain practically full conversion of methanol between 250 ‑ 300 °C. This however depends on geometry, flow characteristics, catalysts loading, and steamto-carbon ratio (S/C). Typically, conversions above 95% are achieved at 250 °C with S/C ≤ 2:1 [3-5] while some researches also achieved 100% conversion with S/C ≤ 1.5:1 [6, 7]. Since conventional PEMFCs operate at lower temperature than the MSR there is a need for development of a nHT PEMFC that could operate at temperatures higher than 250 °C and consequently enable direct thermal coupling of the nHT PEMFC stack and the MSR. Some research efforts in this direction have been published with the development of so-called solid acid fuel cell (SAFC) [8, 9]. The second way is to use cHT PEMFCs and newly developed LT catalyst [10] that will allow the MSR to operate at temperatures below 200 °C and still achieve near to 100% conversions. There are researches devoted to thermal coupling of the catalytic combustor and the MSR [4, 5, 7]. In research [6] even coupling of cHT PEMFC stack and conventional MSR was attempted. There are numerical studies on the integrated systems but they are more focused on the implementation [11] and dynamics [12] of the whole system. More in-depth insight to the design of small compact systems was presented in [13] although the PEMFC stack was excluded from the analysis. Also an experimental system was constructed by the same author in his previous work [14]. Concept and experimental data of compact, highly integrated system was presented in [15] although in experiments the stack of cHT PEMFCs was not thermally coupled to the fuel processing units. In a small system fuel processing reactors are necessarily extremely proximate, and yet must be allowed to attain unique temperatures optimal for their operation. Because elements in thermal contact tend to approach the same temperature a selection of suitable materials and a strategy for thermal insulation of the system is a necessity. Also, finding appropriate sealing materials is crucial. Relatively high temperatures and various fluids present in the system may pose certain problems for conventional sealing materials.

a) MSR combined with LT PEMFC that operate between 70 ‑ 90 °C; b) MSR combined with cHT PEMFC that operate between 130 ‑ 200 °C; c) MSR combined with nHT PEMFC that operate between 190 ‑ 300 °C. Two different pathways are used to more thoroughly examine the conceived systems. First, applies a zerodimensional model based on mass and energy balances of the systems. This is done by using Aspen Plus® where all the system model units are studied at their operating, isothermal and stationary conditions (example shown in Figure 1). Second, uses COMSOL Multiphysics® to simulate physical models of integrated systems (example shown in Figure 2). Here, heat transfer is also studied at stationary conditions where the volume average temperature of model units closely corresponds to the temperatures used in mass and energy balance calculations. The two pathways are interdependent and coupled together. The calculation process and modeling of the systems are explained more in detail in [17].

Figure 1: Flowchart of the system with LT PEMFC stack used in Aspen Plus modelling

Numerical model The objective of numerical model is to study and show the concept of compact and highly integrated system with conventional MSR (operating between 230 ‑ 250 °C) and PEMFC stack. All fuel processing reactors have planar configuration and are stacked into a compact form and arranged in a thermal cascade. Three possible system configurations are analyzed:

Figure 2: Conceptual design of the system with LT PEMFC stack used in COMSOL modelling

Yearbook 2014 Laboratory for Heat and Power

Mass and energy balances A model of PEMFC stack is based on User-2 model unit where a subroutine is written in FORTRAN programing language. Depending on the hydrogen flow into the PEMFC stack the model calculates the rate of power and heat production. Based on polarization curves from commercial suppliers [1, 2] the power density polynomial is calculated. Since no true data are yet available for nHT PEMFC the polarization curve of cHT PEMFC is used instead. Based on the power density curves and the active surface area A = 25 cm2 it is concluded that a stack of four PEMFCs is needed in all three cases to achieve gross electric power of 25 W. The heat produced by the PEMFC stack is calculated as a difference between available energy and the power produced by the stack. The available energy is calculated as a difference in enthalpy of incoming and outgoing streams. Hydrogen over stoichiometric ratio lH2 is equal to the specifications of suppliers, which is 1.2 for LT PEMFC and 1.4 for cHT PEMFC. For the nHT PEMFC lower lH2 = 1.2 is adopted since problems with adsorption of CO are expected to be smaller. All the systems are air cooled. Consequently, the excess air ratio lAIR is greater as specified by suppliers. To properly describe the methanol conversion as a function of temperature the kinetic model is used which was first proposed in [16] and later also used by [3]. A plug flow reactor (RPlug) model unit is used to simulate the MSR. The reaction equations are integrated into the unit in the form of Langmuir-Hinshelwood-Hougen-Watson (LHHW) kinetic model. The PROX reactor is modelled with RStoic model unit and used only in case of LT PEMFC system. It is adopted that the PROX reaction proceeds at 155 °C with excess oxygen ratio l02 = 2.5. It is also assumed that all of the oxygen present in the reactor is consumed and the selectivities for methanol, hydrogen and CO are 0.3, 0.3 and 0.4, respectively. Physical modelling of the systems The same modelling approach based on FEM is used in all three systems. Within the systems only heat conduction is taken into account because all heat exchanges with fluids are modelled either as heat sources or heat sinks. A volume heat source is used for PROX reactor and catalytic combustor but for each individual PEMFC a surface heat source is used. Vaporizer, MSR and convective air cooling of bipolar plates are modelled as volume heat sinks. A generic form of heat conduction equation (Eq. (1)) is used in COMSOL modelling: (1) where k represents thermal conductivity which is isotropic and Q represents heat sources or sinks. Natural convection is considered only in the case of heat transfer from the outer layer of insulation towards surroundings and in the case of LT PEMFC stack where stack is directly exposed to surroundings. A generic form of heat transfer equation (Eq. (2)) is used:

Laboratory for Heat and Power Yearbook 2014

(2)

where n represents normal vector of outer surface and h heat transfer coefficient. Due to heat conduction in plate-type components the temperature gradients appear. Therefore, to properly estimate the temperature of individual component its volume average temperature is used. Results The performance characteristics of integrated systems are summarized in Table 1. Since all input and output streams are at ambient conditions the calculated gross efficiencies in Eqs. (3) and (4) are based on higher heating value:

(3)

(4)

gives the efficiency of integrated PEMFC stack and MSR system while ηCOG gives the system efficiency after the exhaust flue gases are cooled down to 40 °C, where heat flow is used in an external process (e.g. using thermo-electric generators). The highest methanol consumption and consequently are achieved in the system with cHT PEMthe lowest FC. This is due to the fact that the system uses the highest lH2 among all. Even though the LT PEMFC stack uses the is lower smallest amount of methanol the achieved than in the case of the system with nHT PEMFC stack because additional methanol is required to provide heat for the MSR. As a result of low operating temperature considerably lower ηCOG is attained in case of the system with LT PEMFC stack. Due to higher operating temperature both systems with HT PEMFC stacks are able to use the heat of flue gases more efficiently and hence achieve higher system efficiencies. Operating temperature also affects lAIR. This is clearly seen in the case of LT PEMFC stack which requires substantially more air than both HT PEMFC stacks to sustain the required temperature. The ratio would be even higher if the LT PEMFC stack would not be allowed to cool by natural convection. This way considerable part of energy flow is lost, however, it is of no practical use to the system because all the processes take place at higher temperatures. Higher temperature difference between the cooling air and the HT PEMFC stacks results in more efficient cooling, hence lower lAIR is used. This means that the blower will use less power which will further help to enhance net system efficiencies. High excess air can also cause problems in reaching the operating temperature of the catalytic combustor. A viable solution is to bypass part of the air past the combustor as is done in case of LT and cHT PEMFC stacks. However, this adds certain complexity to both of the systems. Heat losses are the lowest in case of the cHT PEMFC stack because the system is fully insulated. Because of higher operating temperature, higher heat losses occur in case of the system with nHT PEMFC stack. Heat losses are

Integrated system

Stack temp.

Gross electric efficiency

Gross cogeneration efficiency

Methanol consumption

Methanol conversion

Excess air ratio

Heat losses

LT PEMFC

70 °C

39.2%

50.5%

7.34 ⋅10-5 mol s-1 a 1.64 ⋅10-5 mol s-1 b

99.0%

9.0

16.1 W c 3.6 W d

cHT PEMFC

180 °C

35.9%

79.2%

9.8 ⋅10-5 mol s-1

97.1%

7.4

3.4 W

nHT PEMFC

255 °C

42.2%

81.6%

8.3 ⋅10-5 mol s-1

98.3%

4.7

3.8 W

a used in the MSR; b used in the catalytic combustor; c uninsulated part of the system (natural convection); d insulated part of the system

the highest in the system with LT PEMFC stack because it is left to cool by natural convection. The key component for all modelled systems is the thermal insulation with extremely good insulating properties and resistance to compression forces. This enables the systems to achieve desired temperature distributions using relatively thin insulation layers; hence, it helps to keep the systems small and in compact form. It also acts as an electric and most importantly as thermal insulator of aluminum alloy end plates. Aluminum is a very good thermal conductor and would contribute to substantial heat losses if end plates were left to reach high temperatures. Experimental setup Since the development of nHT PEMFC is in its early stage we are not able yet to produce such fuel cells with sufficiently large power densities. On the other hand, Laboratory of Catalysis and Chemical Reaction Engineering located at National Institute of Chemistry Slovenia successfully synthetized LT catalyst for MSR. The laboratory also advised on the geometry and design of the MSR. Screening for gasket and thermal insulation materials revealed suitable candidates for cHT and nHT PEMFCs. Lower operating temperature of cHT PEMFC increases possibilities as more conventional materials can be used. Based on these facts, the decision was made to proceed with experimental coupling of the LT MSR and the stack of cHT PEMFCs. MSR and vaporizer micro-reactors have already been manufactured using laser cutting and welding of stainless steel sheets. The design of the system and blueprints for individual components were made in 3D modelling software. Commercially available and custom made gaskets have already been acquired. Works still in progress are strategy for thermal insulation of the system and selection of appropriate measuring equipment which will allow performing adequately accurate measurements. Conclusions The numerical model proves the concept of integrated MSR–PEMFC-stack systems and provides the basis for the design of experimental setup. The results show it is more reasonable to use HT PEMFC in the integrated system because it has advantages over LT PEMFC (no water management, no reformate clean-up, higher tolerance to CO) and also shows higher system efficiencies.

Since nHT PEMFC are still in development and a novel LT MSR catalyst is already available the decision has been made to thermally couple the LT MSR and stack of cHT PEMFCs. References [1] GORE, GORE® PRIMEA® MEAs, http://www.gore. com/en_xx/products/electronic/fuelcells/series_56_ mea_fuel_cells.html, Accessed on: 28.1.2014 [2] BASF, Celtec® P1100W, http://www.fuel-cell.basf. com/cm/internet/Fuel_Cell/en_GB/, Accessed on: 28.1.2014 [3] A.V. Pattekar, M.V. Kothare, Journal of Power Sources, 147 (2005), 116-127. [4] T. Kim, International Journal of Hydrogen Energy, 34 (2009), 6790-6798. [5] K.-F. Lo, S.-C. Wong, International Journal of Hydrogen Energy, 36 (2011), 10719-10726. [6] C. Pan, R. He, Q. Li, J.O. Jensen, N.J. Bjerrum, H.A. Hjulmand, A.B. Jensen, Journal of Power Sources, 145 (2005), 392-398. [7] J.D. Morse, R.S. Upadhye, R.T. Graff, C. Spadaccini, H.G. Park, E.K. Hart, Journal of Micromechanics and Microengineering, 17 (2007), S237. [8] C.R.I. Chisholm, D.A. Boysen, A.B. Papandrew, S. Zecevic, S. Cha, K.A. Sasaki, Á. Varga, K.P. Giapis, S.M. Haile, Interface, 18 (2009), 53-59. [9] A. Hindhede Jensen, L. Qingfeng, E. Christensen, N.J. Bjerrum, Journal of the Electrochemical Society, 161 (2014), F72. [10] W. Tong, K. Cheung, A. West, K.-M. Yu, S.C.E. Tsang, Physical Chemistry Chemical Physics, 15 (2013), 7240-7248. [11] E. Romero-Pascual, J. Soler, International Journal of Hydrogen Energy, (2013). [12] H. Chang, H.C. Chiang, Y.H. Chen, Y.Y. Chang, S.H. Cheng, Chemical Engineering Science, 74 (2012), 27-37. [13] R.S. Besser, International Journal of Hydrogen Energy, 36 (2011), 276-283. [14] K. Shah, R.S. Besser, Journal of Power Sources, 166 (2007), 177-193. [15] D. Wichmann, P. Engelhardt, R. Wruck, K. Lucka, H. Köhne, ECS Transactions, 26 (2010), 505-515. [16] B.A. Peppley, J.C. Amphlett, L.M. Kearns, R.F. Mann, Applied Catalysis A: General, 179 (1999), 31-49. List of publications [1] A. Lotrič, M. Sekavčnik, S. Hočevar, Journal of Power Sources, 270 (2014), 166-182. [2] A. Lotrič, S. Hočevar, Methanol steam reformer - high temperature PEM fuel cell system analysis, Military Green 2012, Brussels, 2012, 115-126 Yearbook 2014 Laboratory for Heat and Power

Catalytic Biomass Gasification In Fluidized Bed With Superheated Steam Abstract → The summary article presents the scientific work of a young industrial researcher between years 2008 and 2014. It summarizes the optimization of catalytic biomass gasification in fluidized bed by minimizing the flow of superheated steam into the gasification reactor. We have proposed a new modal identification-control method for detecting and controlling fully fluidized conical beds with a wide size distribution of particles in the vicinity of the minimum fluidization velocity, irrespective of the composition of the bed, its temperature and density. The method was experimentally tested. A

mathematical model of Fast Internal Circulating Fluidized Bed (FICFB) biomass gasification with total power of 1 MW was developed. Before making a pilot system the particle dynamics were studied in a scaleddown cold-flow laboratory unit. The pilot gasifier was scaled-up and built on the basis of the cold-flow model study. A study of thermodynamics and syngas analysis was carried out. Theoretical results were compared to the experimental. By reducing the flow of superheated steam into the gasification reactor lower energy consumption was achieved.

Introduction Jernej Mele, Ph.D. Laboratory Laboratory for Heat and Power E-mail melejernej@gmail.com Company Bosio d.o.o., Obrtniška cesta 3, SI-3220 Štore Phone +386-31-799-674 Status Former PhD student (Finished post-graduate study: June 2014), Head of research and development in company Bosio d.o.o. Research area Biomass gasification, pyrolysis, fluidized beds Mentor Assoc.Prof.Dr. Andrej Senegačnik, Co-Mentor: Ass.Prof.Dr. Dušan Klinar

The present research work is focused in the area of gasification of solid fuels in fluidized beds with superheated steam. Gasification system FICFB (Fast Internal Circulating Fluidized Bed) in Güssing (Austria) with a total output of 8 MW (2 MWe and 4.5 MW thermal power) is one of the state-of-the-art gasification technology. This technology achieves 6,000 hours of annual operation and the annual operating cost of the system presents approximately 15 % of investment costs [1]. Gasification technology can be combined with cogeneration units, whereby a relatively high efficiency of energy conversion can be achieved. It can also reduce the consumption of fossil fuels by co-use of product gas in industrial processes [2]. While researching the 1 MWth FICFB process, certain questions concerning particle dynamics in gas flows arose. The gas-solid flow of FICFB is almost impossible to precisely model either analytically or numerically. Building a pilot plant is very expensive and there is no certainty that the pilot apparatus will perform the planned task. Therefore, building a much cheaper, small-scale cold flow model in order to design a

Laboratory for Heat and Power Yearbook 2014

large scale system more confidently can be economically justified. The FICFB gasifier, studied and discussed in this article, is divided into two fluidized bed reactors: gasification and a combustion reactor (riser). The solid fuel is fed into the gasification reactor and fluidized with superheated water steam. The bed material together with some remaining char moves to the riser though the chute. In the riser, the remaining carbon burns and heats up the bed material. Hot bed material is pneumatically transported through the riser to the cyclone where it is separated and collected in a siphon. The siphon fully filled with hot bed material acts as a gas barrier between reactors. Hot bed material supports the endothermic gasification reactions with heat. A higher solid circulation rate lowers the temperature difference between the gasification and the combustion zone. Moreover, higher solid flux between the fluidized beds conveys more char from the gasification to the combustion reactor, which reduces the required amount of additional fuel [3]. A dual fluidized bed system allows the treatment of separated gas streams with the same circulating solid and the production of a high-grade product gas that is practically nitrogen free.

The described system is patented under European patent EP 2 146 143 A2 [4]. A more detailed description of this process is given in Hofbauer et.al. (2002) [1]. The particle dynamics of such a system is essential due to design and operation. Some studies of CFB loop predictions can be found in literature [5, 6]. The most commonly described approach in the literature to achieve particle dynamic and reactive similarity in two fluidized beds is to use sets of dimensionless numbers which have to be kept constant in both scales. These sets of dimensionless numbers have been modified and extended by several authors in order to provide a more integrated view of the diverse phenomena taking place in fluidized bed reactors. Objectives and methods By increasing the flow of superheated steam in the gasification reactor, almost linearly decreases the chemical efficiency of gasification [7]. To maintain fluidized bed, a higher flow of superheated steam is needed than for the complete gasification of solid fuel. The moisture in the product gas is therefore high and has to be removed. The main objective is to optimize the process of biomass gasification in fluidized bed by minimizing the flow of superheated steam and achieve two favorable impacts from the standpoint of quality and cost of operation: I) less energy for superheated steam generation is needed, II) moisture in the product gas is reduced. We have designed a system for detecting and monitoring fluidized beds, while minimizing the inlet of superheated steam. The main result of the research is a new method of controlling gas-sand conical fluidized bed with wide size distribution of particles. Very complex FICFB gasifier has been constructed and built, which was also an outstanding engineering challenge and achievement. Doing that a mathematical model for scaling-up each section of the dual fluidized bed system was designed to study the operational behavior, the pressure profile of the circulation loop and the solids circulation rate, as well as for detailed engineering purposes. Evaluation of the model was carried out with experimental data obtained using a cold flow model and then compared to the operational pilot plant. In this way, the operating parameters were studied in detail and the key parameters were determined. The key point and novelty of our study is that the scale-up criteria compares the units for different fluidized bed regimes and considers the change in the process parameters. Stable particle circulation can be achieved with different pressure profiles but with a high impact on the thermo-chemical part of the process.

Cold-flow model First and basic engineering plans were made for the 1 MWth pilot gasifier. To avoid the high costs of probable corrections to the pilot system the decision was made to test the process in a small-scale cold-flow unit. Therefore, the laboratory unit was designed and built in order to simulate the hydrodynamic process of FICFB gasification with air under arbitrary conditions. It was made from stainless steel and glass, so that the particle behavior could be observed. Fig. 1 shows a laboratory unit. The dimensions of the cold flow laboratory unit were determined by downscaling an industrial sized 1 MWth system by applying the scaling criteria of Glicksman (1984) [12]. Accordingly, the cold-flow model was operated with ambient air and lower B Geldart’s particles of quartz sand as the bed material.

Figure 1: Laboratory unit

FICFB pilot plant The pilot plant (Fig. 2) was designed to take into account the heavy thermal and abrasive conditions which occur in it. Gas chambers are made of heat resistant stainless steel and protected by thermo concretes so that the hot gas-solid flow does not trite the steel walls. The main gas inlets are superheated water steam and preheated air; the main outlets are hot flue gas from cyclone and syngas from the reactor. The 1 MWth FICFB pilot plant was designed on the basis of a study carried out on a cold-flow model. At the end both systems were tested. The scale-up model was evaluated by comparing these two systems.

Experimental work The experiments were performed in cold-flow model, 1 MWth pilot gasifier that operates in Celje, Slovenia. To achieve hydrodynamic similarity between the cold-flow model and the pilot plant, the full set of Glicksman scaling rules [8] were applied.

Figure 2: Pilot plant in operation

Yearbook 2014 Laboratory for Heat and Power

Table 1.: Technical and process data for the cold-flow model and the pilot plant reactor Parameter

Symbol

Cold-flow model

Pilot plant

Reactor diameter at gas entrance

dreak,1 [mm]

100

300

Reactor diameter above the fluidized bed

dreak,2 [μm]

190

500

Conical bed angle

α [°]

Gas distributor type

Sandwiching nets

Bubble cap

Quartz sand

100–300

400–600

Bed material Particle size

dp [μm]

Particle density

ρp [kg/m3]

2650

Bulk density

ρp,b [kg/m3]

1575

1550

Stationary bed height

L [mm]

130

400

Air

Steam* Syngas**

Fluidization medium Temperature

Tg,reak/Tsyn[°C]

Density [17]

ρg [kg/m3]

Dynamic viscosity [17]

ηg [Pas]

550

780

1.124

0.288

0.192

1.8·10-5

3.1·10-5

4.6∙10-5

Volume flow of fluidization medium

ΦV,reak

[m3/h]

31.1

517.1

4275

Gas velocity (experimental)

vg0 / vg1 [m/s]

1.11

2.03

6.05

Gas velocity (predicted)

vg0 / vg1 [m/s]

0.397

1.40

1.66

Pressure drop (experimental) p2,3 [mbar]

13.5

47.8

p2,3 [mbar]

13.9

42.6

Pressure drop (predicted)

Figure 3: Characteristic dimensions of a reactor

* Gas inlet at the bottom of the reactor; ** Gas outlet above the fluidized bed in the reactor Table 2.: Technical data for the cold-flow model and the pilot plant combustion cone Parameter

Symbol

Cold-flow model

Pilot plant

Diameter of combustion cone below diffuser

dcomb,1 [mm]

140

Diameter of combustion cone above diffuser

dcomb,2 [mm]

190

Height of combustion cone

hcomb [mm]

1350

4880

Angle of diffuser

γ [°]

Gas distributor type Bed material

Sandwiching nets

Nozzles

Quartz sand

100—300

400—600 2650

Particle size

dp [μm]

Particle density

ρp [kg/m3]

2650

Bulk density

ρp,b [kg/m3]

1575

1550

Gas-oil burner

Burner type Number of burners Volume flow of combustion air

ΦV,co,a [m3/h]

524

diesel

ΦV,a,fuel [L/h]

11.1

Auxiliary fuel Total auxiliary fuel consumption Fluidization medium

Air

Air*

Temperature

Tg,comb [°C]

900

Density [17]

ρg [kg/m3]

1.124

1.09

0.294

Dynamic viscosity [17]

ηg [Pas]

1.8·10-5

3.8·10-5

4.7·10-5

Volume flow of fluidization medium

ΦV,comb [m3/h]

698

5913

Superficial gas velocity (experimental)

vc0 / vc1 [m/s]

9.8

12.6

57.9

Superficial gas velocity (predicted)

vc0 / vc1 [m/s]

3.677

8.96

7.24

Pressure drop (experimental)

p5,6 [mbar]

0.7

4.0

* Gas inlet at the bottom of combustion zone; ** Gas outlet at the top of combustion zone

Laboratory for Heat and Power Yearbook 2014

Flue gas**

Figure 4: Characteristic dimensions of the riser

Figure 5: Comparison of analytically and experimentally acquired chemical composition of product gas

Conclusions Modern gasification systems use fluidized beds. These provide a more homogenous temperature profile in the bed and better heat transfer, as well as allow the possible use of the catalyst as the bed material and allow the use of a gasifying fuel with a wide size distribution of particles. At the outset, we carried out basic engineering procedures for the actual 1 MWth FICFB gasifier. Due to high cost that would be incurred if there were errors in our model, the decision was made to test the process in a low cost small-scale unit. The unit was designed according to full set of Glicksman scaling laws. Similarity between units was achieved by using different particle sizes as the bed material. Therefore the assumption of particle flow similarity is based on a direct comparison of particle Reynolds numbers, Froude numbers, the ratio between gas ρg and particle ρp density, the particle size distribution (psd), the ratio between particle dp and the reactor diameter D, the particle sphericity ϕS, superficial gas velocity vg, dynamic viscosity ηg as well as the geometric similarity in terms of reactor shape, diameter, height, etc. Chemical reactions during gasification and combustion cause variations in temperature, density, and dynamic viscosity, all of which affect particle dynamics. It is practically impossible to create similarity by equalizing all the criteria in the scale-up set. Even though, if we would have been able to equalize all the scale-up parameters, it would be still made on assumptions, e.g. the composition of syngas is predicted on the basis of a mathematical model until the chemical composition is known. When the process was stabilized and a smooth circulation was established, the pressure relations were as expected and all of the gas flows were moving in the appropriate directions. We have acquired experimental data of chemical composition of product gas and compared it to mathematically predicted values (Fig. 5). The scale-up protocol and new modal method to control of gas-sand conical fluidized bed with wide size distribution of particles were experimentally tested with the successful demonstration of the FICFB pilot plant located in Celje, Slovenia.

References [1] Hofbauer, H., Rauch, R., Löffler, G., Kaiser, S., Fercher, E., Tremmel, H. (2002). Six years experience with the FICFB-gasification process. 12th European conference and technology exhibition on biomass for energy, Industry and Climate Protection; Amsterdam, June 2002 [2] Senegačnik, A., Mele, J. Uplinjanje lesne biomase perspektivna tehnologija za večjo uporabo obnovljivih virov tudi v industriji. 15 let na poti energetske odličnosti : zbornik. Ljubljana: Časnik Finance, 2013. 81-87. [3] Kaiser, S., Weigl, K., Schuster, G., Tremmel, H., Friedl, A., Hofbauer, H. (2000). First world conference and exhibition on biomass for energy and industry, Sevilla, June 2000. [4] Bosch, Klaus, Verfahren und Vorrichtung zur Erzeugung eines stickstoffarmen bzw. nahezu stickstofffreien Gases, EP 2 146 143 A2, filed Jul 16, 2009 and issued Jan 20, 2010. (In German) [5] Kaiser, S., Weigl, K., Aichernig, C., Friedl, A., Hofbauer, H. (2001). Simulation of a highly efficient dual fluidized bed gasification process. 3rd European Congress on Chemical Engineering, Nurnberg, June 2001. [6] Rüdisüli, M., Schildhauer, T. J., Biollaz, S. M. A., Van Ommen, J. R. (2012). Scale-up of bubbling fluidized bed reactors – A review. Powder technology, 217, 2012, 21-38. [7] Schuster, G., Loffler, G., Weigl, K., Hofbauer, H.. Biomass steam gasification - an extensive parametric modeling study. Bioresourse Technology 77 (2001): 71-79. [8] Glicksman, L., R. (1982), Scaling Relationships for Fluidized Beds. Chemical engineering science, 39, 1373-1384. List of publications [1] Mele J., Golobič I., Senegačnik A., A method to detect and control fully fluidized conical beds with a wide size distribution of particles in the vicinity of the minimum fluidization velocity. Thermal science, ISSN 0354-9836, 2014 [2] Mele J., Oman J., Krope J. Scale-up of a cold flow model of FICFB biomass gasification process to an industrial pilot plant - hydrodynamics of particles. WSEAS transactions on fluid mechanics, Jan. 2010, vol. 5, iss. 1, str. 15-24. [3] Mele J., Senegačnik A. Teorija podobnosti pri načrtovanju uplinjevalnika biomase v lebdečem sloju. V. Voršič, Jože (ur.). Komunalna energetika : oskrba z energijo. Maribor: Fakulteta za elektrotehniko, računalništvo in informatiko, 2013, str. 1-12 [4] Mele J., Senegačnik A. Minimal velocity of full fluidization a new determination method. 21. Mednarodno posvetovanje Komunalna energetika, 15. do 17. maj 2012, Maribor, Slovenija. Voršič, J. (ur.). Proceedings. Maribor: Fakulteta za elektrotehniko, računalništvo in informatiko, 2012, str. 1-9

Yearbook 2014 Laboratory for Heat and Power

Active Energy Network with Combined Heat and Power Systems and Hydrogen Technologies Abstract → The thesis deals with numerical design of active energy networks. At the beginning influencing factors and their interactions are presented. Planning is performed by five-step decision-making method. The method comprises defining user demands, availability of energy sources and possible network configurations. The last step provides an optimal configuration after decision-making parameter is defined. Quasi-dynamic model of the test network case has been assembled of energy consumer, production

units and saving capacities. Operation of the individual components has been performed by the network management module, which works on the principle of a matrix. The whole model has been implemented using the software package Matlab Simulink. The model allows the simulation of operating, the optimization of individual configurations according to WSM method and sensitivity analysis of different scenarios.

Introduction Andrej Pirc Laboratory Laboratory for Heat and Power E-mail andrej.pirc@vipap.si Room / Phone +386-31-794-554 Status PhD. (started: October 2009, completed: March 2014) Research area Active Energy Network Mentor Prof.Dr. Mihael Sekavčnik

Due to growing global energy demands, the world’s reserves of fossil fuels are steadily decreasing, while their price is increasing. These facts, together with the known environmental issues relating to the use of fossil fuels, are expanding the interest in using alternative sources of energy for the production of electricity and heat [1], [2]. The future development of energy systems will see the more widespread use of renewable energy sources, which will require a significant change to the structure of future electricity networks [3], [4]. Power production from renewable energy sources usually depends on the local weather conditions and is only partially predictable. The resulting fluctuations in supply are not negligible and have a significant impact on the stability of the entire energy system [5], [6]. Due to the partial unpredictability of the power supply, an appropriate technology for energy storage should also be included in the system [7], such as hydrogen technologies. A complex energy network, such as a system for energy supply, storage and consumption, requires advanced planning [8], [9]. In the process of

Laboratory for Heat and Power Yearbook 2014

planning, the following facts should be taken into consideration: • energy source (sort, time and quantity availability), • energy consumer (time and quantity-dependent consumption), • energy production units (efficiency, regulation, operating properties), • economic background and effects on the environment. Five-step decision-making method The goal of the presented research was to develop a method for finding the optimal system configuration that satisfies the given requirements and takes into consideration the specific boundary conditions. The method itself should be as general as possible in order to provide applicability to a broad range of different combinations of energy requirements and energy availability. The procedure is divided into several steps. 1. step: definition of the energy consumers that can require various forms of energy – electrical energy, heat and/ or a secondary form of energy, e.g.,

hydrogen. In general, the energy consumption is dynamically varying. 2. step: definition of the available energy sources with respect to the actual geographical location of the consumers or the energy system. The energy sources can be conventional (natural gas, gasoline, etc.) and renewable (sun, hydro, wind, wood, biogas, etc.). 3. step: selection of the particular energy-system configurations that will be analysed. 4. step: selection of the decision parameters that will be used to find the optimal configuration from the analyzed ones. The decision parameter can be the levelized cost of energy, the renewable fraction, the operating and maintenance properties, and the initial cost. 5. step: simulation of the selected system configurations and comparison of the decision parameters. The optimal configuration is selected based on the results of the system simulations using the weighted-sum method (WSM) [10].

controller to respond appropriately to every situation. The controller must monitor the values of the observed parameters, find in which of the predefined ranges each of the parameters fits and locate the appropriate cell in the system-state matrix according to the parameters’ values. The contents of the cell describe the actions that will, in a given situation, provide the optimal response and operation of the system, Fig. 1.

Energy-system configurations The sizing of the particular production facilities was chosen during the optimisation process. The particular system components were set up for the specific power consumer due to energy-source availability, considering the following: • a stable and secure energy supply throughout the entire operating period, • the operation of particular production facilities (i.e., weather conditions, water level in the accumulation reservoir, charging and discharging of the battery), • the minimum sizes of particular production facilities with regards to the initial capital, • the maximum utilization of the renewable source of energy, • the minimum operating costs and the levelized cost of the electricity.

Figure 1: System regulation by means of the system-state matrix.

DECISIon PArAMEtErS Levelized cost of electricity (LCOE) A stable system operation should not be the only monitored parameter when selecting ‘the best’ configuration for the system. Financial aspects have a great (if not the greatest) influence on the energy system’s planning, too. The information that predicts the viability of an individual power plant is its own levelized cost of electricity. This includes the initial capital costs, the costs of operation, the maintenance, the fuel as well as the environmental taxes and the costs for system decommission. (1)

SYStEM-StAtE MAtrIx rEGULAtIon The regulation of a complex system depends on several representative parameters that sufficiently and uniquely describe the state of the system. Each measured parameter has a value Pi that fits in one of the Ri ranges of various sizes, where the number and sizes of ranges can vary by parameter. If N is the number of observed parameters, an N-dimensional matrix is formed with a size of R1 x R2 x … x RN. Each cell of the matrix represents one unique situation in which the system’s operation could be classified regarding the combination of the observed parameters. It is very important that all possible situations are considered in the system-state matrix and at the same time the matrix eliminates the possibility of the system’s state fitting in more than one situation. When the system-state matrix is constructed its cells need to be filled with the appropriate rules for responses of the system that will ensure optimal operation under a given set of conditions and will also retain the stability of the system. The system-state matrix and the corresponding list of rules are the information that will enable the system

Renewability factor (RF) The renewability factor could be the next decision parameter during the process of finding an optimal system configuration. This factor shows how ‘green’ is a particular solution and is presented as the ratio between the amount of electricity produced from renewable energy sources and the total amount of produced electricity. (2) Electricity-consumption production ratio (CPR) The electricity-consumption production ratio shows the amount of energy lost through the storage capacities. The output of the storage capacities is smaller than the input due to its charging and discharging efficiency. (3)

YEARBOOK 2014 LAborAtorY For HEAt AnD PowEr

Numerical model and results To test the observed system’s operation under the predefined parameters and rules, a numerical model of the system was set up using MathWork’s Simulink software [11], while Matlab was used to import the data and set up the operating parameters and rules. The numerical model of the system was given a one year time series of power consumption and the response of the system was observed. Although the entire time series was known in advance, the controller’s decisions were only based on past data. The main task of the controller was to supply a sufficient amount of power to cover the current load by starting/stopping different energy systems and charging/discharging saving capacities. Results are given in numerical values (decision parameters, costs, fuel use, emissions…) and as operating diagrams (i.e. Fig. 2) for each element of energy network.

includes only two energy system (S1 and S2) can be determined the optimum size of the two systems, so as to record the combination of the two energy systems in the form of a table (11, 12, ..., 45, 55), as shown in Fig.3.

Figure 3: Diagram of sensitivity analysis for power optimization.

These combinations may be arranged as graphical in x and y axis. Calculated parameters of sensitivity analysis (i.e. levelized cost of electricity) are applied to the z axis of a 3D chart. Values, which both correspond to ‘low-power’ system (such as 11, 12, 21), usually do not satisfy a stable supply of energy; values which correspond to ‘largepower’ (such as 45, 54, 55) are oversized. The calculated parameters of other combinations represent a stable power supply. The optimal configuration accords to the lowest point (33 in Fig. 3) of represented area. Conclusions

Figure 2: Energy-system operating diagram, one-week scale.

Optimization and sensitivity analysis The objective solution for optimal system configuration can be determined by multi-criteria decision making method (MCDM), where weighted sum method (WSM) is proposed. Particular criteria xi (i is number of criteria), of such as investor’s financial capabilities, levelized cost of electricity, local law on the minimum amount of use RES, are valued with factor wij (j is number of system configuration) according to their priority (the greater priority is valued with greater number). Optimal configuration is determined after finding the maximum sum of particular products A.

(4)

The numerical model allows us to make sensitivity analysis between and after simulation of the energy network operation too. It allows us to observe the impact of various parameters on the final results. It enables us to optimize the network configuration. Assuming that the analysis

Laboratory for Heat and Power Yearbook 2014

In this work a description of a numerical approach to designing an energy system is presented, where a five-step energy-system design procedure has been developed. The applicability of the developed procedure was demonstrated on a test case. The particular production units of the system were individually modelled (i.e., the photovoltaic power plant, the hydro power plant, the wind power plant, the internal combustion engine and the storage capacities). The energy system was controlled using a system-state matrix approach. The testing of the design procedure and the numerical modelling of the described energy system were performed with Mathwork’s Simulink code. Using system simulations, the optimal configuration of a selfsufficient energy system and its regulation can be found by considering different criteria or decision parameters. The selection of the decision parameters is very important since the choice of these decision parameters usually results in a completely different optimal output system. The results and conclusions of this test-case simulation cannot be generalized because of its specific input parameters, most notably the decision parameter. On the other hand, the principle of using a numerical approach for designing an energy system can be generalized and used for even more complex systems, where other approaches are not as transparent and straightforward.

References [1] Asmus, P.: Microgrids, virtual power plants and our distributed energy future, The Electricity Journal 23 (2010), p. 72 – 82. [2] Tuma, M., Sekavčnik, M.: Energetski sistemi, Fakulteta za strojništvo, Ljubljana, 2004, (in Slovenian). [3] Franco, A., Salza, P.: Strategies for optimal penetration of intermittent renewables in complex energy systems based on techno-operational objectives, Renewable Energy 36 (2011), p. 743 –53. [4] Hammons, TJ.: Integrating renewable energy sources into European grids, Electricity Power Energy Systems 30 (2008), p. 462 – 75. [5] Kanase-Patil, A. B., Saini, R. P., Sharma, M. P.: Integrated renewable energy systems for off grid rural electrification of remote area, Renewable Energy 35 (2010), p. 1342 – 1349. [6] Pavlas, M., Stehlik, P., Oralb, J., Sikula, J.: Integrating renewable sources of energy into an existing combined heat and power system, Energy 31 (2006), p. 2499 – 2511. [7] Prodromidis, G., Coutelieris, F.: Innovative energy storage for off-grid RES-based power systems: integration of flywheels with hydrogen utilization in Fuel Cells, Journal of Energy Engineering (2013). [8] Cormio, C., Dicorato, M., Minoia, A., Trovato, M.: A regional energy planning methodology including renewable energy sources and environmental constraints, Renewable and Sustainable Energy Reviews 7 (2003), p. 99 – 130. [9] Connolly, D., Lund, H., Mathiesen, B. V., Leahy, M.: A review of computer tools for analysing the integration of renewable energy into various energy systems, Applied Energy 87 (2010), p. 1059 – 1082. [10] Pohekar, S.D., Ramachandran, M.: Application of multi-criteria decision making to sustainable energy planning – A review, Renewable and Sustainable Energy Reviews 8 (2004), p. 365 – 381. [11] Mathworks: Simulink Help (2011). List of publications [1] Pirc, A., Sekavčnik, M., Drobnič, B., Mori, M.: Use of hydrogen technologies for saving electric energy in combination with renewable energy systems, 6 th International Workshop on Deregulated Electricity Market Issues in South-Eastern Europe, Demsee 2011. [2] Drobnič, B., Pirc, A., Mori, M., Sekavčnik, M.: A novel approach to the regulation of a self-sufficient energy system using a system-state matrix, International Journal of Electrical Power & Energy Systems 53 (2013), p. 893 – 899. [3] Lacko, R., Pirc, A., Mori, M., Drobnič, B.: Step By Step Numerical Approach to a Self-Sufficient Micro Energy System Design, Int. Journal of Engineering Research and Applications 4 (2014).

Yearbook 2014 Laboratory for Heat and Power

Prof.Dr. Branko Širok

Laboratory for Hydraulic Turbomachinery The laboratory is involved in general research in development of water and turbine machinery for various applications. The work is performed under funding of national in international projects and industry funded research. In the following we will present applications, processes and methods involved in laboratory activities.

Applications cover broad field of turbine machinery, where we focus on aerodynamic and hydrodynamic processes; among them are water turbines, large domestic appliances, small domestic appliances, pumps and fans for special purposes or applications, HVAC fans, suction units, aerodynamic tunnels design, collection chambers for mineral wool and glass wool productions, drying chambers, polymerization chambers, cooling applications, noise improvements of turbine machinery, agricultural spraying machinery, gas transmission systems, wastewater treatment systems etc. Among the processes we study cavitation and cavitation erosion, turbulence, image processing, risk prediction etc. Processes are in most cases related to turbine machinery applications. Among others, we pioneer cavitation process for the use in wastewater treatment and water processing applications and for modeling of cavitation erosion in turbine machinery and precision spraying for agricultural applications. Above mentioned applications and processes are addressed in a variety of ways using numeric modeling, experimental modeling, laboratory testing or field testing. In the laboratory dedicated equipment is available. Available are several laboratory measurement stations, high performance computer, modeling software, fast black and white camera, fast thermal camera, various data acquisition systems, sensors, scanners, actuators, analysers etc. Work for industry includes cooperation with small and medium enterprises and large international companies. Among our clients from abroad were Philips, Saint Gobain, Auma, Gamma Meccanica, Uralita, KSB, Electrabel, Knauf Insulation, Bosch Siemens and many other Slovenian companies, Gorenje, Hidria, Skupina Kolektor, Soške Elektrarne, Termoelektrarna Šoštanj, Izoteh, Domel, TIP95, LIMOS, Turboinštitut etc.

We have in the past participated in several international projects, 5th, 6th and 6th EU Framework projects, CORNET, European Space Agency, EUREKA and COST projects and several national projects. We are open for future cooperation and application to EU projects. In the Laboratory for Hydraulic Turbomachinery we try our best to join above mentioned research and development activities with teaching activities. Students are able to work during laboratory exercises with modern measurement equipment. Students are invited to participate in the research and development activities for research projects and industrial projects during diploma and M.Sc. thesis seminary work. Also in the future we will aim to make teaching process more efficient and appealing to students.

Quantitative risk assessment on transmission network for natural gas Abstract → Ever increasing length of transmission pipelines for natural gas that also spread into urban areas has spawned the need for safety and assessment of risk, which is posed on the environment by such a gas transmission network. In order to achieve reliable, safe enough and continuous operation of transmission system for natural gas, its influence on environment should be minimized. Apart from that, early recognition of less reliable parts of the system that endanger the operation of the gas network is desirable as well. Therefore, certain tools for risk assessment and risk management should be applied. In this way, the system operator is able to monitor continuously the operation of the system as well as to detect potential risks due to technical and operational issues or environmental

influences. Such tools usually present a part of an integral system for safety management, which serves for safety management of employees and public, protection of urban, natural or industrial environment as well as for operational reliability of natural gas transmission network. Risk is usually assessed by special models, which should be developed in accordance with standards and legislation. Their main task is to model consequences of hazardous events, to predict the frequency of such events and to assess risk with the ability to incorporate different risk mitigation measures. Particularly important is the method of quantitative risk assessment that enables the magnitude of risk to be assessed as a numerical value.

Introduction Tom Bajcar Laboratory Laboratory for Hydraulic Turbomachinery E-mail tom.bajcar@fs.uni-lj.si Room N-4 Phone +386-1-4771-422 Status PhD researcher Research area Risk Assessment

The age of existing natural gas networks is getting higher. At the same time, more and more new high pressure pipelines are introduced due to ever rising energy needs in EU countries. Accordingly, the risk that these systems pose on the environment in their vicinity rises as well. In order to keep the risk in acceptable levels, a system or model for risk assessment and risk management should be applied whenever new pipelines are laid close to existing inhabited buildings or vice versa (i.e. when new building are to be built close to existing pipelines or pipeline objects). Risk is generally defined as a measure for likelihood and severity (injuries or harm) of an event. It can be therefore written as [1, 2]: Event risk = Event frequency × Event consequences

(1)

Individual risk is annual probability that a person in the vicinity of a hazardous object dies due to potential hazardous events on that object. On the other hand, societal risk represents the annual expected number of 44

Laboratory for Hydraulic Turbomachinery Yearbook 2014

casualties due to hazardous events. Safety results from the judgement of acceptability of risk: an activity is assessed as safe if the level of its risk is assessed as acceptable. General acceptability criterion (= upper acceptable margin) in many EU countries for individual risk due to natural gas network is 1.10-6/year. Acceptable risk levels are usually provided by national legislation. Risk assessment procedure generally consists of the steps shown in Fig. 1. If the assessed risk is too high, it should be reduced with the use of different additional risk mitigation measures or (rarely) by changing operational parameters (“what if ” scenarios). The whole above procedure is then repeated until the acceptable risk level is reached. A general concept of a model for individual risk assessment on transmission pipeline network for natural gas is presented below. It comprises quantified risk assessment (QRA) on pipelines and on other gas transmission objects (such as metering-regulation stations - MRS). The model is a result of broad European experiences as well as domestic (local) knowledge.

Introduction of risk mitigation measures

Selection of pipeline segment / object

Input parameters

Determination of hazardous event consequences

Mathematical models or numerical simulations

Determination of hazardous event frequencies

Historical databases or local models

The length L rises together with the pipeline diameter and natural gas pressure. The amount of escaped natural gas from a damaged pipeline can be determined analitically or through numerical simulations using computational fluid dynamics (CFD) methods (as shown in the example in Fig. 3).

Quantitative risk assessment

Risk too high Risk evaluation Risk acceptable Monitoring of risk on different segments of transmission system

Figure 1: Schematic algorithm of risk assessment procedure.

Risk analysis Determination of event consequences The most unwanted event on natural gas transmission network is the uncontrolled gas leak and ignition of escaped gas due to failures on the system. Consequences of such a hazardous event are usually heat radiation of burning gas jet (in unobstructed areas) and/or explosion of flammable mixture of natural gas and air (in obstructed areas); they both cause negative effects on inhabitants in the vicinity (injuries, casualties). Heat radiation of burning gas jet generally occurs in the case of pipeline failure, while explosions potentially happen in enclosed spaces (objects such as MRS). Effects of heat radiation on people depend on heat flux and time of exposure to heat radiation [3]. According to past experiences [2, 3], the exposure time of 20 s is usually taken as an average time period for a person needed to find appropriate shelter. In this time period, the amount of radiative heat flux that surpasses 35 kW/m2 is deemed as 100% lethal, while 9.84 kW/m2 represents 1% lethality [2]. Table 1 shows approximate effects of different amounts of radiation heat flux. Radiation heat flux and its consequences diminish with distance from the radiation source; consequences can have effect on the recipient (person) up to a certain distance rh (Fig. 2). The distance rh defines the effective length L of pipeline, within which a hazardous event (HE) can still have detrimental effects on a person at a distance h from the pipeline.

Figure 3: Numerical (CFD) simulation example of velocity field of escaping natural gas through a hole (damage) in a high pressure gas pipeline.

In order to facilitate the calculations (and stay on the conservative side), one can assume that the heat flux source from a jet fire concentrates in one point at the ground surface level. Thus, the radiation heat flux density I can be determined by the following expression [4]: ,

where ε denotes the portion of radiation heat in all the heat during the process of combustion. Q denotes the mass flow rate of the leaking gas, Hc is the heat of combustion of natural gas and r is the distance from the heat source to the location of recipient. τa denotes the atmospheric transmissivity; it depends on the quantity of water vapour in the air [3]: ,

(3)

where pv denotes the saturation pressure of water vapour, h denotes the relative humidity and r the distance from the heat source. When a hazardous event is in the form of a jet fire, probit functions are usually applied for the assessment of the damage level. Probit functions connect the probability of death with the probability unit Pr; the latter comprises the relationship between the dose of heat and consequences on the recipient (death) [2]. Distribution of the probability of death P can then be assessed with the following equation: ,

Figure 2: Effective pipeline length L, where an unwanted event (burning jet fire and its heat radiation) can still affect the recipient (person).

(2)

(4)

where the argument Pr denotes the probability unit and x equals (Pr – 5)/σ with a standard deviation σ = 1. The probability unit Pr for heat radiation of a jet fire is determined as [2]: Yearbook 2014 Laboratory for Hydraulic Turbomachinery

(5)

where I denotes the radiation heat flux at the location of recipient and te the exposure time. Table 1: Effects of heat radiation. Effect

Radiation heat flux [kW/m2] â&#x2030;Ľ 35,0

Collapse of (concrete) buildings.

~15,0

Ignition of wood, smelting of plastics.

9,8

2nd degree burns on people after 20 seconds of exposure, approx. 1% lethality

Table 2: Incident data for natural gas transmission pipelines in central and western Europe between 1970 and 2010 (source: EGIG [5]). Risk source

Percentage of all incidents [%]

Third-party interference

48.4

Construction/material defects

16.7

Corrosion

16.1

Ground movement

7.4

~4,0

Pain after more than 20 seconds of exposure.

Hot-tap

4.8

1,4

No harmful consequences on people even after prolonged exposure (close to max. sun radiation heat flux on Earth).

Sum

Other

Explosion of natural gas â&#x20AC;&#x201C; air mixture occurs in obstructed (enclosed) space. Its effects (i.e. pressure waves) depend on the highest level of overpressure at the location of the recipient (person). There are many different approaches and models to assess the lethality of explosion; one of them is the so-called multienergy method, where there are two boundary pressure levels: p0.3 = 0.3 bar or more (100% lethality if exposed) and p0.1 = 0.1 bar or lower (0% lethality) [3]. Consequences of explosion diminish as well with the distance from the explosion centre; their magnitude depends mainly on the amount of the explosive mixture (i.e. the size of the explosive cloud) and its combustion heat. If the quantity of explosive substance (i.e. natural gas) is known, the distances from the center of explosion to both limit pressure levels can be calculated using the socalled multienergy method [3]: ,

(6)

(7)

where r0.3 and r0.1 are distances to pressure levels p0.3 and p0.1, respectively. pa denotes the ambient pressure (usually atmospheric pressure) and Hc denotes the heat of combustion of natural gas in the confined flammable mass me. Determination of event frequencies Hazardous event frequencies can generally be extracted from historical databases that contain information on already occurred past events [5, 6]. The event frequencies are affected by different risk sources that cause damage on natural gas transmission network. In the case of pipelines, these sources can be usually classified into one of the following groups: third-party interference (usually the risk

source linked with the highest hazardous event frequencies [5]), construction/material defects, corrosion, ground movements, hot taps and others (such as lightning strike etc.). The average percentage of pipeline incidents (damage to pipeline) in some EU countries that appertains to each risk source is shown in Table 2.

Laboratory for Hydraulic Turbomachinery Yearbook 2014

6.6 100.0

On the other hand, in enclosed secured gas installations (such as MRS) with the risk of explosion, the sources of risk comprise the inspection intervals of gas installations and the quality of ventilation [7]. When event consequences (Section 2.1) and event frequencies are determined, one can assess the risk on the selected section of the pipeline or pipeline object according to Eq. 1. Risk mitigation measures Mitigation of risk is needed whenever the assessed risk exceeds the acceptable value. In such cases, risk mitigation measures should be applied to chosen gas transmission network segments with unacceptable high risk. After that, the whole procedure of risk assessment should be repeated to see if the value of newly assessed risk lies within the acceptable range. Risk mitigation measures are usually applied in order to reduce the event frequencies rather than event consequences. If the transmission pipelines are in question, the most used risk mitigation measures comprise the inspection enhancement (aerial and ground inspection, pipeline pigging), warning enhancement (buried warning tapes and surface line markers above the pipeline â&#x20AC;&#x201C; Fig. 4) and mechanical protection enhancement (increased pipe wall thickness and protective slabs) [6, 8].

Figure 4: Warning sign for presence of high pressure natural gas pipeline: surface line marker.

While the mitigation of risk in closed installations (such as MRS) is possible by more frequent inspections, it is usually more efficient to enhance the ventilation in such objects; the consequence is lower probability of occurrence of explosive atmosphere [7]. If these measures are not enough, the risk levels can be further lowered by installation of protective walls around the risky object. The function of these walls is to obstruct the propagation of pressure waves and thus to shorten the range of explosion consequences. Results Following the above described steps the model can be able to assess individual risk due to transmission network for natural gas. With the appropriate input of all required parameters (together with pipeline coordinates and locations for particular gas installations, the final result can be seen in Fig. 5. The nature of the model, as presented in Fig. 5, can serve for risk-mapping the whole desired area and can further be enhanced by connections with geographic information system (GIS).

500

Transmission pipeline 1

400

Regional pipeline 1

200 100 0

Residential object

Regional pipeline 2

-100 -200 2.1

x (m)

2.17

2.16

2.15

x 10 x 10

200

1.4 e00 6

100

1.5e-006 1.4e-006

-200 2.1

2.11

1. 3e -0 06

06 e- 0 1.5 06 -0 e 4 1.

06 -0 5e 1.

7 -00 7e

1.3 e- 0 06 06 e- 0 1.2

9e-007

-006 1.1e06 1e -0 -007 8e 07 7e -0 -007 07 5e 4e -0

06 5e - 0 1. 4e 1.

6 00

1.1 e- 0 1.4 e-00 16.5 06e - 00 6 3e 5 1.5 -00e - e7007 09016.4 4e e1-0.e1- 0 -0 0e71-100.62 07 6 e0e-6- 1 e - 80e0.03 00 -606e 7 0- 00 7 6

7 00 e7 -00 07 8 6e e -0 -00-6006 7 e e 07 11.61 9e -0 e- 060 0 1.e2- 0 -006 e 3 1. 1.5

1e -007 3e -007 07 2e -007 07 5e -0 -0 4e 07 6e -0 7 07 7e -00 8e -007 8e -0 07 07 07 1e-006 6e -07 9e -0 9e -0 61.1e-006 e -0007 - 00 -006 07 0742e -0 .2e1.4e 106 5e -0 3e -0 1.3e-0

300

07 -0 06 9e e- 0 2 1.

2.12

07 -0 2e

2.13

2.14

x (m)

2.15

2.16

-7

14 12 10 8 6 4

07 -0 1e

-1 Tveganje za posameznika (1/leto) Individual risk (year )

400

1.4 e1.4 006 e- 0 06

6 7ee -00 -00 7 7 11.1e 8 e-006 e -007 00 6

500

-100

2.14

2.13

2.12

2.11

1.1 5e -0 1.2 e8e 07 -0 1e -0 006 e- 00 3e 06 6 -00 4 6e -70e71e 7 e -0 007007 9e 07 -00 00 7 2 7 e -0 07

y (m)

300

y (m)

Transmission pipeline 2

Measurementreduction station

Conclusions The main benefits from the presented concept of a model that enables the risk assessment on the whole transmission network for natural gas can be presented as follows:. • Ability of joint quantitative risk assessment on both, natural gas pipelines and natural gas installations (such as metering-reduction stations). • Apart from operational and constructive parameters, the layout of the pipeline network also plays important role in the presented model. • The model can show different iso-risk contours in ground projection, among others the one that represents acceptable risk margin. • Quantification and evaluation of mitigation measures effects on risk. • Ability to change gas network parameters („what-if “ scenarios). • Ability to produce risk maps in connection with GIS. References [1] IGEM – Institution of Gas Engineers and Managers (2008). Steel Pipelines and Associated Installations for High pressure Gas Transmission, Recommendations on Transmission and Distribution Practice. IGEM/ TD/1 Edition 5, Loughborough. [2] CPR 18E Purple Book (1999). Guideline for Quantitative Risk Assessment, Committee for the Prevention of Disasters, The Hague. [3] CPR 14E Yellow Book (2005). Methods for the calculation of physical effects, 3rd edition, Committee for the Prevention of Disasters, The Hague. [4] Jo Y.-D., Ahn B. J. (2005). A method of quantitative risk assessment for transmission pipeline carrying natural gas, Journal of Hazardous Materials A123, str. 1-12. [5] EGIG – European Gas pipeline Incident data Group (2011). Gas Pipeline Incidents 8th Report 1970-2010, Groningen. [6] Mather J., Blackmore C., Petrie A., Treves C. (2001). An assessment of measures in use for gas pipelines to mitigate against damage caused by third party activity, Contract Research Report 372/2001, Health and Safety Executive. [7] Bajcar, T., Cimerman, F., Širok, B. (2014). Model for quantitative risk assessment on naturally ventilated metering-regulation stations for natural gas. Safety Science, vol. 64, p. 50–59. [8] Bajcar, T., Širok, B., Cimerman, F., Eberlinc, M. (2008). Quantification of impact of line markers on risk on transmission pipelines with natural gas. Journal of loss prevention in process industry, vol. 21, no. 6, p. 613-619.

2.17 x 10

Figure 5: Presentation of individual risk due to natural gas transmission network on a specified location.

Yearbook 2014 Laboratory for Hydraulic Turbomachinery

Mineral wool melt fiberization on the spinner wheels Abstract â&#x2020;&#x2019; This paper presents the research work regarding the mineral wool fiber formation from the melt spun by a spinner wheel. Due to restrictions in the study of the real process, melt fiberization was modeled on a single wheel spinning machine supplied by cold Newtonian liquids. High-speed camera visualization was employed to conduct the measurements in different operating points defined by the variation of the wheel rotational speed, liquid flow rate and physical properties. Based on the analysis of acquired photographic material, we identified the typical spinner operating regimes and formed regression models for characteristic integral fiberization parameters such as the number, length and diameter of ligaments

and related flow structures. Additionally, ligament growth kinematics was studied in conjunction with the hydrodynamic stability of fiberization. Based on the qualitative and quantitative analysis results of the process, operating characteristics were compared to other types of rotary fiberization devices and to known theoretical models of the relevant hydrodynamic phenomena. In relation to the undesired phenomena of the industrial spinner wheel surface abrasion and the unfiberized material formation, we also investigated the mechanisms of the liquid film velocity slip against the wheel and the ligament end pinch-off, respectively.

Introduction Benjamin Bizjan Laboratory Laboratory for Hydraulic Turbomachinery E-mail benjamin.bizjan@fs.uni-lj.si Room N/5 Phone +386-1-4771-423 Status PhD student (started: October 2012, to be completed: December 2014) Research area Mineral wool production Mentor Prof.Dr. Branko Ĺ irok, Co-Mentor: Ass.Prof.Dr. Alen OrbaniÄ&#x2021;

Mineral wool is an inorganic, nonflammable insulation material with low heat and noise conductivity as well as low density. There are several different types of mineral wool, including the rock- (stone-), glassand slag wool. Among these types, the rock wool has the most desirable mechanical properties (good strength of insulation panels and a very high melting point), but is also the most difficult one to manufacture. Rock wool melt fiberization is usually achieved by the means of a spinner wheel where the melt stream flows onto the mantle surface of a rotating spinning wheel, is drawn in motion and scattered to fibers. Mineral wool quality (e.g. fiber diameter and length, mass fraction of unfiberized material) largely depends on the quality of the fiberization process on spinner wheels. However, the accurate modeling (either analytical, experimental or numerical) of the fiberization process is extremely difficult due to complex multiphase hydrodynamic phenomena, high process temperatures and a large variety of relevant length scales. The fiberization modeling has so far been restricted to integral experimental

Laboratory for Hydraulic Turbomachinery Yearbook 2014

regression models [1] and simplified numerical models not dealing with hydrodynamics and solidification [2]. Therefore we decided to perform an experimental study on a model spinning wheel, observing liquid ligament formation from the liquid film as a process very similar to the initial phase of fiber formation. Theoretical background Industrial spinner operation The spinner (i.e. the melt fiberization device) is located between the cupola furnace where melt is prepared and the collection chamber where the primary layer of mineral wool is formed. On a spinner wheel, there are three main regions of interest for (Fig. 1). Region A denotes the melt stream flow onto the wheel up to the impingement point. The impingement point position and melt flow fluctuations significantly affect the formation of the thin melt film (region B). The film forms in a radial direction and is scattered to molten ligaments under the combined action of the inertial, aerodynamic and surface tension forces. The ligaments are then transported away from the wheel

Experimental setup

Figure 1: Left: Visualization image of the first spinner wheel; right: fibers under the microscope

(region C) where they solidify to mineral wool fibers and are torn away from the melt film. In our study, we focused on regions B and C. Theory of hydrodynamic instabilities When a liquid stream falls onto the surface of a spinning wheel with radius R and rotational speed f0, it is drawn in motion by the viscous and adhesive forces, forming a thin film of thickness h and width B (Fig. 2). Velocity slip between the film and the surrounding air causes the development of initial (embryonic) disturbances by the Kelvin-Helmholtz instability. The main wave formation mechanism however is the Rayleigh-Taylor instability [2, 3]. It occurs due to the fact that the denser fluid contained in the rotating liquid film is pushed towards the less dense fluid (surrounding air) by centrifugal force acting upon the film. Consequently, unstable waves are formed from the initial film surface disturbances.

Subsection Experiments were conducted on a model spinning wheel with 45mm radius spun by an electric motor (Fig. 3). Spinner wheel rotational speed was varied between 5Hz and 30Hz. Three different glycerol-water mixtures (85%, 75% and 60% glycerol concentration) were used as working media and supplied at two different flow rates Q (1.63mL/s and 3.27mL/s) through a nozzle with 3mm diameter. As the liquid stream impacted the wheel mantle surface, it first formed a radial film which then disintegrated to ligaments and droplets.

Figure 3: Experimental setup for the visualization of ligament formation and breakup

The model ligament formation process was visualized in 36 different operating points using a high speed camera (Fastec Hispec 4 mono 2G). Observed liquid structures were illuminated from behind by diffuse coaxial light from a ring of light emitting diodes. In each operating point, images were recorded in one large window at 652 frames/s and in two smaller windows at 4400frames/s (Fig. 4). Image resolution was 0.11 m per pixel.

Figure 4: An example of recorded liquid disintegration images in different acquisition windows

Results Figure 2: Simplified presentation of ligament formation

The unstable waves grow at different growth rates (depending on their wavelength 位) and the fastest growing wave becomes predominant [3]. The spacing between these waves (s in Fig. 2) is transformed into the spacing between ligaments s. It is increased by liquid viscosity and reduced by the surface tension. Ligament spacing has been determined for spinning discs [3] and cups [4] [5], but not for spinning wheels as in Fig. 2. Under the action of centrifugal and Coriolis force, the ligaments are formed from unstable waves and elongated up to the point when they pinch off or disintegrate. In case of melt as a working medium, the ligaments usually cool rapidly and solidify to fibers [1]. However, if the cooling rate is insufficient or if the liquid is not cooled below its solidification temperature, a breakup of ligaments to droplets occurs [6].

Qualitative image analysis The first step of experimental data processing was a qualitative analysis of recorded images (Fig. 2). Liquid disintegration regimes (Fig. 2) were determined to be significantly different from those on centrally-fed rotary devices (e.g. spinning discs and cups). At low rotational speeds, there is no disintegration at all as all the liquid flows from the wheel in a single continuous stream. As f0 increases, the liquid flow becomes unstable and ligament formation begins, initially occurring intermittently and irregularly. When f0 is further increased, ligaments form continuously, but the formation of a single ligament is a transient phenomenon. First, a liquid bulge is developed on the film and slowly grows into a bulbous-like shape. From this liquid structure, the ligament is elongated at an increasingly rapid rate while also becoming thinner in Yearbook 2014 Laboratory for Hydraulic Turbomachinery

comparison of Weber number exponents suggests a formation mechanism fundamentally similar to spinning discs and cups.

Figure 6: Mean ligament number on the wheel perimeter as a function of dimensionless operating parameters

Another important parameter is the mean ligament diameter dL which should be as low as possible to attain optimal mineral wool properties. We have determined dL to significantly depend only on Weber number (Fig. 7), decreasing when We is increased.

Figure 5: Different liquid disintegration regimes on a spinning wheel (85% glycerol, Q = 3.27mL/s)

diameter. At a certain time, the bulbous ligament end (i.e. the head droplet) is pinched off from the ligament body. From this point onwards, the ligament continues to grow up to the point when its base is detached from the liquid film. This is quickly followed by a capillary breakup of the ligament. As the wheel rotational speed is further increased, the ligaments become thinner and more numerous and at a certain f0, liquid sheets start to form on a part of the wheel. When f0 is further increased, the mass fraction of liquid disintegrated through sheets becomes larger and the liquid film adhesion to the wheel surface is deteriorated up to a point when most of the incoming liquid stream is instantly deflected from the wheel. Quantitative image analysis In this step, characteristic properties of ligaments and other flow structures were studied as a function of the following dimensionless operating parameters:

Figure 7: Mean ligament diameter as a function of dimensionless operating parameters

Besides the diameter, the mean ligament length L is also a very important parameter for assessment of mineral wool quality. In a real industrial process the ligament length measurements are not possible due to high temperatures and a spatially highly complex fiber flow. Final fiber length measurement is also difficult due to the mutual intertwining of fibers and sample fragility. However, in our experiments the mean ligament length was relatively easy to measure and is shown in Fig. 8.

• Weber number: We = ρ(2πf0)2R3/σ (1) • Ohnesorge number: Oh = μ/(ρRσ)1/2 (2) • Dimensionless flow rate: q = (Q/B)*(ρ /Rσ) (3) In equations (1)-(3), liquid density, viscosity and surface tension coefficient are denoted by ρ, μ and σ, respectively. First, the mean number of ligaments N on the wheel perimeter was investigated. N was determined to rise with Weber number as well as liquid flow rate (Fig. 6). Ligaments may form in several parallel planes, but their number in a single plane is significantly lower as with spinning discs and cups with the same R and f0. Nevertheless, a 50

Laboratory for Hydraulic Turbomachinery Yearbook 2014

Figure 8: Mean ligament length as a function of dimensionless operating parameters

From Fig. 8 it can be seen that the ligament length significantly increases with the liquid flow rate and viscosity while the Weber number has little effect on L. The ligament length distribution was non-uniform with a standard deviation between 20% and 60%. Apart from the integral ligament properties discussed so far, we have also investigated the kinematics and dynamics of the ligament growth. Ligament head trajectory was found to closely resemble an involute. However, as pointed out by several authors, the ligament strain rate γ* (equation 4) is also important as it affects the stability of ligament growth. Ligament surface is stabilized (less prone to capillary breakup) when γ* >> 0. (4) In equation (4), Lr is the instantaneous ligament length at time t and dHD is the head droplet diameter.

Figure 9: Ligament strain rate curves for different wheel rotational speeds (Oh = 0.0183, Q = 1.63mL/s)

In all operating points, a similar development of the ligament strain rate was observed. After the onset of ligament growth, γ* increases steadily until the head droplet pinchoff (red dots in Fig. 9). At that moment, γ* is rapidly reduced and drops to a nearly zero value in case of a slow wheel rotation. After a certain time, ligament growth is resumed and γ* once again increases in time. The value of γ* rises until the moment when the ligament base is detached from the film (green triangles in Fig. 9). Ligament detachment is followed by a rapid decrease of γ* until the moment when the capillary breakup to droplets occurs (i.e. γ* curve termination in Fig. 9). It can be noted that in most cases the ligament breakup occurs when γ* is still positive and close to the limit for supercritical capillary wave damping (γ* ≈ 0.58). This indicates that the ligament elongation stability theory developed by other authors [6] is inadequate in our case. It is likely due to the fact that it only accounts for an initial disturbance at t = 0, but not for later disturbances induced by the head droplet and ligament pinch-off. Conclusions Mineral wool fiberization mechanism was studied on a model spinning wheel to overcome the limitations of measurements in an industrial environment. Analysis of process images showed several notable differences in operating mechanism compared with spinning discs and

cups, therefore justifying our experimental modeling. Ligament formation process was characterized by regression models for several integral parameters which can be applied to a wide range of operating conditions due to their dimensionless formulation. Especially important are the measurements of ligament length as this property has, until now, not been measured directly. The same is true for ligament strain rate measurements which have so far only been performed on fundamental 1D cases but not in practical applications (e.g. atomizers). However, our results have also exposed the limitations of existing theoretical models as well as of our experimental modeling approach. Further research should include experiments with a solidifying liquid (e.g. molten sugar) and the blow-in air flow. Moreover, numerical modeling with computational fluid dynamics programs should be employed to more accurately investigate the phenomena difficult or impossible to study experimentally. References

[1] ŠIROK, B., BLAGOJEVIĆ, B., BULLEN, P.R. Mineral wool: production and properties. Cambridge: Woodhead Publishing Limited, 2008. [2] WESTERLUND, T., HOIKKA, T. On the modeling of mineral fiber formation. Computers & Chemical Engineering, 1989, 13(10), 1153-1163. [3] KAMIYA, T. Analysis of the ligament-type disintegration of thin liquid film at the edge of rotating disk. Journal of Chemical Engineering of Japan, 1972, 5(4), 391–396. [4] LIU, J., YU, Q., GUO, Q. Experimental investigation of liquid disintegration by rotary cups. Chemical Engineering Science, 2012, 73, 44–50. [5] LIU, J., YU, Q., LI, P., DU, W. Cold experiments on ligament formation for blast furnace slag granulation. Applied Thermal Engineering, 2012, 40(1), 351–357. [6] VILLERMAUX, E. The formation of filamentary structures from molten silicates: Pele’s hair, angel hair, and blown clinker. Comptes Rendus Mecanique, 2012, 340(8), 555-564.

List of publications

[1] BIZJAN, B., ŠIROK, B., HOČEVAR, M., ORBANIĆ, A. Ligament-type liquid disintegration by a spinning wheel. Chemical Engineering Science, 2014, 116, 172-182. [2] BIZJAN, B., ŠIROK, B., HOČEVAR, M., ORBANIĆ, A. Liquid ligament formation dynamics on a spinning wheel. Chemical Engineering Science, 2014, 119, 187-198. [3] BAJCAR, T., BIZJAN, B., ŠIROK, B., ORBANIĆ, A. Mineral wool melt fiberization on a spinner wheel. Chemical engineering research and design, 2014, 92(1), 80-90. [4] BIZJAN, B., ŠIROK, B., GOVEKAR, E. Nonlinear analysis of mineral wool fiberization process. Journal of Computational and Nonlinear Dynamics, ISSN 1555-1415, 2014. [5] BIZJAN, B., ŠIROK, B. Experimental investigation and modeling of mineral wool melt adhesion on a spinner wheel. B. ŠARLER (ed.), N. MASSAROTTI (ed.), P. NITHIARASU (ed.). Proceedings of the 3. International Conference on Computational Methods for Thermal Problems, Bled, June 2-4, 2014, 321-324. [6] BIZJAN, B., ORBANIĆ, A., ŠIROK, B., KOVAČ, B., BAJCAR, T., KAVKLER, I. A computer-aided visualization method for flow analysis. Flow measurement and instrumentation, 2014, 38, 1-8. [7] BIZJAN, B., ORBANIĆ, A, ŠIROK, B., BAJCAR, T., NOVAK, L., KOVAČ, B. Flow Image Velocimetry Method Based on Advection-Diffusion Equation. Strojniški vestnik – Journal of Mechanical Engineering, 2014, 60(7-8), 483-494. [8] BIZJAN, B., ŠIROK, B. ORBANIĆ, A., BAJCAR, T. Flow Image Velocimetry Method Based on Advection-Diffusion Equation. 18th ERCOFTAC ADA PC Meeting, Budapest, November 15, 2013.

Yearbook 2014 Laboratory for Hydraulic Turbomachinery

Study of erosive cavitation detection in pump mode of pump- storage hydropower plant prototype Abstract → An experimental investigation has been performed to detect cavitation in reversibile pump-turbine hydro power plant. The prototype runner suffers moderate cavitation erosion on the suction side of the runner in the pump mode operation. Measurements of structural vibrations on the housing of turbine bearings, on the top of guide vane, and a measurement of high- frequency hydrodinamic

pressure in the draft tube have been performed. High frequency spectral content of the signal was analysed. Results are shown for different operational points of a prototype turbine. The results of a high- frequency pressure measurements are also compared to results of measurements with similar installation in a cavitation tunnel.

Introduction Tine Cencič Laboratory Laboratory for Hydraulic Turbomachinery E-mail tine.cencic@seng.si Phone +386-1-4771-723 Status PhD student (started: September 2011, to be completed: February 2015) Research area Water turbines Mentor Assoc.Prof.Dr. Marko Hočevar, Co-Mentor: Prof.Dr. Branko Širok

In the recent years energy production from renewable sources has increased. The increased dependence of weather conditions on renewable energy production led to design of units with high power and operation away from best efficiency point. In water turbines unwanted cavitation phenomena may occur at off-design operation and lead to degradation of efficiency, excessive vibrations and cavitation erosion. The traditional approach for cavitation detection is based on measurement and analysis of vibrations or acoustic emission on stationary parts of the machine. Escaler et al. [1] performed a test in a prototype Francis turbine using a rotation sensor sensor fixed on the rotation shaft. Avellan et al. [2] compared model test and prototype, emphasizing the importance of model testing. Escaler et al. [3] carried out an experimental investigation to evaluate the detection of cavitation in hydraulic turbines both for models and prototypes. The methodoloy was based on the analysis of structural vibrations, acoustic emission and hydrodynamic pressure measurement in the machine. Bajić

Laboratory for Hydraulic Turbomachinery Yearbook 2014

et al. [4] performed vibroacoustic measurements of turbine cavitation and multidimensional analysis. The aim of this paper is to find the estimator able to quantify cavitation intensity in the prototype turbine from the vibration measurements on the bearing housing and pressure measurements in the draft tube. Section Pump- storage hydro power plant The power house of pump turbine Avče (Fig. 1) is located on the terrace on the left bank of Soča river, Slovenia. The pump turbine Avče started operation in 2009. Instaled power in turbine mode is 185 MW, while in pump mode it is 180 MW. Effective head in pump mode is Hep 521 m, while rated discharge is Q = 34 m³/s. The available volume of water storage is Vk = 2,170,000 m³. The upper basin accumulation is located at 597 m (lower level) to 625 m (upper level) above the sea level. The pipeline and the high-pressure tunnel connect the upper basin accumulation with the power house of the power plant. The total length of

the pipeline and the high-pressure tunnel is 2216 m. The pipeline concreted in the vertical shaft and in part provided as the inclined tunnel. The high-pressure pipeline is at least 2.6 m and maximum 3.3 m in diameter. For assesment of cavitational characteristics of pump turbine Avče, lower basin accumulation is of high importance. Lower accumulation basin of pump turbine Avče is the lower basin of the Ajba accumulation HPP Ajba. The pump turbine Avče lower accumulation basin is at 104.00 m (lower level) and 106.00 m (upper level) above sea level, which is dependent on the operation of the chain of power plants on the Soča River, the HPP Doblar and the HPP Plave. The available volume amounts to 416,000 m3, if necessary it is possible to add water for the pump turbine Avče from the accumulation of HPP Doblar. The turbine is submerged below lower accumulation basin while located at 49 m above the sea level. In pump operation mode operation is possible with a variable rotational speed with the nominal rotational speed of 600 rev/minute and the possibility of the speed variation from –2 to +4 %. The variable speed enables an increased adaptability to the requirements of the electric power grid system. The pump turbine Avče is connected to the existing 110 kV network of the northern Littoral mesh.

sensor is < 1 ms and pressure range is 0-100 bar. The sensor was connected to the pump turbine flow tract through a stainless steel pipe of length 20 cm and diameter 12 mm. A manual valve was fitted to the steel pipe for instalation or deinstalation. Two accelerometers were installed on turbine cover near the turbine bearing (AC13 and AC14). They were positioned at an angle 90° relative to each other. They were both oriented in the radial direction perpendicular to the pump turbine shaft. The third accelerometer was mounted on the pump turbine governor guide vane (ACV13). It was mounted on the top of the guide vane, oriented axially with the pump turbine axis. All three accelerometers were stud mounted. The accelerometers used were CTC type AC240-2D. The frequency interval of three sensors was 0.6 Hz to 25 kHz, dynamic range was ± 50g and sensitivity was 100 mV/g. Sensors were connected to CTC type PS03 power supply. For data acquisition, a National Instruments NI 6351 data acquisition board was used with 16 bit resoulution and sampling rate up to 1.25 MHz. Low frequency sensors were sampled simultaneously at a frequency of acquisition 200 kHz per channel (low frequency pressure sensor, all three accelerometers). High frequency sensor was sampled with frequency 1 MHz non simultaneously. Samplig interval was for all sensors 7 s. Measurement data was stored to disk for later analysis. Analysis of measurement data was performed using National Instruments Labview software. During operation, all operational data from pump turbine was recorded.

Figure 1: The power house of pump- storage hydro power plant Avče

Measurement equipment Two pressure sensors were installed, one for high frequency (HPS1) and one for low frequency (PS1). Both sensors were installed in a metal service door opening, approximately 1.5 m below the runner. Mounting of both pressure sensors from closer to the runner was not possible, because above the door the pump turbine flow tract is sealed with concrete (see Figure 2). Above the runner, a thick turbine cover also prevents installation. Drilling of bores, large enough for sensor installation, through the turbine cover was not possible due to strength issues. The high frequency sensor used was an ICP general purpose quartz PCB Piezotronics type 111A26 sensor. The characteristics of the sensor are pressure range 0-35 bar, 145 mV/bar sensitivity and 0.000689 bar resolution. A signal conditioner for ICP sensors PCB Piezotronics type 480C02 was used. The high frequency pressure sensor was connected to the pump turbine flow tract through a volute of diameter 6 mm and length 2 mm, while the bore from the volute to the flow tract was of diameter 2 mm and length 3 mm. The low frequency absolute pressure sensor used was an ADZ Nagano type SML-10. The response time of the

Figure 2: Locations of sensors installation

Operational points Model testing in laboratory has identified the area of cavitation bubbles presence in both turbine and pump mode of operation. During operation in the first year, moderate cavitation damage was observed in locations, which corresponds to cavitation during pump mode laboratory testing. In locations of cavitation laboratory testing in the turbine mode of operation, no cavitation damage was observed. Based on this, we will in the following focus measurements and analysis only on the pump mode operation. The flowrate has been evaluated as a dimensionless number. According to standard IEC60193 we used a discharge coefficient Qnd. Discharge coefficient was calculated as (reference IEC60193) Yearbook 2014 Laboratory for Hydraulic Turbomachinery

(1)

discharge coefficient but a higher cavitation intensity depending on the lower discharge coefficient.

Here N is rotational speed (rpm), Q is discharge (m3/s) and D is discharge diameter of runner (m). There are different ways to calculate the Thoma number. In this paper we use: (2)

where He is an effective head and NPSH is the net posotove suction head. rESULtS AnD DISCUSSIon As the best estimator was used standard deviation (σd). Standard deviation was compared with a visualization in a cavitation tunnel and it was noticed as one of the best estimators for the intensity of cavitation [6].

(3) N …. Number of all measurements X …. Measured value -n X …. Average of all measurements Results in pump mode: High frequency pressure sensor is showing a big influence of standard pressure deviation on the discharge coefficient and Thoma number (Figure 3).

Figure 4: Estimator standard pressure deviation in acceleration sensors

Figure 3: Influence of the discharge coefficient to the cavitation intensity

Two accelerometers installed on turbine cover near the turbine bearing (AC13 and AC14) positioned at an angle 90° relative to each other are measuring almost the same measurements (Figure 4). The third accelerometer on the top of guide vane (ACV13) is not showing the same results as the sensor in radial directions. The sensor is not showing a big influence on the thoma number in lower 54

LAborAtorY For HYDrAULIC tUrboMACHInErY YEARBOOK 2014

The type of cavity is correspond to a low incidence angle of the flow and depending on the design of the impeller. For a lower discharge value, the flow incidence is increased and then a leading edge cavity appears. In centrifugal pump, leading edge cavity is the main type of cavitation development that is producing cavitation erosion. The results in pump mode are showing that the cavitation is occured when the pump is operating in lower discharge coefficient and it is increasing when the thoma number is decreasing. From the results we can notice that the cavitation is exponentally increasing in part when the discharge coefficient is getting lower. A high corelation can be seen from all the vibration sensors and a high frequency pressure sensor.

Conclusions A study of vibration, dynamic pressure levels in the high frequency range is already a well- known technique to detect cavitation activity. The use of high frequency pressure sensor serves to extend the analyse tu upper frequencies that the accelometers cannnot reach. In this examination it was shown a way with standard deviation estimator to detect the cavitation intensity. The results are showing that on the cavitation in pump mode has a big influence the discharge coefficient, even more than the Thoma number. Compare to the theory and the model test the come results are showing the real situation of the cavitation intensity. References [1] X., Egusquiza, E., Farhat, M., Avellan, F., Cussirat, M. Escaler, “Detection of cavitation in hydraulic turbines,” Mechanical Systems and Signal Processing, vol. 20, pp. 983-1007, 2006. [2] Francois Avellan, “Introduction to cavitation in hydraulic machinery,” 2004. [3] Mohamed Farhat, Eduaer Egusquiza, Francois Avellan Xavier Escaler, “Vibration cavitation detection using onboard measurements,” in fifth international Symposium on Cavitation , Osaka, Japan, November 1-4, 2003, 2003. [4] B., Keller, A. Bajić, “Spectrum normalization method in vibro-acoustical diagnostic measurement of hydroturbine cavitation,” Journa of Fluids Engineering, vol. 118, pp. 756-761, 1996. [5] International Electretrotechnical Comission, “IEC 60193:1999,” Geneva, Hydraulic turbines, storage pumps and pumps- turbines- Model acceptance tests 1999. [6] Zhaohui Li, Xuezheng Chu, Qingfu Sun Huixuan Shi, “Experimental Investigation on Cavitation in Large Kaplan Turbines,” in Third International Conference on Measuring Technology and Mechatronics Automation, 2011, pp. 120-123. List of publications [1] T., Hočevar, M., Širok, B. Cencič, „Study of erosive cavitation detection in pump mode of pump-storage hydropower plant prototype,“ Journal of Fluids Engineering, vol. Vol. 136, May 2014

Yearbook 2014 Laboratory for Hydraulic Turbomachinery

Direct measurements of thermal delay in cavitating flow for the European Space Agency (ESA) Abstract → Optimal operation of turbopumps is crucial for all liquid fuel rocket engines. To reduce weight, these pumps often operate at critical conditions, where dynamic instability and cavitation are unavoidable. In cryogenic engines, the fuel and oxidizer used are liquid usually hydrogen and liquid oxygen at very low temperatures (about 14 and 90 K, respectively). Usually we treat cavitation as an isothermal phenomenon, but this assumption is not valid for such propellants: flows are characterized by a substantial cooling during the vaporization process due to cavitation. This phenomenon delays the further development of cavitation, so it plays a moderation role in the cavitation increase. The numerical prediction of the thermal effect is therefore a major industrial issue.

carry out direct measurements of thermodynamic effects in the developed cavitating flow. Previous studies (mostly from 1960’s and 70’s) have addressed this issue only indirectly by comparing results obtained in water and in thermosensible media. The equipment and experience we have enables to measure the temperature effects directly and thus to construct a better basis for evaluation of the results of numerical predictions. Direct users of the results are researchers and industry that are dealing with cavitation erosion and specific topics of cavitating flow. The paper presents the first ever time resolved measurements of thermodynamic effects in cavitating flow in a Venturi nozzle.

The presented study is supported by the European Space Agency (ESA). In the scope of the project we

Introduction Matevž Dular Laboratory Laboratory for Hydraulic Turbomachinery E-mail matevz.dular@fs.uni-lj.si Room 314 Phone +386-1-4771-314 Status Associate professor Research area Cavitation

Cavitation is characterized by vapor generation and condensation due to pressure changes at approximately constant temperature of the fluid. Is it justified to use the isothermal approach when we are dealing with liquids such as cold water, where the influence of the temperature variations on the integral liquid properties is negligible [1]. A detailed look in the formation of a cavitation bubble shows that it is formed by the local pressure drop, which causes the gas inside the cavitation nucleus to expand what consequently triggers evaporation. The latent heat is then supplied from the surrounding liquid, creating a thermal boundary layer around the bubble. The heat transfer causes a local decrease of the liquid temperature, which results in a slight drop of the vapour pressure [2]. This phenomenon delays the further development of the bubble, because now a greater pressure drop is needed to maintain

Laboratory for Hydraulic Turbomachinery Yearbook 2014

the process. This phenomenon is known as thermal delay [3]. When the local surrounding pressure rises, the bubble starts to collapse. During the collapse the condensation occurs and in the final stages the gases also violently compress, which leads to considerable rise of the temperature inside the bubble [4]. Due to complexity of the experimental investigation of the local temperature variations, the past studies mostly concentrated on the consequences of the thermodynamic effects, rather than on the investigation of the mechanism itself. Only few simple measuremtns existed [5] before Dular & Coutier-Delgosha [6] introduced the high-speed infrared (IR) camera technique to measure the temperature on a single cavitation bubble Until now the thermodynamic effects were usually estimated by rudimentary models, which most of them were proposed between 1960 and 1990. The most commonly used

nondimensional parameters are the B-factor [7], Σ parameter [8], α [9] and Σ* parameter [10]. The present study shows innovative, direct measurement of thermodynamic effects in cavitating flow by a non-invasive method. A high speed IR camera was used to measure the temperature field in the cavitating flow; simultaneously visualization by a conventional high-speed camera was made.

Considering the combination of inaccuracies of pressure and temperature measurements, the cavitation number could be determined within ±0.02 of global uncertainty. Figure 2 shows about one period of cavitation cloud shedding process at cavitation number s = 1.3.

Measurements Experimental set-up The experimental set-up is shown in Fig. 1.

Figure 2: Cavitation as seen at cavitation number s = 1.3 (the flow direction is from the left to the right)

Figure 1: Experimental set-up

The cavitation tunnel consists of two 2 L reservoirs (1 and 2), a convergent-divergent Venturi nozzle (3) and a ball valve (4). The first reservoir (1) is filled with the working fluid (hot water) and pressurized to the desired level (through pressure pipe connection (5)). Similarly, in the second, empty, reservoir (2), the pressure level can also be adjusted by the second pressure pipe (6). At a rapid opening of the ball valve (4), the working fluid is pushed from the first reservoir through the Venturi nozzle (3), where the cavitation occurs, to the second reservoir (2). During the 3 to 5 second long experiment the pressures in both reservoirs were recorded by Hygrosens DRTR-AL-10VR16B pressure transmitters at a rate of 1000 Hz. The geometry of the Venturi test section is a constriction with a converging angle of 18° and diverging angle of 8° was used. The cross-section of the test section channel is reduced from 6×5mm2 at the inlet, to the 1×5mm2 at the throat. Downstream of the throat of the Venturi nozzle an observation window, made out of sapphire glass, was installed – sapphire glass had to be used since it is transparent in both visible and infrared light spectrum. The flow velocity changes during the experiment as a result of changing pressure difference between the two reservoirs. Nevertheless, during a short period of time (about 0.3s) the pressure difference remained constant at a desired level so that measurement point could be easily determined. Developed cavitating flow was observed at several pressure differences which give values of cavitation number (s=1.3, 1.8 and 2.3).

One can see that cavitation is first attached (at t = 0.0 ms). It then slowly grows while at its closure a reentrant jet forms – it flows upstream and consequently causes the cavitation cloud to separate (at this instant at t = 4.2 ms the maximum size of the attached part of the cavity is reached). The cloud separates from the attached cavity at t = 4.9 ms. Interestingly, it seems that at t = 2.8 ms the reentrant jet splits into two branches – a more persistent branch travels further upstream and causes the large cloud separation, wile the smaller one dies out at t = 4.9 ms. Cavitation visualization and temperature measurements Fastec Imaging HiSpec4 2G mono high speed camera was used to capture images of cavitating flow from the side view. The camera enables capturing images at 523 fps (frames per second) at 3Mpixel resolution. For the present experiments, it was synchronized with the IR camera and recorded at a higher frame rate. A high-speed IR camera CMT384SM – Thermosensorik was used to record the dynamics of the temperature fields. The camera is sensitive in wavelength range between 3 and 5 mm. To analyze the dynamics of the temperature field and its relationship to the cavitation structure dynamics a higher frequency (3550 fps) of the acquisition was chosen. The uncertainty of IR temperature measuring was inspected by comparison with measurements by an A-class Pt100 sensor and a discrepancy of ±0.2 K was found. However, our main goal was to quantify relative differences in non-uniform (and possibly time-dependant) temperature field. For a single element on the temperature sensor of the IR camera, the noise equivalent temperature difference (NETD) was less than 20 mK. Experimental results As already mentioned two sets of measurements were conducted. We first investigated the mean features of the temperature fields at a low acquisition frequency, later we increased the frame rate of the IR camera to capture the temperature dynamics of cavitating flow. Yearbook 2014 Laboratory for Hydraulic Turbomachinery

Mean temperature fields During this set of experiments conventional and IR high speed cameras operated at 10000 and 790 fps, respectively. Figure 3 shows a representative instantaneous image of cavitation (a), a mean of a series of cavitation images (b) and the mean temperature field in the layer near the sapphire window (c) for cavitation numbers σ = 1.3 (top), 1.8 (middle) and 2.3 (bottom). The flow is from the left to the right; x = 0 mm corresponds to the position of the throat of the Venturi channel. The reference temperature of approximately 95°C (slight variations (±1 K), which do not influence the magnitude of the thermodynamic effects, appeared between the tests) was measured 20 mm upstream of the throat of the Venturi by a type J thermocouple.

ture depression is associated to the cavitation number (the cavitation extent) – the larger the cavitation number the smaller the depression. At the present time we feel that more measurements are needed for a clear conformation. Also the temperature fields (images c) closely resemble the mean of a series of cavitation images (images b) – when cavitation extent is greater a larger volume of the fluid is at an elevated temperature and the fall to the freestream temperature occurs further downstream. It is common for all three cases that cavitation structures cannot be seen just downstream of the throat. The reason for this could be that a large bubble (a few mm in length) first forms in this region, from which smaller bubbles shed – the light scattering on smaller bubbles is more intense, hence they appear whiter in the image. Also one can observe the sheet cavitation in this region must be very thin (less than 1 mm) what implies that it was poorly illuminated during the experiment. All three cases also share the fact that the bulk fluid is hotter than the one near the bottom and top channel walls. This could be related to the heat flow due to the different temperatures of the fluid and the channel walls. Prior to the experiment the system (test section) was stabilized at a temperature close to the temperature of the water (95°C) by running non cavitating flow through it for a couple of minutes. Since the flow is highly turbulent also the thermal boundary layer should be very small but it seems that we were unable to entirely prevent the heat transfer between the walls and the fluid. Temperature field dynamics In the second set of measurements we investigated the dynamics of the temperature field. For this we needed to increase the acquisition frequency of the IR camera to about 3550 Hz. Consequently we had to decrease size of the observation window, which now extended only 15 mm downstream of the throat. Figure 4 shows the instantaneous images captured with the high-speed (left) and IR (right) cameras.

Figure 3: Representative instantaneous image of cavitation (a), mean of a series of cavitation images (b) and the mean temperature field (c) for σ = 1.3 (top), σ = 1.8 (middle) and σ = 2.3 (bottom)

As a consequence of cavitation bubble growth by evaporation and gas expansion in the vicinity of the throat a clear temperature depression can be observed. The bubbles then start to collapse what causes rapid recuperation of the temperature which finally exceeds the initial (freestream) temperature. After the peak temperature is achieved, it slowly falls and limits to the freestream temperature at the downstream end of the observation window. Similarly to the findings of Franc et al. [6] the results imply, to some extent, that the magnitude of the tempera-

Laboratory for Hydraulic Turbomachinery Yearbook 2014

Figure 4: Instantaneous images of cavitation and the corresponding temperature field at σ = 1.8.

As already mentioned cavitation near the throat of the Venturi cannot be clearly seen. One can see that cavitation clouds, which consist of a large number of tiny bubbles, posses a higher temperature than the rest of the flow. This again confirms the hypothesis the bubbles inside the cloud undergo several rebounds in their lifetime. If a bubble would undergo only one growth and collapse, the cloud should either be cold (if the bubble would still be growing) or of the same temperature as the rest of the fluid (if the bubble would be at a quasi stable state before its collapse), the “hot” region would then appear only for a short period of time, just after the cloud collapse.

Conclusions

Comparisons to the theory of thermal delay The Rayleigh-Plesset equation with consideration of the thermal delay theory was developed only for a case of a single cavitation bubble and more complex approaches are scarce – mainly due to the lack of experimental data.

(1) In the present work we limited our investigation to the question if experimental results roughly comply with the theory. As a test condition we took the case at σ = 1.3. First we calculated the pressure evolution along the mid height (see Fig.4) of the Venturi channel. To limit the number of adjustable variables and consequently to leave the investigation as clear and as simple as possible this was done by a potential flow theory approach. Here one calculates that as the flow approaches the throat of the Venturi the pressure rapidly drops from p1 to vapour pressure, which persists about 6 mm downstream of the throat. During the next 12 mm the pressure then recovers to approximately pressure p2. The pressure losses were also considered. Since we are dealing with cavitating flow, one should consider the possibility of interaction between the bubbles. This was intentionally omitted, again for the sake of clarity of results. We do not know the size of the cavitation nuclei as they enter the Venturi channel. Hence an ensemble of nuclei sizes was chosen (R0 = 1, 1.5, 2, 2.5 and 3 mm). Figure 5 shows the calculated temperature evolutions (according to Eqn. 3) for 5 cavitation nuclei sizes together with experimental results for the case of σ = 1.3.

Figure 5: Calculated and measured temperature evolutions in the channel for the case of σ = 13.

One can see that the bubbles undergo one or more rebounds as they flow through the low pressure zone. As already mentioned a single growth and collapse would result in a rapid cooling and heating of the fluid. On the other hand a number rebounds reveals a temperature profile which first drops below the freestream temperature and then gradually rises to a higher temperature – a trend which was observed during experiments.

In the present work we show high speed IR camera measurements of the temperature fields in cavitating flow of hot water, which already exhibits measurable (if not significant) thermal delay. For the first time the temperature fields across the Venturi section were obtained and the temperature dynamics was observed. Results show that the idea of the thermal delay can be applied to the present problem, but one needs to consider the possibility (probability) that a bubble rebound occurs. References [1] Hord, J., Anderson L. M., Hall W. J. 1972 Cavitation in Liquid Cryogens I – Venturi. NASA CR-2054. [2] Franc, J.-P. & Michel, J.-M 2004 Fundamentals of Cavitation. Fluid Mechanics and Its Applications, 76, Springer. [3] Brennen, C.E 1995 Cavitation and Bubble Dynamics, Oxford University Press. [4] Hauke, G. , Fuster, D. & Dopazo, C. 2007 Dynamics of a single cavitating and reacting bubble. Physical Review E, 75, 066310. [5] Fruman, D.H., Reboud, J.L. & Stutz, B. 1999 Estimation of thermal effects in cavitation of thermosensible liquids. Int. Journal of Heat and Mass Transfer, 42, 3195-3204 [6] Franc, J.-P., Boitel, G., Riondet, M., Janson, E., Ramina, P. & Rebattet, C. 2010 Thermodynamic Effect on a Cavitating Inducer-Part II: On-Board Measurements of Temperature Depression Within Leading Edge Cavities. ASME J. Fluids Eng., 132 (2), 021304. [7] Stepanoff A. J. 1961 Cavitation in centrifugal pumps with liquids other than water. J. Eng. Power, 83, 79-90. [8] Brennen, C. E. 1973 The Dynamic Behavior and Compliance of a Stream of Cavitating bubbles. ASME J. Fluids Eng., 95 (4), 533-541. [9] Kato, H. 1984, Thermodynamic Effect on Incipient and Development of Sheet Cavitation. Proceedings of International Symposium on Cavitation Inception, New Orleans, LA, 127-136. [10] Watanabe, S., Hidaka, T., Horiguchi, H., Furukawa, A. & Tsujimoto, Y. 2007 Steady Analysis of the Thermodynamic Effect of Partial Cavitation Using the Singularity Method. ASME J. Fluids Eng., 129 (2), 121-127. List of publications [1] Dular, M., Coutier-Delgosha, O. Thermodynamic effects during growth and collapse of a single cavitation bubble. Journal of Fluid Mechanics, 2013, vol. 736, pp. 44–66. [2] Petkovšek, M., Dular, M. IR Measurements of the Thermodynamic Effects in Cavitating flow. International Journal of Heat and Fluid Flow, 2013, vol. 44, pp. 756-763.

Yearbook 2014 Laboratory for Hydraulic Turbomachinery

Precision spraying in orchards and vineyards Abstract → In year 2014 we finished an important research project, in September, CROPS EU 7th FP project has ended. In workpackage 7 of this ICT Robotic project, the following objectives were successfully met. First objective was to achieve canopy optimised spraying where the spraying operating parameters are continuously adapted to the canopy characteristics, such as volume, shape, density, etc., varying in space and time with the aim to reduce pesticide application by 30%. The second objective was to achieve high-

precision close range precision spraying to selectively and precisely apply chemicals solely on targets susceptible to specific diseases/pests with a 90% success rate and a 90% reduction in pesticide use for selected varieties.The first objective was performed in Ljubljana, the second objective was performed in Milano, while Marko Hočevar was the workpackage leader.

Introduction Marko Hočevar Laboratory Laboratory for Hydraulic Turbomachinery E-mail marko.hocevar@fs.uni-lj.si Room 314 Phone +386-1-4771-314 Status Associate Professor Research area Water Turbines, Agricultural Engineering, Measurements

Application of agrochemicals is at present, the primary method used to protect plants from pests (diseases, fungi, insects, weeds). To do this, active ingredients formulations are diluted in water and distributed to vegetation in form of spray droplets. To protect plants from diseases and pests, agrochemicals are sprayed uniformly to ensure coverage of susceptible targets at the appropriate time in the season. In orchard crops and grapevine susceptible targets (fruits, bunches, new sprouts, younger leaves etc.) can be located anywhere in the vegetation, consequently current spraying techniques aims to cover all parts of plants, front and behind, top and bottom, as well as within the canopy. Consequently, a high volume air-flow assists the transport and the depositing of pesticides droplets to internal and hidden parts of plants inside canopy of permanent crops. Large droplets fall down before reach the target or slip down from leaves because of great weight. Small droplets also fly away from canopy. In locations of plants with very thin canopies, droplets travel through the canopy. The significant amount of pesticide can be lost at such type of application. This has a negative effect on quality of execution of agrotechnical operations, production economics and the environment. Precise application supported by up-to-date technologies of target application is one of the most promising

Laboratory for Hydraulic Turbomachinery Yearbook 2014

options of pesticide quantity reduction. The concept of precise application of pesticides involves the possibility of momentary modification and adjustment of chemical application to the needs of the target (plant) on which the application is being performed. Hence, there is a need to develop and introduce techniques and systems for disease detection and pesticide application able to optimize the distribution of pesticides according to the specific characteristics of the target, such as disease’s susceptibility, or the presence of infection symptoms. Crops EU project The main objective of CROPS EU project was to develop a highly configurable, modular and clever carrier platform comprising a carrier plus modular parallel manipulators and “intelligent tools” (sensors, algorithms, sprayers, grippers) that can easily be installed onto the carrier and that are capable of adapting to new tasks and conditions. Both the scientific know-how and a number of technological demonstrators was developed for the agro management of high value crops like greenhouse vegetables, orchard fruits, and grapes for premium wines. The CROPS robotic platform is capable of site-specific spraying (discussed here) and selective harvesting of fruit (i.e., detect the fruit, determine its ripeness, move towards the fruit and grasp it and softly detach it).

Canopy optimised spraying In the following, we will review the design of canopy optimised sprayer. It is composed of several parts, prime mover, sensing equipment, sensing information processing manipulator and end effectors. Canopy optimised sprayer in 2014 configuration is shown in Fig.1. We will also provide information on activities in the current year 2014. Prime mover As prime mover we name the platform that will drive the canopy optimized sprayer. The canopy optimised sprayer is a trailed design with one axle so one part of the weight is carried by prime mover. The prime mover provides also hydraulic power for robotic arm movements, electric power for electric and electronic components and mechanic power for moving and spray pump drive. The specifications for min. power, max. wheelbase, max. width, weight, drive, working driving speed range, PTO speed, hydraulic fluid pressure, hydraulic fluid flow and electric power source meet specifications of a normal agriculture orchard tractor.

Figure 1: Canopy optimised sprayer

The sprayer is connected to the prime mover, while prime mover provides hydraulic connections, electrical power supply and PTO shaft for spraying fan operation. The sprayer will be trailed design with spraying tank capacity of at least 200 l. Design of the sprayer and the platform is open, it is not covered because of sensing requirements. This was selected because hail nets prevent operation with any type of coverage. Sensing devices The task of sensing devices for the canopy optimized sprayer is to detect presence, distance to and density of the canopy. Sensing system must measure, analyze canopies and provide the following information to the spraying arm: 3D surface of the canopy, density of canopy in dependence of the height, distance from the sprayer in dependence of the height. All distances are measured from the sensing devices, this will allow for calculation of distances from the spraying arm. All sensing devices must be able to operate with 12-14 V DC supply voltage, as available on the orchard tractor. With density of the canopy we name amount of leaves in the canopy, the green mass. For measurement of canopy density we see the following possibilities. With laser scanner, dense canopy does not allow the laser beam to pen-

etrate deep in the canopy, so output profile will follow the contour without much scatter. Canopy with small density (less leaves per canopy volume), will allow penetration of laser beam deeper in the canopy and the output profile will follow canopy with a lot of scatter.The following sensors were selected for sensing tasks for canopy optimized spraying: laser scanner, RGB cameras, and ultrasonic sensors. Sensing information processing To achieve design objectives of canopy optimised sprayer, one must allow for accuracy of canopy optimised spraying of regions of around 30x30 cm. This specification we set based on distance between plants in a row, height of plants and properties of average plant. The selection of spraying region 30x30 cm sets specifications for accuracy of sensing and speed of processing of information. However, it is not possible to translate sensing accuracy to the same actuating accuracy. This has several sources: airflow deviation due to ambient wind velocity and direction, with trailed designs, there is an increase in uncertainty due to uneven terrain and poor tracking of the mean path between left and right row, difference in measured distance between sensors and tree row and actual distance between actuator and tree row as result of non straight driving or turning, finite response time of spraying arm, etc. We therefore assume that three times increase in uncertainty appears when going from sensing to actuation uncertainty. Uncertainty of canopy detection was thus 10 cm in both x and y direction. Uncertainty here is given as absolute measurement, not as perimeter or thickness. Canopies usually have irregular shape to such extent, that perimeter simply can not be established. Uncertainty relates to locations individual canopies, branches, edges etc. Processing speed of sensing information must allow derivation of 3D surface of the canopy, density of canopy in dependence of the height, and distance from the sprayer in dependence of the height in 50 ms (path/velocity) for max spraying velocity that was set to 2 m/s. The frequency of measurement data analysis was therefore better than 20 Hz in all cases, the worst case of driving velocity 2 m/s and 10 Hz in the case of driving with velocity 1 m/s. Actuation Actuation is required to bring the aerodynamic and pesticide nozzles close in appropriate position relative to the plant and spray the plant according to the canopy characteristics. Actuation is composed of the following two components: manipulator and end effector. Manipulation is movement of the spraying arm to the desired position relative to the plant and end effector composed of the aerodynamic and pesticide nozzles for spraying according to the canopy characteristics. To achieve appropriate pesticide coverage into tree canopy we used three spraying arms with one aerodynamic deflector (spout) each with two different pesticides nozzles on each. Number of spraying arms with deflectors (spouts) was be discussed with regard to the spray coverage uniformity. More spraying arms should most probably improve the uniformity of spray coverage, but may also lead to spray overlapping. Spray overlapping may lead to poor localization of the spray, something we did not want. Because of this we have limited the design to three spray-

Yearbook 2014 Laboratory for Hydraulic Turbomachinery

ing arms with one aerodynamic deflector each. However, since the principle of operation was proven, the design can be upgraded to more spraying arms. If deflectors are fixed on straight vertical holder, then are the distances between them and target area (tree canopy surface) different. Also air flow directions are not perpendicular to target area because of arc shape of tree canopy of normal slender spindle shape of the tree. This approach is preferred on few commercial designs currently available. Consequently, part of spray flow does not reach the canopy especially at the top and bottom of the tree and this part represent spray loss. We solved this problem with tilting spraying arms/actuators whose upper and lower lever was extended and positioned towards the tree. Manipulator The structure of the arm is strong enough that it can carry the weight of deflectors and air supply lines and forces resulting from acceleration due to the movement. The structure also has to be stable enough, that deformations are small, preferably less that 1 cm. For all movements hydraulic cylinders were be used because they are easy to control, can be stopped in each position and may persist in this position. We set specification to hydraulic cylinders over the electric operation of the spraying arm due to the favorable behavior in the case of blockages, leading to better reliability. Hydraulic cylinders are controlled through 4/3 solenoid valves. They are used in closed loop control so each movement (linear or rotational) with analog transducer. On the sprayer there is also system for driving speed measurement. The driving speed is required to calculate the time delay between sensing and actuation. The canopy is divided into three parts, low, middle and high. From the coordinates, provided by the measurement system in front of the spraying arm, measured points on each vertical part of canopy surface, the average (fitted) line distance, angle and center of gravity are calculated. Then the spraying arm moves every single deflector toward the tree such, that the deflector axis is at the appropriate distance and perpendicular to the fitted line. This is shown in Fig. 2. The distances between the end of deflectors and center of gravity is adjustable. To achieve this goal we will use a structure with two levers, which are swivel mounted on the middle holder and extendable.

Figure 2: Degrees of freedom of the canopy optimised spraying. To achieve the described functionality it is required to provide eight degrees of freedom.

Laboratory for Hydraulic Turbomachinery Yearbook 2014

At the end of each arm the deflector is swivel mounted, while the middle deflector is also swivel mounted on the middle holder so that just angle can be adjusted for this deflector. Middle holder is fixed on central extendable beam which allows movements of whole structure towards or away from trees. Development in current year 2014 In year 2014, based on results and analysis several small modifications of the software were performed, resulting in better performance of the canopy optimised sprayer (Fig 3). Most modifications were made on the algorithm and software level, while hardware was mostly retained in 2013 configuration.

Figure 3: Canopy optimised sprayer in operation in 2014.

Among the modifications performed are most important arms positioning algorithm modifications, power management, installation of fast valves for continuous flow control, and selection of algorithms for continuous flow control. Close range precision spraying With state of the art spraying methods in vineyards and orchards plants are evenly sprayed as the sprayer travels along the row of plants. Airflow support generates large movement of branches and leaves, in the large extent exposing both leaves sides to the spray. In addition, airflow generation at the large scale generates large coherent vortices, further contributing to the improved spray coverage on both sides of leaves. In comparison to the state of the art spraying methods, a new close range precision spraying method is evaluated here. Close range precision spraying of discrete foci of diseases at their early development stage is a promising method for reduction of pesticides use. Depending on disease development stage, close range precision spraying airflow targets are of size around 15 cm in diameter, preventing generation of very large turbulent coherent structures in the airflow. To enhance local turbulence and leaves fluctuations and to enable spraying of both sides of leaves, rotating airflow screen was used. With rotating airflow screen, airflow velocity power spectrum at location of the target shows discrete peaks of velocity fluctuations. Close range precision spraying experiments were performed on young vines. Fast camera, image analysis and water sensitive papers were used for analysis. Natural

frequencies of individual leaves showed sharp peaks of fluctuations at discrete frequencies. Leaf RMS velocity fluctuations corresponds well with airflow turbulence levels. Spraying is evaluated as spray coverage percentage and number of impacts. Spraying of adaxial leaves surfaces was good, while spraying of abaxial surfaces should be further improved. For actual spraying application, the rotating airflow screen should be replaced with more compact device with similar characteristics. In the following we will review our contribution to the project regarding spraying end effector, while we will skip the part provided by other project partners. Flow aerodynamics around plants and leaves Fluid flow carries pesticide droplets and exhibits pressure forces to surfaces of leaves and branches. Leaves and branches deform according to these pressure forces. Flow field of fluids around deformable bodies such as plants and their interaction with fluid flow are very complex and challenging problem in mechanics of fluids. Both experimental and CFD methods are used for evaluation. The main goal is to establish flow conditions in canopies, required for good pesticide application. Studies of aerodynamic properties inside canopies have been initiated by studies of drag properties of plants, competing for light which drives the evolution of taller and taller plants and their reconﬁguration in response to this drag. In last two decades, attention has been paid to plant canopy flows in attempts to understand, quantify and model chemical, momentum and energy exchanges between the air and the canopy elements. Canopy turbulence is dominated by intermittent coherent structures that are responsible for most of the momentum and scalar fluxes, and whose length scales are of the order of the canopy height. In forests or plantations, the size of these structures may be modulated by inter-tree spacing. In the last three decades, extensive measurements inside and above canopies have been performed using single point sensors, e.g., hotwire and hot-film anemometer, laser Doppler velocimetry and sonic anemometers in the field or in wind tunnels. Results were used to derive an inflection point at canopy height for a wide range of rough boundaries, a property typical to flows over orchards or vineyards. Spraying end effector The prototype of end-effector for spraying is shown in Fig 4. Spraying end effector consists of the following components: airflow generator (axial fan), airflow nozzle, pesticide nozzle with anti-dripping device, electrical connections (power supply and control signals), and chassis. Required for operation of the prototype of end-effector for spraying are also: pump for pesticide, pesticide connection pipe, pesticide connection bypass pipe, electric power supply (24 V DC and 48 V DC) and supply of control signals for fan rotational speed selection and operation of pesticide valve. These devices are located in the manipulator or optionally in a control panel (in the case of operation without manipulator). In the final version the fan installed is EBM Papst 4118 N/2H7P. The current version features PWM control for rotational speed, accessible with the manipulator or control panel for use without manipulator. Other internal parts

are Crydom D1D07 relay module, Jakša 3/2 valve with Schottky diode and pesticide nozzle Steinen (0.5 gal/h, 30°, full cone).

Figure 4: Spraying end effector.

Conclusions The CROPS project was sucessfully finished in September 2014. In workpackage 7, for which Marko Hočevar was workpackage leader, workpackage objectives were largely met. List of publications [1] HOČEVAR, Marko, ŠIROK, Brane, GODEŠA, Tone, STOPAR, Matej. Flowering estimation in apple orchards by image analysis. Precision agriculture, ISSN 1385-2256, Aug. 2014, vol. 15, iss. 4, str. 466-478, [2] STERMAN, Aljaž, MALNERŠIČ, Aleš, HOČEVAR, Marko, GODEŠA, Tone. Moving spraying arm with internal airflow for air-assisted orchard sprayer. V: International Conference of Agricultural Engineering, Zurich, 6-10 July 2014. [3] BERENSTEIN, Ron, BEN-SHAHAR, Ohad, EDAN, Yael, GODEŠA, Tone, HOČEVAR, Marko. Image registration for agricultural sensing tasks. V: International Conference of Agricultural Engineering, Zurich, 6-10 July 2014. AgEng 2014 Nullam euismod lacus facilisis odio viverra, congue pharetra ipsum vehicula [4] OBERTI, Roberto, MARCHI, Massimo, TIRELLI, Paolo, CALCANTE, Aldo, IRITI, Marcello, HOČEVAR, Marko, BAUR, Joerg, PFAFF, Julian, SCHÜTZ, Christoph, ULBRICH, Heinz. The Crops agricultural robot : application to selective spraying of grapevine’s diseases. V: International Conference of Agricultural Engineering, Zurich, 6-10 July 2014. AgEng 2014 [5] JEJČIČ, Viktor, HOČEVAR, Marko, GODEŠA, Tone, HERAKOVIČ, Niko. Sprayer for targeted application of phytopharmaceutical preparations in permanent crops = Zerstäuber zur gezielten Auftragung von phytophatmazeutischen Zubereitungen bei Dauerkulturen = Pulvérisateur pour une application ciblée de préparations phytopharmaceutiques dans des cultures permanentes : European patent specification : EP 2 277 376 B1, 2014-08-13. Paris: Europäisches Patentamt: = European Patent Office: = Office européen des brevets, 2013. 7 str.

Yearbook 2014 Laboratory for Hydraulic Turbomachinery

Investigation of vortex shedding from an airfoil by CFD simulation and computer-aided flow visualization Abstract â&#x2020;&#x2019; An experimental and numerical study of vortex generation and shedding from a NACA 4421 airfoil at low Reynolds number is presented. The experiment was conducted in a low-speed wind tunnel by flow visualization. A high-speed camera was used to record flow structures at the airfoil trailing edge. The recorded images were processed with an in-house developed software based on the advection-diffusion

equation to compute instantaneous 2D velocity fields. These results were compared with results of the CFD simulation which employed the scale-adaptive simulation (SAS) turbulence modelling. The presented work was done in scope of a national applied research project titled Development of a computer-aided visualization method for the diagnostics of velocity fields in hydrodynamic systems.

Introduction Lovrenc Novak Laboratory Laboratory for Hydraulic Turbomachinery E-mail lovrenc.novak@fs.uni-lj.si Room N5 Phone +386-1-4771-423 Status Researcher / Assistant Research area Computational and experimental fluid dynamics

A remarkable insight into the mechanics of flow around bodies is nowadays made possible by the numerous experimental and numerical methods. On the numerical side, new and improved numerical models enable better flow prediction accuracy and detail. At the same time, the constantly increasing computing power enables resolution of flow phenomena down to the smallest spatial and temporal scales. However, new and improved methods emerge also on the experimental side. In the current work, one improved experimental (visualization based) method and two numerical (turbulence related) methods are compared by studying of vortex shedding from an airfoil at low Reynolds number. Experimental approach is based on flow visualization, which enables capturing of complex time-dependant flow structures. Recorded images were analysed by a recently developed method for computer aided visualization, which enables calculation of velocity field from a series of images containing illuminated layer of fluid with a pollutant mixed in. The calculation method employed here is based on the advection-diffusion equation and was first introduced in [1]. The method is implemented into software ADMflow, which provides all the required pre- and post-processing tools for application of the method through a graphical user interface. More information on the theoretical principles

Laboratory for Hydraulic Turbomachinery Yearbook 2014

of calculation engine and a detailed evaluation of ADMflow performance can be found in [2]. In addition to the experiment, numerical (CFD) calculations were performed to calculate flow around the airfoil. Key issue for most applications of CFD, including flows around airfoils, is modelling of turbulence. When detailed information on turbulent structures is required, modelled turbulence by the Reynolds Averaged Navier Stokes (RANS) method does not provide sufficient information. In such cases, turbulence has to be resolved by using methods such as the Large Eddy Simulation (LES). This means substantially increased computing demands due to finer grids required. An alternative approach is to use one of the hybrid RANS-LES models, such as the Detached Eddy Simulation (DES) method, or the Scale-adaptive simulation (SAS), which in many ways behaves similar to the DES. The SAS model was developed by Menter and Egorov [3] and is in fact an improved unsteady RANS model, with LES capability in unstable flow regions. The SAS model is based on introduction of the von Karman length scale into the turbulent length scale equation of a two-equation turbulence model. This provides means for local detection of unsteadiness and automatic balancing between contributions of modelled and resolved turbulence stresses. In regions of unsteady flow, the SAS term in the turbulence length scale equation will act to reduce the turbulent viscosity

and its damping effect on the resolved fluctuations, therefore promoting the momentum equations to operate in unsteady mode. The detailed equations for the SAS model are found in the literature [3]. The SAS model, coupled with the widely popular SST model, is available in code Ansys Fluent 13.0 which was used for the simulations. Experimental setup Measurements with flow visualization were performed in a low-speed wind tunnel as shown on Fig. 1. The test section of the tunnel has a square 100 x 100 mm cross-section and is 800 mm long. A NACA 4421 airfoil with a chord length of 30 mm was used as the test object. The airfoil was mounted horizontally in the middle of the test section, spanning its entire width. All experiments were performed with the airfoil at 3° angle of attack, which ensured flow separation and vortex shedding to occur on the airfoil suction side, while the boundary layer on the airfoil pressure side remained stable. 2

10 6

5 3

4 6

Figure 1: Experimental station; 1 – smoke generation wire, 2 – flow straightener, 3 – contraction section, 4 – test section, 5 – airfoil, 6 – LED lighting, 7 – high-speed camera, 8 – flow straightener, 9 – radial fan, 10 – frequency inverter.

Experimental flow visualization was achieved by introducing passive tracer smoke into fully developed flow. Smoke was generated by vaporizing paraffin oil on a heated coil, made of a thin stainless steel wire. A Fastec HiSpec4 2G Mono high-speed camera was used to record visualized flow. Recording frequency was 8103 frames per second using 121 μs exposure time. Recorded images had 8-bit grey level depth and resolution of 752x224 pixels, where pixel size was calculated as p = 0.09 mm. Due to the short exposure time, a powerful source of illumination was needed. For that purpose two LED lights were used, one of them placed above and one below the airfoil. Numerical setup Geometry of the numerical model was made to represent the actual wind tunnel where the experiments were conducted. The computational domain covers the entire height of the wind tunnel (100 mm) in direction normal to the airfoil (y coordinate) and has a length of 200 mm in streamwise direction (x coordinate) with the airfoil leading edge at x = 50 mm. A C-type grid with 473 cells in wall parallel direction and 62 cells in wall normal direction was generated in the x-y plane of the domain (Fig. 2). This grid was extruded in the spanwise direction with 37 cells of constant width, totalling to 50% chord length or 15% of the actual tunnel width. The grid consisted of a total of 1085062 hexa cells. Initial cell height on the airfoil was set to produce values of y+ < 1.6.

Top and bottom of the domain were defined as no slip walls, since they represent the actual wind tunnel walls. Periodic boundary conditions were imposed on the spanwise boundaries. Inlet velocity profile was set according to the measured data, obtained with the hot-wire anemometry at 18 equally spaced locations across the channel height. The average inlet velocity was 4.94 m/s. Turbulence intensity of 1.5% was computed from the hot-wire measurements and set as inlet turbulence, while the inlet turbulent length scale of 5 mm was estimated based on the upstream flow straightener geometry. Outlet was defined as a pressure outlet with constant relative pressure of 0 Pa.

Figure 2: Computational grid Time step for the simulation was chosen with respect to the computed local Courant numbers and to the frame rate of the high speed camera (8103 fps) that was used in the experiment. Calculations at a time step of 4.1137e-05 s, which equals one third of the camera’s time per frame, produced Courant numbers in the wake area ranging between 0.3 and 1.8, with average slightly below the recommended value of unity. Both the SST model in unsteady mode and the SSTSAS model were used for the simulations. Initial conditions for the simulations were always provided by the steady state solution using the SST model. Both models were used with the default coefficients as set in Ansys Fluent 13.0 [4]. All calculations were performed on the HPCFS high performance computer in Ljubljana using 24 CPUs. Computational time per second of simulated time was around 8 hours for both turbulence models. Excellent convergence in time was achieved with periodicity in residual values and main flow variables. Results Flow separation on the airfoil suction side and generation of a Karman vortex street was captured by both the numerical and experimental approach. Fig. 3 shows two experimentally obtained states, separated by a time interval of 1 ms. Examination of subsequent images, which are not presented here due to space constraints, shows that the patterns repeat periodically with only slight differences, caused by the randomness in smoke injection intensity. A sequence of such experimental images were used as an input for the computer aided visualization (CAV) analysis.

Figure 3: A sequence from the experiment, separated by 1ms of time Yearbook 2014 Laboratory for Hydraulic Turbomachinery

Velocity field in the trailing edge region, where vortex generation and shedding takes place, is presented in more detail on Fig 4. Results of the CFD simulations are compared to the results of CAV by the ADMflow code. Only one condition in time is displayed here due to space constraints. Also, vector lenghts are kept constant in order to better visualize the flow patterns.

SST

SST-SAS

CAV

Figure 4: Instantaneous velocity fields by the SST model, the SSTSAS model and the CAV method

Analysis of a sequence of CFD predicted velocity fields reveals the mechanism of periodic vortex generation and shedding. Conditions are in many ways similar to the well investigated case of vortex shedding behind cylinders, but with significant differences due to the specific, non-symmetric geometry. A vortex with clockwise rotation is generated at the interface of the separated layer and recirculating flow above the airfoil suction side. A vortex with counter-clockwise rotation is generated at the airfoil trailing edge. Both vortices are generated alternately. In the growing phase the centres of both vortices move downstream and start to accelerate more rapidly once they reach the trailing edge, where they are shed to form the von Karman vortex street. Calculations with the SST model predict very stable periodic flow conditions, with a simpler and more organized vortex patterns compared to the SST-SAS model calculations, where instability is most evident in the recirculating boundary layer. This instability is best reflected in variable inception location of the clockwise (upper) vortex as well as in variation of its intensity and shape as it moves 66

Laboratory for Hydraulic Turbomachinery Yearbook 2014

downstream. The SST-SAS also gives a highly 3-dimensional, non-homogenous and unsteady velocity field in the region of recirculating flow above the airfoil suction side. Generation and dissipation of smaller vortices can be detected in this region. None of these develops in case of the SST model, where flow above the suction side is steady and uniformly directed reverse to the main flow. Velocity fields calculated by the CAV method follow similar patterns to the CFD results. Most evident differences occur at the lower part of observed window, under the airfoil pressure side, where velocities are significantly under-predicted. Also, resolution of vortical structures by the CAV method is not always successful. Another notable difference is in the extent of the recirculating flow above the airfoil suction side, where CAV calculations show lower (closer to the airfoil) recirculating region, despite the actual visualization images showing conditions more similar to the CFD predictions. Many deficiencies in the CAV results could stem from the actual input images with tracer not carrying enough information (greyness gradients) for the algorithm to be able to calculate proper velocities. This situation is especially evident in larger black regions (no smoke) or in regions of unperturbed smoke trails, such as upstream the airfoil or in the attached boundary layers. Another reason for lost detail is related to the fact that the studied flow is complex and highly unsteady and since the CAV method utilizes some temporal and spatial smoothing it is inevitable to introduce a certain amount of averaging in the results. Further comparison of calculated velocities for the SST, the SST-SAS and the CAV cases was done for three vertical sample lines, shown on Fig. 5. Here, line1 is located at the trailing edge, line0 is one third chord length upstream and line2 one third chord length downstream of line1. Time-averaged values and fluctuation intensities were analysed for approximately 39 vortex shedding periods long data series. Generally, both turbulence models show good agreement in the time-averaged values, however the predicted velocity fluctuations are larger in case of the SST-SAS model. Significant differences occur in spanwise velocities, where the SST model shows almost steady-state conditions, whereas the SST-SAS clearly predicts a time-variable spanwise component. The CAV results indicate significantly less temporal variation in velocity values compared to the numerical results. This could be partly attributed to the methodâ&#x20AC;&#x2DC;s inherent timeaveraging. Another influence to be considered in relation to the CAV is the presence of flow in the third (spanwise) dimension, which the method cannot resolve. Line0 Line1 Line2

Figure 5: Sample lines

To determine the accuracy and errors for both the CAV and CFD results it would be essential to have reference measured velocity data. Since this is currently not available, comparisons of calculated velocities are focused only at highlighting similarities and differences between the two turbulence modelling approaches on one side and the CAV method on the other side.

Final comparisons of the CFD, CAV and experimental results were done by carrying out frequency analysis on sampled data at line0, line1 and line2. Time series for each line were created as sequences of average values of each velocity component and greyness level along the line. Fast Fourier transform (FFT) algorithm was used on all the time series to generate the power spectra. The experimental greyness spectrum at line2, shown on Fig. 6a, has strong peaks at 250 Hz, 500 Hz and 750 Hz, corresponding to the fundamental vortex shedding frequency and its harmonics. Peaks at exactly the same frequencies can be seen for the CAV vertical velocity at line2 (Fig. 6b). This means that the CAV method was time-accurate at capturing dynamic properties of the observed fluid flow. In case of the SST and SST-SAS calculations, the first peak occurrs at around 280 Hz, which is an overprediction of 12% compared to the experimental data. First harmonic can be clearly seen at around 560 Hz in case of the SST model, where the spectra are very smooth compared to the SST-SAS results which reflect higher scatter of frequencies in the data.

Figure 6: Frequency spectra of (a) experimentally obtained average greyness level along line0, line1 and line2 and (b) computed average vertical velocity along line2 for different models.

Conclusions Vortex generation and shedding from a NACA4421 airfoil at low Reynolds number was investigated numerically and experimentally. Numerical simulations were performed with two different turbulence modelling approaches, the SST model and the SST-SAS model. Experimental flow visualization was enhanced by calculation of velocity fields from the resulting images by the computer aided visualization (CAV) code ADM-flow. Comparisons between results by the different methods showed significant differences. Both the SST and the SST-SAS models predicted periodic shedding of large scale vortices and formation of

flow patterns, consistent with the experimentally observed flow. Investigation of velocity fields showed increased complexity and non-homogeneity predicted by the SSTSAS model, with a higher degree of velocity fluctuations present. The SST-SAS calculations produced a fully 3D velocity field, whereas the SST calculations resulted in a quasi 2D velocity field with marginal spanwise velocity component. The SST model produced a highly periodic and stable flow, whereas the SST-SAS model predicted slight instabilities in the recirculating boundary layer. The increased complexity of turbulent flow produced by the SST-SAS model also reflects in the frequency domain, with spectra showing a wider range of contained frequencies compared to the SST model. The fundamental vortex shedding frequency predicted by the numerical simulations is around 12% higher than the frequency, calculated from experimental data. The CAV method produced velocity fields similar to the CFD predicted flow. Most notable differences between the CAV and the CFD velocity fields could be seen in the extent of recirculating flow above the airfoil and in the resolution of large scale vortices, where the CAV method performed inconsistently. Reasons for decreased accuracy of the CAV method could partly be related to its sensitivity to the quality of input images, which should ideally contain sufficient greyness gradients across the whole area of interest. Another major reason affecting the CAV performance could be attributed to the fact that the studied flow was complex and highly unsteady, which probably led to higher errors due to the temporal averaging and spatial smoothing, inherent to the method. However, the dynamics of the flow in terms of vortex shedding frequency was captured by the CAV method with excellent agreement to the experiment. Other effects on the CAV uncertainty, such as reduction of 3D flow to a 2D plane, are also not to be neglected and should be evaluated in the following investigations, preferably by including accurate local velocity measurements to serve as a reference for validation of the method. References [1] Bajcar, T.; Širok, B.; Eberlinc, M. (2009). Quantification of flow kinematics using computer-aided visualization, Strojniški vestnik – Journal of Mechanical Engineering, Vol. 55, No. 4, 215-223 [2] Bizjan B.; Orbanić A.; Širok B.; Kovač, B.; Bajcar T.; Kavkler, I. (2014). A computer-aided visualization method for flow analysis, Flow Measurement and Instrumentation, Vol. 38, Aug. 2014, 1-8 [3] Menter, F. R.; Egorov, Y. (2010). The scale-adaptive simulation method for unsteady turbulent flow predictions, Part 1: Theory and Model Description, Flow, Turbulence and Combustion, Vol. 85, No. 1, 113-138 [4] ANSYS Inc. (2010). ANSYS Fluent Release 13.0 User’s Guide List of publications [1] Bizjan B.; Orbanić A.; Širok B.; Bajcar T.; Novak L.; Kovač B. (2014). Flow Image Velocimetry Method Based on Advection-Diffusion Equation, Strojniški vestnik – Journal of Mechanical Engineering, Vol. 60, No. 7-8, 483-494

Yearbook 2014 Laboratory for Hydraulic Turbomachinery

Aerodynamics analysis of close range precision spraying Abstract → Close range precision spraying of discrete foci at their early development stage is a promising method for reduction of pesticides use. Close range precision spraying airflow targets are of size around 15 cm in diameter, preventing generation of very large turbulent coherent structures in the airflow. To enhance local turbulence and leaves fluctuations and to enable spraying of both sides of leaves, rotating airflow screen was used. With rotating airflow screen,

airflow velocity power spectrum at location of the target shows discrete peaks of velocity fluctuations. Close range precision spraying was performed on young vines. Fast camera and image analysis were used for analysis. Natural frequency of individual leaves showed sharp peaks of fluctuations at discrete frequencies. Leaf RMS velocity fluctuations corresponds well with airflow turbulence levels.

Introduction

Aleš Malneršič Laboratory Laboratory for Hydraulic Turbomachinery E-mail ales.malnersic@fs.uni-lj.si Room N4 Phone +386-1-4771-422 Status PhD student (started: September 2014, to be completed: November 2016) Research area System for precision spraying in orchards and vineyards Mentor Assoc.Prof.Dr. Marko Hočevar

Precise application of pesticides supported by up-to-date technologies of target application is one of the most promising options of pesticide quantity reduction. The concept of precise application of pesticides involves adjustment of chemical application to the needs of the target. Hence, there is a need to develop and introduce new techniques and systems for disease detection and pesticide application able to optimize the distribution of pesticides according to the specific characteristics of the target, such as diseases susceptibility, or the presence of infection symptoms. Flow aerodynamics around plants and leaves Fluid flow carries pesticide droplets and exhibits pressure forces to surfaces of leaves and branches. Flow field of fluids around deformable bodies such as plants and their interaction with fluid flow are very complex and challenging problem in mechanics of fluids (Delele et al., 2007; Endalew et al., 2010). Studies of aerodynamic properties inside canopies have been initiated by studies of drag properties of plants. Reviews of the turbulence structure above and within the canopy were presented by Raupach et al. (1991) and Finnigan (2000). Canopy turbulence is dominated by intermittent coherent structures that are responsible for most of the momentum and scalar fluxes, and whose length

Laboratory for Hydraulic Turbomachinery Yearbook 2014

scales are of the order of the canopy height. Large coherent structures, enable good propagation and interaction with the plant. To some extent, large coherent structures are responsible for flux of pesticide droplets to adaxial sides of leaves. In comparison with conventional spraying, coherent structures formed by SEF are too small to effectively carry spraying droplets around plant leaves and spray adaxial sides of leaves. Another option for spraying of adaxial side of leaf is to include fluctuations in the spraying airflow. Fluctuations of the spraying airflow cause non-uniform pressure loading and movement of the plant. Close range spraying end effector The spraying end effector (SEF) was designed with the aim to perform precision spraying of small patches of infected areas. SEF consists of the following components: airflow generator, airflow nozzle, pesticide nozzle with anti-dripping device, pump for pesticide, electrical connections (power supply and control signals), pesticide connection, and chassis with optional electronics. The SEF

Figure 1: Close range spraying end effector. Left: back view of the model, right: front view of the model.

1.0

1Hz 0,4m SEF 1Hz 0,4m 2xSEF

0.8 0.6

amplitude [-]

provides narrow spray angle around 30° and 150 µm droplets size. SEF includes a two stage axial fan with counter rotating rotors.

0.4 0.2

MEASUrEMEnt AnD AnALYSIS MEtHoDS

0.0 0

frequency [Hz]

Figure 4: Spectra of plant leaves displacements at distance 0.4 m: with excitation, screen rotational frequency is 1 /s. 1

1Hz 0,6m SEF 1Hz 0,6m 2xSEF

0.8 0.6 amplitude [-]

Measurements of plant/airflow interaction were performed according to setup shown in Fig. 2. One or two SEF were used with or without rotating airflow screenbto introduce periodic fluctuations. We used a ficus pot plant with height 70 cm. Plant movements were recorded by a fast camera and later analyzed by image analysis algorithm. The camera used for image acquisition was B&W FASTEC HISPEC 4. Camera operated at resolution 640x600 pixel, 200 Hz frame rate and exposure time 50-600s, number of images acquired was 5000. Nikkor 50 mm F1:1.2 lens was used. Illumination was provided by eight LED lights CREE XM-L T5. Image analysis was performed on recorded images to estimate displacements and velocities of leaves from two successive images using normalized cross correlation method.

0.4 0.2 0 0

frequency [Hz]

Figure 5: Spectra of influence of distance and rotation speed on leaves displacements, screen rotation 1/s. 1

0,6Hz 0,6m SEF

0.8 0.6 amplitude [-]

0.4 0.2

Figure 2: Measurement setup: 1) SEF, 2) rotating airflow screen with four holes, driven by electric motor, 3) electric motor variable drive, 4) PC, 6) ficus plant, 7) high speed camera.

rESULtS Fig. 3 shows behaviour of plant without periodic excitation and Fig. 4 with periodic excitation. Case with periodic excitations relates to the case with rotating screen and case without periodic excitations relates to case without rotating screen, that is continuous operation of SEF.Without periodic excitation amplitudes of leaves displacements are low. With excitation, peaks of plant oscillations are around 10x higher and located at narrow frequencies bands. Two SEF with exitation produce more peaks of fluctuations. Influence of spraying distance for the case with excitation is shown in Fig. 4 (0.4 m) and Fig. 5 (0.6 m). Intensity of leaves fluctuations is reduced for about 2x.To study influence of frequency of airflow fluctuations on leaves fluctuations, we varied rotational speed of the screen. Influence of frequency of airflow fluctuations is shown in Fig. 5 and 6. Reduction of screen rotational speed causes frequency of leaves fluctuations to shift to lower frequencies. 1.0

0,4m SEF 0,4m 2xSEF

0.8 amplitude [-]

0.6 0.2 0.0 5

frequency [Hz]

Figure 6: Spectra of influence of distance and rotation speed on leaves displacements screen rotation 0.6 /s.

ConCLUSIon For experiment, a spraying end effector was build and equipped with rotating airflow screen, which induced discrete frequency peaks of velocity fluctuations. Measurements of displacement of leaves in the airflow have shown that leaves fluctuate with discrete frequencies. For actual spraying application, the rotating airflow screen should be replaced with more compact device. references [1] Delele, M.A., Jaeken, P., Debaer, C., Baetens, K., Endalew, A.M., Ramon, H., Nicolaï B.M, and Verboven, P., CFD prototyping of an air-assisted orchard sprayer aimed at drift reduction. Comp. elect. Agric. 55 (1), 16 27, 2007. [2] Endalew, A.M., Debaer, C., Rutten, N., Vercammen, J., Delele, M.A., Ramon, H., Nicolaï, B.M. and Verboven P., Modelling the effect of tree foliage on sprayer airflow in orchards. Boundary-layer Meteorol. 138, 139-162, 2010. [3] Raupach M.R., Antonia R.A., Rajagopalan S., Rough-wall turbulent boundary layers, Appl. [4] Mech. Rev. 44:1 25, 1991. [5] Finnigan, J., Turbulence in Plant Canopies, Annual Review of Fluid Mechanics Vol. 32: 519-571, 2000.

List of publications

0.4

Figure 3: Spectra of plant leaves displacements at distance 0.4 m without excitation.

[1] Malneršič A., Hočevar M., Širok B., Interakcija med zračnim tokom in rastlino pri ciljnem pršenju, Kuhljevi dnevi 2012, 137-144, 2012. [2] Malneršič A., Hočevar M., Širok B., Marchi M., Tirelli P. and Oberti R., Close range precision spraying airflow/plant interaction, Proceedings of the first International Conference on Robotics and associated High-technologies and Equipment for agriculture, 107-112, 2012. YEARBOOK 2014 LAborAtorY For HYDrAULIC tUrboMACHInErY

Advanced orchard spraying ABstRACt → One of several ways to reduce spray consumption and minimize drift in orchard spraying applications is to spray trees using a directed spray which is oriented towards a tree canopy according to its size and shape. By doing so the mist coming from the nozzles does not miss or go past the foliage but is blown through the foliage where active substances

Aljaž osterman Laboratory Laboratory for Hydraulic turbomachinery E-mail aljaz.osterman@fs.uni-lj.si Room n4 Phone +386-1-4771-453 Status researcher with PhD Research area Multiphase flow, cavitation, spraying

should be deposited. In this short paper several orchard applications are presented: orchard spraying with a sprayer with continuously moving geometry of its spraying arms, monitoring of growth of fruit trees during growing season with LIDAR and fruit thinning with selective spraying based on precision GPS.

SPrAYEr wItH MoVInG GEoMEtrY For efficient orchard spraying a sprayer with continuously moving geometry of its spraying arms was developed. Spraying arms were continuously adapting to shape and density of a tree canopy. To realize proper movements spraying arms were equipped with hydraulic cylinders and position sensors. The sprayer itself was equipped with a velocity sensor to automatically cease spraying when the tractor stopped. Tree canopy was measured with 2D LIDAR mounted on the sprayer. Scans were positioned perpendicularly to driving direction. In this way data of canopy height and its horizontal distance from the sprayer

(for the side closer to LIDAR) were obtained. Sampling frequency of LIDAR was 50 Hz so during spraying consecutive scanned planes were only a few centimeters apart. Based on LIDAR measurements with very high spatial resolution and accuracy, the canopy at a given position along the row was characterized by determining its contour. From simplified contour the required nozzle positions and inclinations for each spraying arm were calculated. At the same time LIDAR measurements were used to estimate canopy density on the basis of a number of detected targets – received reflections of a laser beam inside a specific region of interest. Both results were used to control electromagnetic valves for hydraulic cylinders and pesticide spraying.

Figure 1: Hydraulically driven moving geometry of the spraying arms

LAborAtorY For HYDrAULIC tUrboMACHInErY YEARBOOK 2014

conditions. Spray quality was estimated by spray coverage detected with water-sensitive papers mounted on trees. Results showed greatly improved coverage. This means that with such sprayer significant pesticide savings can be achieved. trEE GrowtH MonItorInG Throughout the year growth of trees in an orchard is affected by weather conditions, diseases, human activities, etc. Orchard measurements were performed over a whole growing season and were done with a moving platform on which 2D LIDAR was mounted.

Figure 2: Positioning of the spraying arms based on canopy data obtained with LIDAR

Considering such spraying as a time-dependent process, each scan was saved in a local database with a time stamp and current driving velocity. At a given location the appropriate scan was taken from the database for further processing based on a traveled distance from the point where measurements were taken. This was done to take into account the distance between LIDAR and the spraying arms causing some time offset dependent on sprayer velocity.

Figure 4: LIDAR measurements of apple trees

The platform consisted of a vehicle, a computer and a navigation system. The vehicle went always the same route. It moved at a walking pace so that the distance between consecutive scans was just a few centimeters.

Figure 5: Cross-sectional areas of a tree canopy segment (obtained from rasterized LIDAR measurements) Figure 3: Orchard sprayer with LIDAR (1), computer (2), electronics (3), electromagnetic valves (4), spraying arms and air ducts (5), tank (6) and radial fan (7)

Regulation and control were realized in LabVIEW programming language. The sprayer was equipped with a computer which was in interaction with sprayer hardware via digital and analog National Instruments cDAQ modules. LabVIEW was found to be very suitable for a given purpose because different tasks can be divided into independent subprograms which run similarly as the data flow goes. Finally the sprayer was tested in real orchard

During the measurements a precise location of the vehicle was being determined with a GPS module. Precise positioning with centimeter-range accuracy was based on raw GPS data captured with a precision-timing GPS module mounted on the vehicle, Real-Time Kinematics GPS algorithms running on the computer and the second GPS signal from a referential base station. Each LIDAR scan was tagged with a precise position where it was taken and then combined data were saved to the computer. Comparison of the data obtained during the year for the same positions showed that shapes of trees evolve in a predictable way. YEARBOOK 2014 LAborAtorY For HYDrAULIC tUrboMACHInErY

Such observations through time can be used for planning of summer pruning of fruit trees. Finally, based on growth measurements including size, shape and density of a tree crown, a way to establish a referential starting point for variable-rate spraying was proposed. Fruit thinning based on precision GPS Experimental fruit thinning was performed in an orchard on apple trees of Elstar and Brina variety. Fruit thinning, which gives bigger fruit and reduces the effect of alternate bearing, was done by spraying during development of fruit (phenological growth stage 71 using the BBCH-scale). A precision GPS module was added to an air-blast sprayer previously upgraded with electromagnetic valves, relays, National Instruments Compact Data Acquisition (NI cDAQ) system and a computer.

GPS signals were processed by RTKLib program package. In this way real-time positioning in a cm-range was possible. Precise locations of apple trees were determined from recorded pictures. Location of a tree was determined so that a trunk was in the center of the picture. Locations and corresponding number of flowers were stored on the computer mounted on the sprayer. During thinning GPS positioning was done in the same way (as during the recording of trees). Based on a current sprayer location the number of flowers of the nearest tree was compared against a threshold value. When the number of flowers was higher than the threshold value the electromagnetic valves, which were mounted before the spraying nozzles, opened and the tree was sprayed. In this way only overflowering trees were thinned.

Figure 6: Low-cost precision positioning system based on Real-Time Kinematics GPS:1) laptop, 2) antenna cable, 3) GPS antenna, 4) USB GSM modem, 5) USB cable, 6) precision-timing GPS receiver module

Selective spraying was based on a number of flowers on each individual apple tree. Flowers were manually counted during full flowering when trees were also recorded with a camera. For each picture a corresponding GPS position was logged. Precise GPS positioning was based on a u-blox LEA-6T module and Real-Time Kinematics for which a base station signal was obtained via GSM modem from Slovenian GPS network SIGNAL.

Figure 8: Relative GPS positions of Brina trees with corresponding numbers of flowers

For manipulation of electromagnetic valves based on a number of flowers a LabVIEW program was written. Selective spraying of alternating apple trees resulted in reduced spray consumption, more evenly loaded trees and higher quality of apples. It is also expected that with such spraying alternate bearing is reduced in the future. Acknowledgement Parts of described work were funded by the EU as a part of the 7 FP research project CROPS (grant agreement number 246252).

Figure 7: Tractor AGT 835 with sprayer: 1) electronics box, 2) electromagnetic valves, 3) GPS antenna, 4) speed indicator, 5) fan, 6) 5 spraying nozzles, 7) tank, 8) pump 72

Laboratory for Hydraulic Turbomachinery Yearbook 2014

List of publications [1] Osterman A. et al.: Introducing low-cost precision GPS/GNSS to agriculture. Actual tasks on agricultural engineering] : proceedings of the 41st International Symposium on Agricultural Engineering Opatija, Croatia, 19-22 February 2013. [2] Osterman A. et al.: Real-time positioning algorithm for variable-geometry air-assisted orchard sprayer.Â Computers and electronics in agriculture, doi:Â 10.1016/j.compag.2013.08.013, 2013 (98) 175182.

[3] Osterman A., Godeša T., Hočevar M.: Canopy adapted orchard spraying. Actual tasks on agricultural engineering : proceedings of the 42nd International Symposium on Agricultural Engineering, Opatija, Croatia, 25-28 February 2014. [4] Osterman A. et al.: Fruit thinning with selective spraying. Actual tasks on agricultural engineering : proceedings of the 42nd International Symposium on Agricultural Engineering, Opatija, Croatia, 25-28 February 2014 [5] Ostermna A., Hocevar M., Godeša T.: LIDAR-based control of automated orchard sprayer. International Conference of Agricultural Engineering, Zurich, 6-10 July 2014. [6] Osterman A. et al.: Monitoring of growth of fruit trees during growing season with LIDAR. Proc. of RHEA2014: new trends in mobile robotics, perception and actuation for agriculture and forestry. Madrid, 2014, p. 123-129. [7] Osterman A. et al.: Moving spraying arm with internal airflow for air-assisted orchard sprayer. International Conference of Agricultural Engineering, Zurich, 6-10 July 2014. AgEng 2014 [8] Osterman A. et al.: Unilateral characterization of tree canopies in orchards with lidar. Proc. of RHEA2014: new trends in mobile robotics, perception and actuation for agriculture and forestry. Madrid, 2014, p. 93-100. References [1] Balsari, P et al. (2008). A system for adjusting the spray application to the target characteristics. Agricultural Engineering International: The CIGR Ejournal, 10. [2] Brown, D. L. et al. (2008). Targeted spray technology to reduce pesticide in runoff from dormant orchards. Crop Protection, 27(3–5), 545–552. [3] Centner, T. J., Colson, G., & Lawrence, A. (2014). Assigning liability for pesticide spray drift. Land Use Policy, 36, 83–88. [4] Cross, J. et al. (2001). Spray deposits and losses in different sized apple trees from an axial fan orchard sprayer: 1. Effects of spray liquid flow rate. Crop Protection, 20(1), 13–30. [5] Downey, D. et al. (2011). “Smart” sprayer technology provides environmental and economic benefits in California orchards. California Agriculture, 65(2), 85–89. [6] Escolà, A. et al. (2011). Performance of an Ultrasonic Ranging Sensor in Apple Tree Canopies. Sensors, 11(3), 2459–2477. [7] Gil, E. et al. (2007). Variable rate application of plant protection products in vineyard using ultrasonic sensors. Crop Protection, 26(8), 1287–1297. [8] Harmsen, S.R., and Koenderink, N.J.J.P. (2009). Multi-target tracking for flower counting using adaptive motion models. Computers and Electronics in Agriculture 65, 7–18. [9] Hehnen, D. et al. (2012). Mechanical flower thinning improves fruit quality of apples and promotes consistent bearing. Scientia Horticulturae 134, 241–244.

[10] Hocevar, M. et al. (2010). Design and testing of an automated system for targeted spraying in orchards. Journal of Plant Diseases and Protection (JPDP), 117(2), 71–79. [11] Krasniqi, A.-L. et al. (2013). Quantifying key parameters as elicitors for alternate fruit bearing in cv. “Elstar” apple trees. Plant Science 212, 10–14. [12] Lead Holder, H. et al. (1997). Compendium of growth stage identification keys for mono-and dicotyledonous plants. Novartis, Basel, ISBN 3-9520749-3-4. [13] Llorens, J. et al. (2011). Ultrasonic and LIDAR Sensors for Electronic Canopy Characterization in Vineyards: Advances to Improve Pesticide Application Methods. Sensors, 11(2), 2177–2194. [14] Meier, U. et al. (1994). Phänologische Entwicklungsstadien des Kernobstes (Malus domestica Borkh. und Pyrus communis L.), des Steinobstes (Prunus-Arten), der Johannisbeere (Ribes-Arten) und der Erdbeere (Fragaria x ananassa Duch.). Nachrichtenbl. Deut. Pflanzenschutzd. 46, 141–153. [15] Rosa, U.A. et al. (2008). An electro-mechanical limb shaker for fruit thinning. Computers and Electronics in Agriculture 61, 213–221. [16] Rosell Polo, J. R. et al. (2009). A tractor-mounted scanning LIDAR for the non- destructive measurement of vegetative volume and surface area of treerow plantations: A comparison with conventional destructive measurements. Biosystems Engineering 102, 128–134. [17] Sanz, R. et al. (2013). Relationship between tree row LIDAR-volume and leaf area density for fruit orchards and vineyards obtained with a LIDAR 3D Dynamic Measurement System. Agricultural and Forest Meteorology, 171–172, 153–162. [18] Sauerteig, K.A., and Cline, J.A. (2013). Mechanical blossom thinning of “Allstar” peaches influences yield and quality. Scientia Horticulturae 160, 243–250. [19] Takasu, T. (2009). RTKLIB: Open Source Program Package for RTK-GPS, FOSS4G 2009 Tokyo, Japan, Nov 2, 2009. http://www.rtklib.com/, accessed Sept 26, 2013. [20] Walklate, P. J., Cross, J. V., & Pergher, G. (2011). Support system for efficient dosage of orchard and vineyard spraying products. Computers and Electronics in Agriculture, 75(2), 355–362.

Yearbook 2014 Laboratory for Hydraulic Turbomachinery

Rotation generator of hydrodynamic cavitation Abstract → Nowadays, due to lack of freshwater resources a sufficient wastewater management is an environmental concern. This global issue is resulting in the rapid growth of technologies for wastewater treatment. One very promising technology is usage of cavitation. In our laboratory, several studies have been made considering treatment of different kind of wastewater with hydrodynamic cavitation. For this purpose a novel machine (rotation generator of hydrodynamic cavitation – RGHC) has been designed,

built and tested. RGHC presents a machine, which is based on a modified centrifugal pump and can be used in many different processes. Till now several experiments have been performed using hydrodynamic cavitation as a tool for treating different kind of water, e.g.: use of cavitation for sludge disintegration, for pharmaceuticals removal, for disinfection of swimming pool water and for killing the bacteria Legionella pneumophila.

Introduction Martin Petkovšek Laboratory Laboratory for Hydraulic Turbomachinery E-mail martin.petkovsek@fs.uni-lj.si Room N4 Phone +386-1-4771-422 Status PhD student (started: September 2011, to be completed: 2015) Research area Cavitation in thermosensible liquids, wastewater treatment Mentor Assoc.Prof.Dr. Matevž Dular, Co-Mentor: Prof.Dr. Branko Širok

Cavitation as a phenomenon is characterized by a formation, growth and collapse of bubbles within a liquid. It forms, when the local pressure drops below the vaporization pressure for which two main reasons are possible. By ultrasonic cavitation, the acoustic waves cause the local pressure fluctuations, which are the cause for local pressure drop, while by hydrodynamic cavitation, the geometry of a system is the reason for velocity fluctuations in a liquid flow, which can cause local drop in pressure. Locally seen the cavitation is a process of evaporation, gas expansion, condensation and gas compression [1]. The cavitation bubble forms due to evaporation and gas expansion and therefore collects the energy from the surrounding liquid. During the cavitation bubble collapse, where the condensation and gas compression occurs, this energy is released. The energy can be released by: (i) spherical bubble collapse in shape of pressure impulse in an order of several hundred bars [2], (ii) at solid boundary the cavitation bubble collapses asymmetrically, where consequently micro jet is formed, which can reach the velocities in order of 100 m/s [3]. By spherical bubble collapses also very high temperatures of several thousand kelvins occur (theoretically), but these last for very short time – in 1μs the temperature

Laboratory for Hydraulic Turbomachinery Yearbook 2014

drops to the temperature of the surrounding liquid [4]. Such extreme conditions are adequate to rupture or kill biological structures or they can cause water molecules to dissociate into OH and H radicals [5]. Till now series of studies have been made using cavitation for treating different kind of wastewaters. The majority of the experiments have been conducted with acoustic cavitation, but also several studies base on hydrodynamic cavitation. To gain the effect on treating wastewater, the acoustic and hydrodynamic cavitation can be combined. Also different kind of hybrid techniques, so called advanced oxidation processes have been developed for even more sufficient treatment, where they use different kind of chemicals like hydrogen peroxide, carbone tetrachloride, Fenton›s reagents and others with combination of cavitation. There is no general presumption among the techniques, different types of pollutants demands different treatments and have different optimum conditions, due to different mechanisms of removal or destruction. In the paper are presented four areas, where our RGHC was successfully used: (i) at municipal wastewater treatment plant (WWTP) the RGHC was used as a tool for sludge pretreatment to improve and accelerate the anaerobic digestion, which benefits in higher biogas production; (ii) RGHC was also used for treat-

ment, removal of different kind of pharmaceuticals from municipal wastewater; (iii) in swimming pool complex or spa, many different chemical compounds can be formed due to combination of chlor and other organic materials, which were successfully removed with RGHC; (iv) also studies with removal of bacteria Legionella pneumophila with hydrodynamic cavitation were performed, where very encouraging results were achieved. Machine presentation On basis of knowledge and results from the first designed cavitation generator [6], a novel rotation generator of hydrodynamic cavitation (RGHC) was built at Laboratory for Water and Turbine Machines (Faculty of Mechanical Engineering, University of Ljubljana). Based on centrifugal pump design it has a modified rotor and added stator in the housing of the pump (Fig. 1). The RGHC consists of an electric motor (1) with power of 5.5 kW, which drives the modified rotor (2). The stator (3) is placed opposite to the rotor in the axial direction in the housing of the pump. The housing of the pump preserves the original inlet (4) and outlet (5) to retain the possibility of standard installation. The housing, the modified rotor (2) and added stator (3) are forming the cavitation treatment chamber. In addition to treating wastewater with cavitation, the whole RGHC partly still preserves the pumping function, which means that the machine does not require an additional circulation pump to operate.

er diameter and the same number of grooves as the rotor. The difference is at the teeth geometry, where the stator teeth have inclination and they have barriers from three sides of each tooth. The purpose of these barriers is to retain the high pressure outside the cavitation zone, which results in intensification of the cavitation bubble collapse. The main advantage of the presented RGHC is its double function, which means that it works as a cavitation generator and simultaneously as a pump. In compare with conventional hydrodynamic cavitation generators, such as Venturi restrictions or orifice plates, the presented RGHC does not cause additional pressure drops in consisting system. By restrictions, especially by orifice plates is also a risk of potential obstruction, due to small hole diameter. Figure 2 presents visualization with high-speed camera of formed cavitation between the rotor and stator of the machine, where the images were taken at 8000 fps. Hatched line presents the stator tooth on the left side of the images (staying still), while on the right side of the images the rotors teeth are moving upwards, due to rotation.

Figure 2: Visualization with high-speed camera of formed cavitation in RGHC

Figure 1: Rotation generator of hydrodynamic cavitation

Modified rotor and added stator have specially designed geometry, which causes periodically repeating pressure oscillations. Alternately low (below the vapor pressure) and high pressure cause the cavity formation. The type of cavitation which is forming inside the treatment chamber is so called shear cavitation, where cavitation structures are formed due to shear forces, which are caused by the relative movement of rotor, stator and the liquid in between them. The rotors geometry â&#x20AC;&#x153;dragsâ&#x20AC;? the liquid partially in the tangential and partially in the radial direction. The radial velocity of the liquid gives the machine the suction function, while the tangential velocity of the liquid causes the liquid to rotate in the treatment chamber. The rotor has on its diameter of 190 mm a certain number of grooves. These grooves consequently form the so called teeth, which are sticking out of the main core of the rotor in axial direction. The stator has the same out-

With the distance between the rotor and the stator, teeth geometry and with the frequency of the rotor, we are able to control the size and the aggressiveness of the formed cavitation. Results Disintegration of waste activated sludge (WAS) Wastewater treatment plants use the cavitation as a tool for sludge pretreatment to improve and accelerate the anaerobic digestion, which benefits in higher biogas production, mass reduction, pathogen reduction and odour removal. Nowadays, most of commercial technology for sludge treatment, that use cavitation, base on acoustic cavitation (ultrasonication), which is in comparison with hydrodynamic cavitation less complex for implementation, but also less energy efficient. The problem by ultrasonic cavitation devices is that the cavitation effects occur only close to the vibrating surface, which decreases the possibility of the whole sludge to be treated. With our machine we are able to provide a sufficient disintegration of the sludge, which were confirmed with Yearbook 2014 Laboratory for Hydraulic Turbomachinery

our experimental work. At the first stage the analysis of hydrodynamics of the RGHC were made with tap water, where the cavitation extent and aggressiveness was evaluated. At the second stage RGHC was used as a tool for pretreatment of a WAS, collected from Wastewater treatment plant Domžale-Kamnik. In case of WAS the disintegration rate was measured, where the soluble chemical oxygen demand (SCOD) and soluble Kjeldahl nitrogen were monitored and microbiological pictures were taken. The SCOD, one of the parameters for disintegration, increased from initial 45 mg/L up to 602 mg/L and 12.7 % more biogas has been produced by 20 passes through RGHC. The results were obtained on a pilot bioreactor plant, volume of 400 liters.

Figure 3: Disintegration rate measured with SCOD by number of passes through RGHC

experiments were conducted in deionized water and also in real wastewater, where the results show, that for real wastewater effluents matrix composition reduces removal efficiency. Nevertheless the removal efficiency of the real wastewater effluents, by hydrodynamic cavitation, still reaches form 37% up to 79% for individual selected pharmaceutical. Removal of trihalomethanes from water in swimming pools Chemistry of swimming pool water is fundamentally different compared to drinking water. Whilst drinking water is only treated once – if at all – swimming pool water is used and treated repeatedly over extended periods of time. During the use of swimming pools, organic, inorganic and microbial contaminations are continuously introduced into the water by bathers. Additionally, water preparation chemicals as well as their impurities and decomposition products are added to the water, and microbiological activities in parts of the water preparation system may further impair the situation. Trihalomethanes (THMs) are formed as a by-product predominantly when chlorine is used to disinfect water. According to several studies, exposure to THMs may pose an increased risk of cancer. Some THMs have been identified as mutagens, which alerts DNA. Mutagens are considered to affect the genetics of future generations in addition to being carcinogenic. In addition to these risks, THMs are linked to bladder cancer, heart, lungs, kidney, liver, and central nervous system damage. With our experiments, we showed that cavitation formed in our RGHC could be appropriate for THMs reduction in swimming pools. The results in Fig. 5 are shown for sample volume of 1000L, which could be considered as a pilot experiment.

Figure 4: Total biogas production in case of cavitated sludge vs non-cavitated sludge

Pharmaceuticals removal Pharmaceuticals are an important and indispensable element of modern life but parallel to the continuous rise in their consumption, is the increasing burden on the environment posed by pharmaceutical residues. The main sources of these residues are wastewaters that even after conventional (biological) treatment still contain pharmacologically active compounds. Pharmaceuticals can enter the environment through various routes (hospitals, households, unused medicines, animal excretion, etc.), but generally WWTP effluents are considered the most critical point source. While conventional (biological) treatments are insufficient by most pharmaceutical removal, new, advanced methods are being studied. And one of them is cavitation. In our studies (Zupanc et. al 2013, Zupanc et. al 2014, Petkovšek et. al 2013), the removal of clofibric acid, ibuprofen, naproxen, ketoprofen, carbamazepine and diclofenac residues from wastewater, using a hydrodynamic cavitation, has been systematically studied. The

Laboratory for Hydraulic Turbomachinery Yearbook 2014

Figure 5: Removal of THMs

Bacteria Legionella pneumophila Legionella pneumophila is wide spread in all natural fresh water sources in predominantly low concentrations. The bacteria has also frequently been observed in engineered water systems such as warm water distributing systems, cooling towers, humidifiers and fountains. In low concentrations Legionella pneumophila does not represent a significant risk for the health of humans, however the multiplication of the bacteria in water systems poses a potentially fatal (between 15 and 20 % of those infected) human health risk wherever aerosolisation can occur. Especially people with weakened immune system (e.g. cancer patients) are facing with fatal risk. To ensure the

safe environment in health institutions, most common, the so called thermal shocks are used to disinfect water distribution system, which are very energy wasteful and stressful for the plumbing. In our study (Šarc et. al 2014) we show, that bacteria Legionella pneumophila can be effectively eradicated by hydrodynamic cavitation. We show that it is probably not the pressure peaks or the high local temperatures that cause the eradication of the bacteria, but the rapid pressure decrease which was initiated in supercavitating flow regime. Results of the study show promising ground for further optimization of a methodology for Legionella pneumophila removal by cavitation. Conclusions Several studies have been made in our laboratory regarding to treatment different kind of water with hydrodynamic cavitation. Due to need for a cavitation machine, which has a simple design, low operating cost and could be easily to scale up or down, a novel Rotation generator of hydrodynamic cavitation has been designed, built and tested. Results of our experiments show, that presented machine is appropriate for several implementations by different processes, where the treatment of different kind of water is needed. With the geometry of some parts of the machine and its kinematics we are able to adjust the properties of the formed cavitation, such as size and aggressiveness. Our further work is focused to improve the efficiency of the machine and to make an experimental installations on real systems.

[2] M. Zupanc, T. Kosjek, M. Petkovšek, M. Dular, B. Kompare, B. Širok, Ž. Blaženka, E. Heath, Removal of pharmaceuticals from wastewater by biological processes, hydrodynamic cavitation and UV treatment, Ultrasonic Sonochemistry 20 (2013) 1104–1112 [3] M. Zupanc, T. Kosjek, M. Petkovšek, M. Dular, B. Kompare, B. Širok, Ž. Blaženka, M. Stražar, E. Heath, Shear-induced hydrodynamic cavitation as a tool for pharmaceutical micropollutants removal from urban wastewater, Ultrasonic Sonochemistry 21 (2014) 1213–1221. [4] A. Šarc, M. Oder, M. Dular, Can rapid pressure decrease induced by supercavitation efficiently eradicate Legionella Pneumophila bacteria?, Desalination and Water Treatment (2014). [5] M. Petkovšek, M. Dular, M. Mlakar, M. Levstek, M. Stražar, B. Širok, A novel rotation generator of hydrodynamic cavitation for waste-activated sludge disintegration, submitted to Ultrasonic Sonochemistry in 2014. [6] M. Petkovšek, A. Šarc, M. Dular, B. Širok, Cavitation as an alternative solution to fight the Legionella Bacteria; an insidious threat to water supply systems, Ujma 27 (2013) 159-164.

References [1] M. Dular, O. Coutier-Delgosha, Thermodynamic effects during growth and collapse of a single cavitation bubble, Journal of Fluid Mechanics, 736 (2013) 44-66. [2] Y.C. Wang, C.E. Brennen, Shock wave development in the collapse of a cloud of bubbles, Cavitation and Multiphase Flow, FED vol. 194 (1994) 15-19. [3] J.P. Franc, Fundamentals of Cavitation, Kluwer Academic Publishers, 2004. [4] S. Fujikawa, T. Akamatsu, Effects of the non-equilibrium condensation of vapor on the pressure wave produced by the collapse of the bubble in a liquid, Journal of Fluid Mechanics 97 (1980) 481–512. [5] P. Braeutigam, M. Franke, R. J. Schneider , A. Lehmann , A. Stolle, B. Ondruschka, Degradation of carbamazepine in environmentally relevant concentrations in water by Hydrodynamic-Acoustic-Cavitation (HAC), Water Res. 46 (2012) 2469-2477. [6] M. Petkovšek, M. Zupanc, M. Dular, T. Kosjek, E. Heath, B. Kompare, B. Širok, Rotation generator of hydrodynamic cavitation for water treatment, Separ. Purif. Technol. 118 (2013) 415-423. List of publications [1] M. Petkovšek, M. Zupanc, M. Dular, T. Kosjek, E. Heath, B. Kompare, B. Širok, Rotation generator of hydrodynamic cavitation for water treatment, Separ. Purif. Technol. 118 (2013) 415-423.

Yearbook 2014 Laboratory for Hydraulic Turbomachinery

Hydrodynamic cavitation for wastewater treatment Abstract → PhD research work regarding effects of hydrodynamic cavitation for wastewater treatment applications is presented. The hydrodynamic cavitation generator prototype was experimentally tested. High speed camera was used to record cavitation structures and hydrophone for pressure pulsation measurements.

Comparing the results, a new experimental method is to be developed. The method will be directly used for hydrodynamic cavitation generators development and optimisation.

Introduction Tadej Stepišnik Perdih Laboratory Laboratory for Hydraulic Turbomachinery E-mail tadej.stepisnik@fs.uni-lj.si Room N5 Phone +386-1-4771-423 Status PhD student (started: October 2014, to be completed: March 2017) Research area Hydrodynamic cavitation Mentor Prof.Dr. Branko Širok, Co-Mentor: Assoc.Prof.Dr. Matevž Dular

Cavitation is the formation, growth and immediate collapse of vapour or gas bubbles within a liquid. Cavities form in a flowing liquid, where local static pressure is reduced to the liquid saturation point. This occurs when a liquid is subjected to rapid changes of pressure or velocity [1]. During the cavitation formation phase, cavitation bubbles collect the energy from the surrounding fluid and release it during the collapse. It is known that during a bubble collapse pressure shockwaves can reach up to several hundred bars and temperatures of several thousand K can occur. The shockwaves are strong enough to cause damage on subjected machine parts, therefore cavitation is usually undesirable phenomenon [2]. If cavitation power is controlled however, it can be utilized - for instance for wastewater treatment. Extreme conditions caused by violent bubble collapses initiate physico-chemical reactions, which degrade various substances in wastewaters [3]. There are four types of cavitation: hydrodynamic, acoustic, optical and particle induced. In hydrodynamic cavitation local drop of pressure is caused by the velocity fluctuations due to the geometry of a system, most commonly when a liquid is passing through a constricted pipe or by a rotating object in a liquid. In comparison to the other types, hydrodynamic cavitation is more energy efficient and appropriate to operate at large

Laboratory for Hydraulic Turbomachinery Yearbook 2014

scales [2]. This makes hydrodynamic cavitation attractive for implementation in various industry and domestic machines, which will be the focus of following PhD studies in the Laboratory for Water and Turbine machines. Objectives The goal is to design an energy efficient wastewater treatment system basing on hydrodynamic cavitation, satisfying commercial-use requirements. In order to achieve this we will develop an experimental method for the substances degradation prediction by valuing the pressure pulsations. Methodology An experimental method combining modern high-speed camera visualisation and hydrodynamic acoustic measurements will be investigated. The method will base on the presumption that the structural oscillations of the water vapour fraction in the cavitation clouds and the pressure oscillations of the cavitation cloud shedding are directly correlated. Simultaneously the cavitation cloud structure will be visualised and the pressure pulsations measured. Comparing the variables, prediction model for the local pressure diagnostics will be determined. Model will be directly used to design and optimise hydrodynamic cavitation generators.

Results

Figure 1: Hydrodynamic cavitation generator prototype

Experimental setup Prototype cavitation generator Different hydrodynamic cavitation generators have been used during previous studies on wastewater treatment, namely: orifice plates [4], Venturi tubes [2], or rotor/stator installations [5]. Currently a prototype of a cavitation generator (from now HCP) is being tested. Unlike other cavitation generators, novel prototype require a small finite volume of fluid for its operation (V = 3,67 mL). Such cavitation generator is very suitable for biotechnology or chemical research, where small samples has to be analysed. HCP is presented on Fig. 1. The generator has basic cylindrical shape (80 mm length, 20 mm diameter) with a hollow (also cylindrical, 50 mm long, 10 mm diameter). A metal sphere is inserted into the hollow. HCP operates by applying oscillating movement to the sphere or the cavitation generator housing. Relative motion between the sphere and the housing effects in a flow around the sphere. Due to the narrow gap between housing and sphere, high flow velocity occur which results in low pressure regions, required for cavitation generation.

Preliminary experiments using HCP have been conducted in order to assess generated cavitation. Cavitation characteristics were observed for 3 different absolute static pressure values (1 bar, 1,5 bar and 2 bar) and for 3 different sphere diameters (8 mm, 8,5 mm and 9 mm). Fig. 3 (top) shows the sequence of images for 8,5 mm sphere and 1,5 bar of static pressure. The cavitation structures are clearly visible. Observing pressure signal (Fig. 3 bottom), one can assume that 5 negative amplitude peaks are to some extent related to cavitation.

Figure 3: Visualisation of the generated cavitation (top) and corresponding pressure signal (bottom)

Conclusions

Cavitation station Cavitation station is shown on Fig. 2. HCP was attached to mechanical pulsating machine, applying oscillating motion in axial direction:

Effects of hydrodynamic cavitation for wastewater treatment applications are to be investigated during PhD studies. Energy efficient hydrodynamic cavitation generator is to be designed with support of the novelty experimental (1) method. The developed method will be used to predict the where x0 = 10 mm and ω = 50 Hz. The hydrophone (Reson mechanisms of cleaning effects of hydrodynamic cavitaTC 4013) and the static pressure transducer (ABB 2600T) tion for wastewater treatment. The prototype hydrodywere connected to HCP with a thin plastic tube (l = 200 namic cavitation generator, suitable for small samples, mm). Image sequences were acquired using high-speed has already been preliminary tested. Results imply that the hydrodynamic cavitation is detected both by the highcamera (Fastec HiSpec 4), with frequency 8000 fps. speed camera and the hydrophone. References

Figure 2: Experimental setup

[1] Širok, B.; Dular, M.; Stoffel, B. Kavitacija, Ljubljana : i2, 2006 [2] Arroyo, S.; Benito, Y. A theoretical study of hydrodynamic cavitation, Ultrasonics Sonochemistry, 2008, 15, 203 - 211 [3] Zupanc, M.; Kosjek, T.; Petkovšek, M.; Dular, M.; Kompare, B.; Širok, B.; Blažeka, Ž.; Heath, E. Removal of pharmaceuticals from wastewater by biological processes, hydrodynamic cavitation and UV treatement, Ultrasonics Sonochemistry, 2013, 20, 1104 – 1112 [4] Gogate, P.R.; Pandit, A.B. A review of imperative technologies for wastewater treatment, Advances in Environmental Research, 2004, 8, 553 - 597 [5] Petkovšek, M.; Zupanc, M.; Kosjek, T.; Heath, E.; Kompare, B.; Širok, B. Rotation generator of hydrodynamic cavitation for water treatment, 2013, 118, 415 - 423 Yearbook 2014 Laboratory for Hydraulic Turbomachinery

Kinematics of mineral wool fibers in air flow Abstract → Mineral wool fibers are found in mineral wool insulation material which is widely used as a thermal and acoustic insulation. To increase mineral wool production productivity and improve quality of final product which is demanded in modern market, understanding of fundamentals in process of fibers forming, fibers transportation and primary layer forming

is crucial. Increased quality of a product means that the final insulation material has for instance: lower thermal conductivity, better mechanical properties and high level of homogeneity. Numerous studies of ligament and fiber formation were carried out, however, kinematics of fibers flow in surrounding air and primary layer formation on collecting mash is still quite unknown.

Introduction Marko Peternelj Laboratory Laboratory for Hydraulic Turbomachinery E-mail marko.peternlej@fs.uni-lj.si Company Izoteh d.o.o., Brnciceva ulica 15b, 1231 Ljubljana Phone +386-31-427-783 Status PhD student Research area Kinematics of mineral wool fibers in air flow Mentor Prof. Brane Širok, Ph.D.

Mineral wool production line consists of several sub-processes shown on Figure 1. Raw material is first melted in a cupola furnace and the malt, due to gravity, drops on to high speed rotating spinner discs shown on Figure 1 as a region A. Due to adhesion and viscosity forces a thin layer of melt is formed on a disc circumference – region B on Figure 1 where because of high centrifugal forces and tangential velocity, series of instabilities are formed which leads to fibers forming. Mineral wool fibers are blown off with primary air flow which is located close to the fibers forming zone and are then transported as multiphase flow through collecting chamber. Secondary air flow or suction flow helps to transport mixture of mineral wool fibers and air to the perforated mash where a primary mineral wool layer is formed. Primarily wool layer than exits the collecting chamber and travels thru the pendulum system where a secondary layer is formed.

Figure 1: mineral wool production line 80

In this study the zone of mineral wool fibers transportation from the rotating discs to the perforated mash is examined. Fibers transportation starts in a region C shown on Figure 1. Objectives and methods Hypothesis is that the parameters as air inlet velocity magnitude and its direction from nozzle, mass flow rate of fibers, spinner discs angular velocity and other has most significant influence on fiber kinematic and, consequently, on primary layer forming process. Object of the study is to investigate and evaluated process parameters which significantly affect process performances and, consequently, final product. Due to high complexity of actual industrial process where physical phenomena as fluid dynamics of melt at different temperature gradients, heat transfer, ligament formation out of rotating discs, ligament solidification or fiber forming, binder distribution between fibers and other, numerous of simplifications has to be done. 2D transient numerical simulation of rotating disc and axial blowing nozzle has been performed to examine different parameters on fibers transportation behavior. Fibers were modeled as solid particles exiting from rotating disc circumference in a specified angle. Investigated parameters were: 1. v – inlet air total velocity magnitude from nozzle 2. α – inlet air flow entering angle from nozzle 3. q – fibers mass flow rate

Important result is particles distribution on outlet of the control volume. Ideal is uniform particle distribution which in real processes is difficult to achieve. Variability of certain flow variable φ through specified controlled surface is used to determinate mass-weighted uniformity index ym as an evaluation parameter to compare different boundary settings. ∑

(1). (1). – Surface consists of n surfaces with area Ai, φm is mean value of variable φ through the entire surface, ρi and vi are local density and velocity. If uniformity index is 1 it means that particles are evenly distributed on entire surface. Based on time-dependent mass-weighted uniformity index a histogram has been created to evaluate particle distribution on outlet of control volume when the steady state has been reached. Relationship between particle concentration and distance from center axis are shown on two histograms on Figures 3 and 4. ∑

Future experimental work Experimental model has been designed to validate numerical simulations. Experimental model consists of rotating disc with holes on its circumference through fibers are formed. Close to rotating disc an axial blowing nozzle in mounted, which blows formed fibers in to the chamber constructed as a transparent tube, so observation of fibers flow will be possible. Axial ventilator sucks mixture of fibers and air vertically. On top of the chamber and before the ventilator a collecting mash for fibers deposition is located. Molten sugar will be used as fibers forming media.

Figure 6: longitudinal cross section of experimental model. Figure 3: Particle concentration for: v=120 m/s, q=1750 kg/s, and α = 0°

Figure 4: Particle concentration: v = m/s, q = 1750 kg/s, and α = 0°

Figure 5 shows particle concentration in the control volume where significant influence of inlet air flow entering angle α is seen. Inlet air velocity magnitude v is set to 120 m/s and fibers mass flow rate q to 1750 kg/h.

Figure 5: Particle concentration at α=0° (upper figure); α=45° (lower figure)

Conclusions Numerical simulation of fibers flow and air mixture has been investigated and based on different parameters variations can be concluded, that the most significant influence on fibers kinematics in collecting chamber has the inlet air flow entering angle from nozzle. Inlet air total velocity magnitude and fibers mass flow rate has only minor influence on fibers distribution in collecting chamber and, consequently have less influence on homogeneity of formed primary layer. The same effect can be seen in industrial process, where special deflectors are used to deflect air from nozzle. References [1] Širok B., Blagojević B., Bullen P.: Mineral Wool: Production and Properties; Woodhead Publishing Limited, Cambridge, (2008). [2] Bajcar T., Širok B., Bizjan B., Orbanić A.: Mineral wool melt fiberization on a spinner wheel. Chemical Engineering Research and Design, vol. 91, (2013) [3] Bizjan B., Širok B., Govekar E. Nonlinear analysis of mineral wool fiberization process. Journal of computational and nonlinear dynamics [4] Dimovski D., Širok B., Blagojević B., Kariž Z. Mineral wool primary layer structure influence on the fibre distribution in the final product. Glass technology, (2004) [5] Bajcar T., Blagojević B., Širok B., Dular M. Influence of flow properties on a structure of a mineral wool primary layer. Experimental thermal and fluid science, (2007) 81

Assoc.Prof.Dr. Tomaž Katrašnik

Laboratory for Internal Combustion Engines and Electromobility The research and teaching activities of the Laboratory with more than 45 years of experience is focused on internal combustion engines for stationary and mobile applications and on hybrid and fully electric powertrains including fuel cell electric powertrains.

The activities are structured in five areas: • Advanced engine concepts including engine control, • Advanced and alternative fuels and combustion concepts, • Optimization of real world energy consumption and exhaust emissions, • Hybrid and fully electric powertrains including fuel cell electric powertrains, and • Development of mechanistic simulation models in these fields. In the area of Advanced engine concepts including engine control there is a strong focus on Closed Loop Combustion Control and development of demonstrators using real-time hardware. CLCC modules incorporate selfdeveloped prototypes of model based controllers relying on thermodynamics and combustion parameters as well as on virtual emission sensors, with the aim of optimizing Diesel and Otto combustion processes as well as various types of CAI or LTC combustion concepts, e.g. HCCI, PCCI, RCCi... This area fully interacts with the Development of advanced and alternative fuels and combustion concepts area. In this area we research all commercially available fuels for mobile application including dual-fuel operation and extend this focus on innovative waste or bio derived fuels, where the later ones have a significant potential in stationary application. Knowledge arising from previous areas combined with state-of-the-art equipment and profound expertize in real world testing form a solid basis for research in the area of Optimization of real world energy consumption and exhaust emissions. In this area we perform analyses of exhaust emissions and vehicle dynamics on various vehicle types ranging from ICE powered and hybrid-electric

vehicles to heavy duty and off-road vehicles utilizing conventional and advanced/alternative fuels. In the area of Hybrid and fully electric powertrains that includes fuel cell electric powertrains we are engaged in R&D activities on the component level and on the system level since the optimization on the system level is crucial for achieving high energy efficiency and low exhaust emissions. This forms a crucial interaction with the area Development of mechanistic simulation models where we focus on fully coupled multi-domain mechanistically based models, which feature high level of predictability and high level of accuracy. In this are we have long term cooperation with the market leader on the development of professional modelling tools that are in use by multiple OEMs. In all of the above mentioned areas we have a fully equipped laboratory with state-of-the-art hardware and software facilities. Moreover, in these areas we are also engaged in public funded national and international (EU Framework, COST and FFG-Austria) projects as well as direct industrial projects with domestic and foreign partners, which confirms high research level of the Laboratory. All these engagements form a solid basis for the teaching activities that fully reflect all our research activities. During their studies students work with modern research equipment and their work during seminars and thesis is generally interrelated with research work of the Laboratory establishing a challenging and highly motivated working environment.

Real World Emissions and Vehicle Parameters of Different Bus Powertrain Technologies Abstract → Air quality in urban areas is strongly influenced by exhaust emissions emitted from the internal combustion engine powered segment of the public transport fleet. The aim of the study that was conducted within the FP7 CIVITAS ELAN project was to analyze benefits in terms of energy consumption and exhaust emissions when replacing baseline diesel powered EURO III city busses by the CNG powered EURO V busses and hydraulic series hybrid diesel powered EURO V busses. Experiments were performed on the regular bus routes to access realistic energy consumption and exhaust emissions that are characteristic for the city of Ljubljana. During the test runs instantaneous gaseous emission (CO2,

CO, NOx and THC) were measured together with the instantaneous particulate PM10 emissions. In addition, several vehicle parameters were monitored. First, instantaneous emissions are analyzed. These results form the basis for interpreting lumped cycle averaged emissions, which indicate that application of the CNG busses does not necessary reduce CO2 emissions. As expected CNG bus featured nearly negligible particulate emission, which offers a potential to lower particulate emissions in the urban center. Analyzes of the hydraulic series hybrid bus with EURO V diesel engine revealed that powertrain hybridization does not necessarily result in better fuel economy.

Introduction Samuel Rodman Oprešnik Laboratory Laboratory for Internal Combustion Engines E-mail samuel.rodman@fs.uni-lj.si Room 310 Phone +386-1-4771-310 Status PhD assistant Research area Alternative Fuels and I.C.E. Exhaust Emission Measurements

The mobility of the globalized world is supported by internal-combustion engines, which generally use fossil fuels. City inhabitants have to move daily within the city. Therefore cities have public transport systems, including city public bus transport as one of the most important elements. As the majority of people are concentrated in the city, the atmospheric emissions of local buses have a considerable impact on human health. According to Slovenian Environment Agency report the effect of PM10 particles on health is dependent upon their concentration and time of exposure. Long-term exposure to lower concentrations of PM10 has much greater effect than short exposures to high concentrations. It is also stated that at the Ljubljana city center measuring site the average annual concentration of PM10 in 2011 was 44 μg/m3 that is more than the allowed average annual concentration. Moreover, the allowed average daily concentration was exceeded 94 times in the same year.

Laboratory for Internal Combustion Engines and e-Mobility Yearbook 2014

City of Ljubljana participated, as project coordinator, in the CIVITAS ELAN program which is a European funded initiative with the aim to support cities to introduce ambitious transport measures and policies towards clean, sustainable mobility solutions ensuring health and access for all citizens [1]. Many measures were introduced within the program to achieve above mentioned goals. Among them purchase of five hydraulic hybrid buses that meet Euro V emission standard, opening the first compressed natural gas (CNG) fuelling station in Slovenia and introducing twenty CNG-powered buses and the prospect on purchase more new buses to improve the public transport operation fleet impact on city air quality. To be able to evaluate the impact of introduction of new bus powertrain technologies a set of measurements were performed within CIVITAS ELAN program. Conventional bus with diesel engine that subjects to EURO III emission standard and represents the average fleet condition was tested along with two newly intro-

duced buses: hydraulic hybrid bus and CNG-powered bus. Since real-world exhaust gas emissions can substantially differ from homologation test cycles exhaust gas emission measurements were performed in real-world traffic on regular, representative bus route in Ljubljana. Some were performed with passengers on board, some with substitution mass for passengers where the noise level and volume of the equipment was too high. Methodology Tested buses Three buses of different powertrain technologies were tested. MAN NL223 has been chosen since the bus represents the average fleet condition. The second tested bus was Kutsenits Hydra City II that is a series hydraulic hybrid bus. The third tested bus was CNG powered IRISBUS IVECO CITELIS CNG. All three buses main characteristics are shown in Table 1. Measurement equipment PEMS were used to monitor exhaust gas emissions. HORIBA OBS-2200 on-board emission system was used to measure CO2, H2O, CO, THC and NOx emissions. CO2, H2O, and CO are measured using heated NDIR sensor, THC are monitored with FID sensor and NOx with CLD sensor. System measures also exhaust gas standard volume flow, current position and velocity of the vehicle and offers several AI and DI channels. HORIBA OBS-2000 TRPM on-board emission system was used to measure particulate matter. It comprises from filter gravimetric method and real time instrument that is calibrated with filter gravimetric results. Vehicle dynamics was monitored by RaceTechnology DL2 system that combines results from two-axis accelerometer and 20Hz GPS antenna. Measuring equipment was installed on the bus with all the support systems (electrical generator, filters,...) (see Figure 1). Special care has to be devoted to proper installation of exhaust gas volume flow meter and sampling probes.

Test route Route no. 2 has been chosen as representative, test route as it is typically flat that is characteristic for Ljubljana bus routes where uphill or downhill sections are short and rare. It starts in Ljubljana suburb, where higher average velocities are achieved and leads through the city center where low average velocities are to be expected and ends again in Ljubljana suburb. Itâ&#x20AC;&#x2122;s 12.65 km long. Four measurements were performed on MAN and Irisbus bus, two in each direction. Kutsenits bus has a possibility to turn the hybrid mode and power limiter on or off. So four different settings are possible. Four measurements were performed for each setting.

Figure 1: Installed PEMS for gas and PM10 measurement on Kutsenits bus

Using vehicle dynamics measurement data gathered during real-world operation with passengers on-board thorough analysis was performed to get the insight into average bus operation on a bus route. According to the analysis results the driver was requested to stop at each bus stop for a time of 15s to mimic the real-world operation during PEMS measurements that couldnâ&#x20AC;&#x2122;t be performed with passengers on board due to noise and volume of the measuring equipment. The average number of passengers on board was 23. 1725kg of substitution mass was loaded on the buses during tests, 75kg for each person.

Table 1: Main characteristics of the tested buses engine

MAN D 0836 LUH15, turbo intercooled, DI

Mercedes-Benz OM 924 LA. V/1 turbo intercooled, DI

IVECO Cursor 8 CNG EEV turbo intercooled

swept volume

6871 cm3

4801 cm3

7790 cm3

no. of cylinders

compression ratio

18.0:1

11.5:1

max. power

162 kW

160 kW

213 kW

max. torque

850Nm

810Nm

1100 Nm

emission standard

EURO III

EURO V

power transmission

conventional automatic VOITH-Diwa.3 851.3 E

series hydraulic hybrid

conventional automatic VOITH Diwa.5 854.5

fuel

diesel

CNG

exhaust aftertreatment

EGR

EGR + DOC + SCR

TWC

number of seats

30+1

21+1

25+1

total no. of passengers

max. allowed gross weight

18000 kg

11200 kg

18000 kg

Yearbook 2014 Laboratory for Internal Combustion Engines and e-Mobility

rESULtS Time traces Some typical measured time traces will be analyzed in this section. Engine speed traces versus time are presented in Figures 2-4. MAN bus engine speed trace is typical for city buses with conventional automatic gearboxes that can be found on MAN and Irisbus buses. Quick decreases of engine speed can be observed during gear shift. Engines are idling on the bus stops, in front of traffic lights and in traffic congestions Kutsenits bus incorporates series hydraulic hybrid powertrain and engine speed traces differ substantially from MAN and Irisbus buses.

bled due to the availability of the second hydraulic hybrid power source - high pressure hydraulic reservoirs where excess energy delivered by the engine is stored. On the other hand Figure 4 represents typical series hydraulic powertrain engine speed trace. Figure 5 represents CO2 concentration in the exhaust gases versus time of Kutsenits bus with diesel engine. At idling engine runs with approx. 4 vol % CO2 in the exhaust gases that increases during accelerations and drops to 0 during deceleration when the fuel is cut off. Similar CO2 time trace can be observed on MAN bus.

Figure 5: CO2 vs. time measured on Kutsenits bus in non-hybrid mode with power limiter off Figure 2: Engine speed vs. time measured on MAN bus

When the power limiter is turned on (Figure 3) engine speed is limited to approx. 1800min-1 compared to Figure 4 that represents engine speed trace with power limiter turned off.

On the other hand measured CO2 time trace on Irisbus (Figure 6) is typical for spark ignition engine that runs with stoichiometric homogeneous charge. Engine runs with approx. 9.5 vol % CO2. During decelerations the fuel is cut of and CO2 concentration drops towards 0. CO and THC emissions were measured as well and low concentrations were detected on all three buses. EURO V diesel engine on Kutsenits bus reached the lowest values.

Figure 3: Engine speed vs. time measured on Kutsenits bus in hybrid mode with power limiter on

Hybrid controller tends to operate the engine more frequently at lower speeds and higher load to improve its

Figure 4: Engine speed vs. time measured on Kutsenits bus in nonhybrid mode with power limiter off

Figure 6: CO2 vs. time measured on Irisbus bus

NOx concentration time trace in the exhaust gases of MAN bus is presented in Figure 7. During idling engine is producing approx. 400 vol ppm of NOx. Values increase during higher loads required for bus accelerations and reach top 1600 vol ppm. Similar top values can be observed on Irisbus bus (Figure 8). But high values last for a relatively short time just during first phase of accelerations and donâ&#x20AC;&#x2122;t influence significantly cumulative emissions. The biggest difference in pollutant emissions between analyzed busses is in particulate matter concentration as expected. Euro III diesel engine of MAN bus emits up to 500mg/L PM10 as can be seen from Figure 9.

efficiency (Figure 3). Such operation point shift is ena-

LAborAtorY For IntErnAL CoMbUStIon EnGInES AnD E-MobILItY YEARBOOK 2014

Figure 7: NOx vs. time measured on MAN bus

Maximum values are reached during transients from idling to high loads where the lowest air to fuel ratio was observed.

Cumulative results Average fuel consumption of four drives for each bus and driving mode was calculated by the carbon-balance method and is presented in table 2. Mass fuel consumption of Irisbus is the highest due to the nature of SI engine that has lower efficiency compared to CI engine. That is the case despite the fact that the lower heating value of CNG is approx. 20% higher compared to diesel fuel. Higher H/C ratio of CNG compared to diesel fuel results in similar CO2 footprint of MAN and Irisbus. Interesting results were gained on Kutsenits bus. Power limiter has a positive effect on fuel consumption as expected. But on the other hand hybrid operation of Kutsenits bus has a negative effect on fuel consumption. That confirms the fact that the effort to optimize topology of hybrid powertrain is of crucial importance. As can be seen from Table 2 CNG powered buses can contribute substantially to better quality of air in urban areas. Even if just EURO V diesel and CNG engine are compared NOx emissions are 10 times lower with CNG engine and PM10 emissions are negligible. Table 2: Fuel consumption, CO2, NOx and PM10 emissions Fuel NOx PM10 Power cons. CO2 bus Hybrid limiter g/km g/km g/km g/km

Figure 8: NOx vs. time measured on Irisbus bus

EGR, DOC and SCR used on Kutsenits diesel engine allow higher in-cylinder temperatures and pressures that are favorable to lower PM10 generation. Maximum values are approx. 10 times lower compared to MAN bus.

MAN

Kutsenits

Irisbus

397.2

1249

16.11

1.26

OFF

362.1

1141

5.20

0.05

357.6

1127

6.98

0.08

OFF

371.4

1171

5.61

0.08

OFF

343.8

1084

4.59

0.05

448.6

1224

0.51

0.00

Since the tested Irisbus bus had a low mileage further monitoring of the harmful exhaust emissions during lifetime would be sensible since three-way catalytic converter tends to degrade fast regarding THC and NOx conversion efficiency when CNG is used as a fuel [2]. ConCLUSIonS Figure 9: PM10 vs. time measured on MAN bus

Figure 10 presents PM10 concentration in exhaust gases of Irisbus bus. Homogeneous charge combustion and CNG used as fuel result in low concentrations of PM10.

Figure 10: PM10 vs. time measured on Irisbus bus

Our findings indicate that introduction of CNG powered buses as a replacement for diesel powered buses could contribute to lower particulate matter and NOx pollution in urban areas. It was showed that hybrid technology is not a guarantee for better fuel economy. Optimization of hybrid powertrain topology and fine tuning to real-world routes is of crucial importance. references [1] CIVITAS Elan, Mobilizing citizens for vital cities Introductory brochure; Ljubljana, Brno, Gent, Porto, Zagreb, July 2009. [2] S. Rodman Opresnik et al., Exhaust emissions and fuel consumption of a triple-fuel spark-ignition engine powered passenger car, Transportation Research Part D: Transport and Environment, Vol. 17, Issue 3, pp.221-227, May 2012.

YEARBOOK 2014 LAborAtorY For IntErnAL CoMbUStIon EnGInES AnD E-MobILItY

Use of innovative lignocellulosic biofuels in gas turbines Abstract → Second generation biofuels are increasing their share in the energy source portfolio, which is mainly driven by the availability of low quality residual biomass and other lignocellulosic materials as well as by their inoffensive production in comparison with food competitive feedstock. Here, fuels with low carbon footprint, produced by liquefaction in acidified polyhydroxy alcohols are tested, containing different shares of lignocellulosic biomass- from 0% to 33%. Some fuel types exhibit artificially elevated pH value to analyze the impact of product reactivity on combustion performance. To analyze the impact of fuel properties on the combustion performance and exhaust emissions, multiple parameters are varied in an experimental study conducted on a laboratory scale gas turbine. Viscosity is reduced to acceptable levels, fuels by preheating the fuel to a temperature between 80 and 110°C. During the experiments, turbine inlet

temperature is varied between 750°C and 850 °C, airflow between 0,17 kg/s and 0,20 kg/s. Combustion chamber inlet temperature is kept between 420°C and 480°C. To obtain stable combustion of different fuel types, several upgrades and analyses of fuel injection nozzle were performed and constraints for stable operation in terms of operation parameters were determined. CO emissions between 230 and 1500 ppm were obtained with THC emissions being reasonably low and NOx emissions increasing with turbine inlet temperature. NOx emissions were also closely dependent on elemental composition of the fuel. Although pollutant emissions, especially those of CO, increase with increasing biomass content in the fuel, experimental results indicate that analyzed biofuels might be successfully utilized in professional gas turbines.

Introduction Tine Seljak Laboratory Laboratory for Internal Combustion Engines E-mail tine.seljak@fs.uni-lj.si Room 305 Phone +386-1-4771-305 Status PhD student (started: September 2010, to be completed: April 2015) Research area Combustion / Alternative Fuels Mentor Assoc.Prof.Dr. Tomaž Katrašnik, Co-Mentor: Ass.Prof.Dr. Matjaž Kunaver

In the last decades there is an increasing interest in numerous biofuels originating from wood and other easy accessible local biomass with a goal to replace fossil fuels in specific applications. From technical point of view of present engines and other fuel conversion devices, ethanol and biodiesel are the most desired fuels as they closely resemble physical properties of the two most widely used petroleum distillates, gasoline and diesel fuel. Most of the biodiesel is produced from rapeseed oil and soybean and thus represents a competition to food dedicated feedstock [1]. Ethanol, which is used as a substitute for gasoline fuel is mainly produced by fermentation and subsequent distillation which serves as an additional refining step. These processes are also well known and efforts are being made to improve either the microorganisms which are responsible for fermentation or separation processes with the aim to reduce the production cost. Since most of the ethanol is produced from fermented sugar or starch it is inevitable to compete with food oriented feedstock. As a solution, primary material

Laboratory for Internal Combustion Engines and e-Mobility Yearbook 2014

for second generation biofuels is intended to be mostly lignocellulosic biomass. Following proper pretreatment, cellulose can be fermented in a relatively complex and expensive process, which also yields ethanol as a byproduct. For certain applications, such as stationary gas turbines, medium and low speed diesel cycle engines, it is also possible to utilize heavier biofuels which are usually less expensive to produce. These are characterized by high viscosity and low heating value as well as by low price and low priced feedstock availability. Viscosity of these fuels is often reduced by heating or by mixing with ethanol, although emulsions or blends with diesel fuel might also be used. Fuels, presented here were obtained through liquefaction in acidified polyhydroxy alcohols. The first proposal for this process to be used for fuel production purposes is dating in 1996, however, catalytic hydrotreatement was suggested as a second processing step to reduce viscosity and oxygen content and thereby obtain heating value similar to that of a usual petroleum fuel. Combustion performance of raw, non-hydrogenated liquefied wood

originating from European spruce was first presented in [1]. Successful initial experiments outlined extensive research area in terms of tailored recipes yielding more desirable LW properties, which were presented in [2], where neutralized version of LW with reactant ratios 1:3 (wood to glycols) was analyzed. This analysis provides an insight into phenomena while utilizing the LW with reduced reactivity, which is particularly important during fuel preheating because it improves fuel stability and widens the selection of preheat system component materials. Second, raw LW with reactant ratios 1:2 was analyzed, since this recipe features increased economic value due to lower feedstock costs. Further work was conducted in [4] where constraints considering primary air temperature were revealed. Coking of the fuel injection nozzle and deposit formation on combustion chamber walls were identified as limiting phenomena, determining the suitable primary air temperature.

Table 1: Properties of different LW types. LW type 1

LW type 2

LW type 3

C [wt.%]

item

47.60

47.52

H [wt.%]

7.98

8.00

N [wt.%]

0.19

0.34

S [wt.%]

0.89

O [wt.%]

43.34

43.26

pH value Density [kg/L] Viscosity at 100 째C

2.5

5.5

1.3 kg/L

80 cSt

171 cSt

80 cSt

FUELS AnD MEtHoDS Tested fuels Liquefaction of ground spruce wood passed through a 3 mm screen was carried out in a 200 L vented batch reactor equipped with a mixer, a heater and a 2 kW ultrasonic probe. The ratios of di-ethylene glycol:glycerol:wood were 1.5:1.5:1. Additionally, to reduce the temperature and time of liquefaction reactions, 22.5 g of p-Toluenesulfonic acid (p-TSA) was added per kg of reactant mixture. After the prepared mixture was heated to 160째C, 20 minutes retention with ultrasound excitation was sufficient to obtain a LW with less than 2% of solid residue. The properties of this LW (type 1) are presented in Table 2 beside the properties of two other types of LW. One with 1:1:1 reactant ratio (di-ethylene glycol:glycerol:wood) (LW type 2) and a partially neutralized version of 1,5:1,5:1 formulation (LW type 3). The composition of different formulations is presented in Fig. 1. The elemental compositions between LW type 1 AND LW type 2 is considered unchanged due to very similar composition of glycols and lignocellulosic biomass. The elemental composition of LW type 3 is presented to emphasize the elevated nitrogen content.

Figure 1. Composition of different tested LW types

The elemental composition and properties of different LW types is presented in Table 1. The density is higher than in conventional petroleum-derived fuels, together with high viscosity which is highly temperature-dependent as presented in Fig. 2.

Figure 2. Viscosity of different liquefied wood types

Experimental setup The presented fuel types were tested in a laboratory scale micro gas turbine.

Figure 3. Layout of experimental system

Test were carried out at different turbine inlet temperatures and different fuel temperatures for LW type 1. For LW type 2 and type 3, tests with fixed fuel temperatures were performed.

YEARBOOK 2014 LAborAtorY For IntErnAL CoMbUStIon EnGInES AnD E-MobILItY

Results Influence of fuel temperature CO and THC emissions decrease with the increased fuel preheat temperature (Fig. 3). This is mainly the consequence of two phenomena: Firstly, the lower viscosity at higher temperatures enhances formation of more intense instabilities on the liquid film, which then promotes primary spray breakup, and secondly, the lower surface tension enhances secondary droplet breakup.

Therefore, higher fuel preheat temperature results in a faster evaporation rate and in a faster and more homogeneous air-fuel mixture formation which is beneficial to combustion efficiency which is visible through lower CO and THC emissions. Liquefaction reactions pathways indicate that liquefied wood contains a large amount of cyclic hydrocarbon molecular structures with high ignition energy, which confirms previously mentioned source of THC emissions. Influence of different fuel types In Fig. 4, emissions of different types of fuel are shown. Trends are in line with physical and chemical characteristics of the fuels and can be mainly attributed to: a) Viscosity of the fuels (with LW type no. 2 having the highest, followed by LW type no. 1 and no. 3 and significantly lower viscosity of diesel fuel), b) Density of the fuels (with all LW types having higher density than diesel fuel by roughly for a factor of 1.5 higher densities than diesel fuel). c) Molecular composition of the fuels (with LW type no. 2 having higher content of cyclic hydrocarbons than LW type no. 1 and no. 3 and all LW types having higher molecular weights than diesel fuel).

Figure 3. Emissions at different fuel temperatures Figure 4. Emissions with different fuel types 90

Laboratory for Internal Combustion Engines and e-Mobility Yearbook 2014

Taking into account the upper three facts, atomization of LW is most likely notably impaired by a) and penetration depth is further increased by b). The combination of these two facts also reduces surface to mass ratio of the droplets which is most likely together with c) the reason for reduced evaporation rate of the fuel. Besides decreased volatility due to heavier molecular weights, c) also influences the combustion process due to cyclic hydrocarbon content with its high ignition resistance. The CO emissions are therefore the highest with LW type no. 2, followed by LW type no. 1 and LW type no. 3. The latter two exhibit almost similar concentrations due to similar properties under a), b) and c). With diesel fuel, very low concentrations were achieved due to more favorable physical and chemical characteristics of diesel in comparison to all LW types allowing for fast mixture formation and very low rate of mixture escape into colder parts of combustion chamber. Influence of primary air temperature In Fig. 5 and Fig. 6, CO and NOx emissions are presented for two different operating cycles – simple and regenerative. Regenerative cycle is characterized by high primary air temperatures, whereas in simple cycle mode, primary combustion air temperature equals compressor discharge temperature.

Figure 5.CO emissions for different operating cycles

Figure 6. NOx emissions for different operating cycles.

Temperature of the primary air was shown to have significant influence on the formation of exhaust emissions when firing liquefied wood. CO emissions were notably reduced at higher air temperatures. The reduction was attributed to faster evaporation rate and increased turbulence in primary zone which caused faster mixture formation.

NOx emissions were strongly dependent on the combination of both, primary air temperature and TIT. At low primary air temperatures, the influence of TIT was significantly reduced, while at high air temperatures, TIT played an important role in NOx formation. This was attributed to the competing effects of prompt and thermal mechanism lead by different equivalence ratios and high oxygen content in the LW. Generally, NOx emissions elevated with increased TIT as well as with increased primary air temperature. Conclusions The results of the upper studies suggest significant influence of turbine inlet temperature, fuel temperature and fuel type on exhaust emissions of innovative fuel and on its ability to support stable operation in MGT. The main drivers for this are physical and chemical properties of different LW types and local conditions in primary combustion zone, affected by operating parameters of the turbine. The results indicate that microturbine systems, utilizing LW type 1 as a fuel should aim for high turbine inlet temperatures and fuel temperature of 100°C. Additionally, operation in regenerative cycle mode would be desired to obtain stable operation and lowest emissions possible. List of publications [1] Seljak T, Rodman Oprešnik S, Kunaver M, Katrašnik T. Wood, liquefied in polyhydroxy alcohols as a fuel for gas turbines. Applied Energy 2012;99:40-49. [2] Seljak T, Kunaver M, Katrašnik T. Emission evaluation of different types of liquefied wood. Stroj Vestn – J Mech E. 2014;60(4): 221-31. [3] Seljak T, Rodman Oprešnik S, Katrašnik T. Microturbine combustion and emission evaluation of waste polymer-derived fuels. Energy 2014; In Press. [4] Seljak T, Rodman Oprešnik S, Kunaver M, Katrašnik T. Effects of primary air temperature on emission of a gas turbine, fired with liquefied spruce wood. Biomass Bioenergy 2014; In Press. [5] Seljak T, Rodman Oprešnik S, Kunaver M, Katrašnik, T. Combustion and emission characterization of waste-derived fuels in a gas turbine. Book of abstract. [Çeşme: s.n., 2013] [6] Seljak T, Rodman Oprešnik S, Kunaver M, Katrašnik T. Combustion analysis of liquefied biomass under gas turbine conditions. V: Austrian - Croatian - Hungarian Combustion Meeting - ACH 2012, Zagreb, 12th - 13th of April 2012. Croatian section of the Combustion Institute, 2011. [7] Seljak T, Rodman Oprešnik S, Kunaver M, Katrašnik T. Combustion performance of liquefied lignocellulosic material in a laboratory scale gas turbine. V: International Conference on Applied Energy, July 5-8, 2012, Suzhou, China. [8] Seljak T, Rodman Oprešnik S, Kunaver M, Katrašnik T. Combustion properties of fuels obtained through pyrolysis and liquefaction of fossil and bio-based polymer materials. 6th International Conference on Sustainable Energy and environmental protection, SEEP 2013, 20th - 23rd of August 2013, Maribor.

Yearbook 2014 Laboratory for Internal Combustion Engines and e-Mobility

Hybrid analytic-numerical 3D approach to modelling PEM fuel cells Abstract → Presenting a Hybrid Analytic-Numerical approach (HAN) to fuel cell modelling applied to modelling a straight parallel channel fuel cell. HAN’s core principles are based on two innovative modelling solutions: • numerical 1D pipe flow model coupled to an analytic 2D model for species transport in the plane perpendicular to the flow • model for transport of liquid water comprehended in the modelling formalism for transport of gaseous species

These two fundamental features of HAN enable: • modelling operation in conditions ranging from partially flooded to completely dry • achieving short computational times comparable to those of 1D models • achieving high accuracy of results comparable to that of 3D CFD models

Introduction Gregor Tavčar Laboratory Laboratory for Internal Combustion Engines E-mail gregor.tavcar@fs.uni-lj.si Room 310 Phone +386-1-4771-310 Status PhD researcher (PhD thesis defended: April 2014) Research area Fuel cell modelling Mentor Assoc.Prof.Dr. Tomaž Katrašnik, Co-Mentor: Assoc.Prof.Dr. Viktor Hacker

Mathematical modelling of PEMFC has been receiving great attention ever since the publication of the comprehensive PEMFC model of Springer et. al. [1] in 1991 that is still often used as a key reference. PEMFC are characterized by fully interrelated mechanisms of convection, diffusion, phase transition and electrochemical reactions that govern transports of mass, charge and heat. As systematically analysed in references [2-4] many fuel cell models can be found in literature or are available as packages in commercial software. Their various approaches to modelling the fuel cell behaviour can be categorised according to: • How physical phenomena are described i.e.: Analytically, mechanistically, semi empirically, fully empirically; • The dimensionality of the described physical phenomena: 0D, 1D, quasi 2D, 2D, quasi 3D, 3D; These models span over wide ranges of computational speeds and accuracy.

Laboratory for Internal Combustion Engines and e-Mobility Yearbook 2014

In the last years special attention has been devoted to the so called reduced dimensionality models. The reduced dimensionality models in their broader sense, i.e. any model that does not address all three dimensions equally, are a typical answer to the need for a balance between computational speed and accuracy of results. Common reduced dimensionality models are the 2D models (e.g [5], the 2D model in [6]) that model straight channel fuel cell configurations (such as depicted in Figure 1) solving governing equations in the plane that is parallel to the gas flow and perpendicular to the membrane (e.g. green plane in Figure 1). A simplification of the 2D modelling approach leads to the so called 1D+1D (e.g. [7]) models where only the most dominant physical phenomena (e.g. mass transport via bulk gas flow in channel) are addressed in the dimension that runs along direction of gas flow. Applying these 2D or 1D+1D models to fuel cell geometries such as presented in Figure 1 inherently leads to certain systematic discrepancies: namely, the area through which mass is exchanged between the gas diffusion layer (GDL) and the chan-

nel is considerably smaller than the area through which mass is exchanged between the membrane and the GDL. To compensate for this discrepancy some models are extended to employ additional correction parameters. These correction parameters are typically obtained by calibrating against results obtained with full 3D CFD models or against experimental data. The so compensated 2D models yield the pseudo 3D modelling approach (e.g. [8], [9] and Sherwood number adjusted 2D model in [6]). An alternative to this pseudo 3D approach is the so called 2D+1D approach (as found in [6]) where the physical phenomena are fully addressed in the plane perpendicular to the gas flow and, analogously to 1D+1D approach, only the dominant physical phenomena are addressed in the direction of the gas flow. The pseudo 3D and 2D+1D models aim to take into account all three dimensions whilst achieving considerably shorter computational times than 3D CFD simulations. However, with the exception of [8], they still cannot achieve computational times comparable to the 1D models due to their inherent numerical nature. Owing to its analytic nature the quasi 3D model found in [8] features computational times comparable to the 1D models. However [8] offers an approximate analytic solution for direct liquid fuel cells that is only valid when constant velocity along channels is assumed. HAN Outline Articles of Tavčar and Katrašnik [11] and [10] propose an innovative Hybrid 3D Analytic Numerical approach to modelling species transport (HAN) in a PEM fuel cell as a way to obtain results with full 3D resolution whilst achieving short computational times. The principles of HAN modelling in [11] and [10] are presented on a straight channel co-flow fuel cell and its core principle can be summarized as: A hybrid 3D model constructed by taking a 1D numerical model for the gas-flow along the channel with a superimposed 2D analytic solution for the plane perpendicular to the gas-flow. The additional computational load introduced by the calculation of the 2D analytic solution was shown to be of the same order of magnitude as the computational load of the base 1D calculation proving this hybrid 3D approach to be computationally efficient. It was shown in [11] and [10] that, owing to its partially analytic nature, the HAN approach enables resolving a 3D species transport with very high accuracy while preserving computational speed comparable to 1D models. In terms of model categorization introduced above, HAN is thus a 2D+1D model resembling that found in [5] with the difference that HAN features analytically (instead of numerically) resolved species concentration in the 2D cross-sectional plane for both GDL and channel, making it computationally considerably faster. The nature of HAN’s analytic solution is also fundamentally different from the one in [8] where the 2D analytic solution is obtained for the plane that is perpendicular to the membrane and parallel to the gas flow. The isothermal HAN models in [11] and [10] both assume co-flow configuration, no liquid water in GDLs and

channels and exclusively binary gas mixtures (i.e. either oxygen and water vapour or hydrogen and water vapour). Liquid water build-up at catalyst layers hinders the access of reactants to the reaction sites on catalyst layer causing fuel cell performance decrease. Accurate modelling of liquid water generation and transport in the GDL and channels is thus very important for the overall accuracy of a fuel cell model. Occurrence of liquid water thus requires modelling of two phase flows. The two phase flows are adequately addressed by advanced full 3D CFD simulations; however it presents a great challenge for the reduced dimensionality models. Objectives The presented isothermal HAN model is based on the mathematical formalism derived in [11] and [10] featuring: • the hybrid 3D analytic-numerical model for gaseous species transport in channels and GDLs • membrane model with integrated iterative routines for calculating nonlinear electrochemical processes. • co-flow configuration The here presented HAN model extends this formalism to enable also: • addressing the transport of liquid water in GDLs and channels, • addressing ternary gas mixtures (i.e. cathode fed with air comprising oxygen, nitrogen and water vapour) • simulating counter-flow configurations. Thus two specific operational cases are simulated with this HAN model: • Co-flow with low humidity feed gasses (no liquid water occurring in channels and GDLs). In these conditions the relative dryness of feed gasses causes de-humidification of the membrane leading to elevated proton transport resistance of the membrane. The co-flow regime best exposes the varying membrane humidification throughout the fuel cell. This case is thus useful to validate the model’s ability to accurately model the membrane at varying humidification. • Counter-flow with high humidity feed gases (liquid water occurring in both GDLs and both channels). In these conditions the membrane is fully hydrated practically throughout the membrane leading to optimal and constant proton conductivity. This case is thus useful for validating the model’s ability to model the liquid water transport in GDLs and channels and to assess the efficiency of algorithms for counter-flow computation. The accuracy of this HAN model is evaluated in two ways: • by comparing the results on the 3D distribution of physical variables obtained by HAN with the ones obtained by a professional 3D CFD tool and • by comparing a plot of the polarisation curve obtained by HAN simulation with the laboratory experimental polarisation data for the modelled fuel cell.

Yearbook 2014 Laboratory for Internal Combustion Engines and e-Mobility

This presentation thus aims at showing that the so constructed HAN model is capable of: • modelling operation in conditions ranging from partially flooded to completely dry • achieving short computational times comparable to those of 1D models • achieving high accuracy of results comparable to that of 3D CFD models Modelled fuel cell The modelled fuel cell is a straight hydrogen-air or hydrogen-oxygen type PEM fuel cell (FC) with its geometry taken from a laboratory test fuel cell operated in both coand counter-flow configurations. The fuel cell topology and geometry are illustrated in Figure 1a distinguishing the following sub-elements: the cathode feed part comprising cathode channels and cathode gas diffusion layer (GDL); a thin catalyst layer for oxygen reduction; a hydrated proton exchange membrane; a thin catalyst layer for hydrogen oxidation; anode feed part comprising anode channels and anode gas diffusion layer (GDL). Due to the symmetric geometry of the modelled fuel cell and the assumed isothermal operating conditions the modelling of the whole fuel cell is reduced onto modelling the fuel cell’s representative unit which is a half of one rib, as depicted in Figure 1b.

Figure 1: Polarisation curve. Pink are laboratory measurement results of voltages obtained at current densities: 10000, 8000, 6000, 4000, 2000, 1200 and 400 A/m2. Brown are HAN results of current densities obtained at voltages that match the experimentally measured voltages

Figure 1: Cathode Galvani potential distribution at 0.75V of operational voltage. Green plot are CFD results and brown HAN results.

Figure 1: Fuel cell geometry schematically broken-down onto elementary units for HAN computation. The blue regions represent the membrane and the spotted translucent the GDLs; (a) A fiverib parallel channel co- or counter-flow fuel cell geometry. The two symmetry planes that apply to all ribs are shown delimiting the right half of one rib and so defining the representative unit. (b) Representative unit with one slice (red) indicated. (c) Slice as a sliced-out section of the representative unit. A slice has sufficiently small depth to be treated as a 2D object. (d) Slice split onto computational domains.

Results In this section four specific examples of results are presented: In Figure 2 polarisation results obtained by HAN simulation are evaluated by plotting them against the experimental polarisation measurements; in Fugure 3 HAN vs. CFD results of cathode potential are presented for the dry co-flow configuration and in Fugure 4 HAN vs. CFD results of the GDL surface water concentration are presented for the humid counter-flow configuration; in Table 1 the computational times are reported to evaluate HAN’s computational.

Figure 1: Surface concentration of liquid water on cathode GDL surface. The measuring unit of μm is the water column equivalent of the surface concentration in ml/m2. Table 1: Calculation times (on a desktop computer) regime /model dry co-flow dry counter-flow humid counter-flow CFD(coarser)

No. of iterations

CPU time [s]

5 to 6

0.85*

6.0*

150

75*

5000

2442

* times obtained with HAN programmed in Mathematica – considerably shorter times are expected for HAN programmed in C or Fortran

Laboratory for Internal Combustion Engines and e-Mobility Yearbook 2014

Conclusions The comparative HAN versus CFD 3D plots have demonstrated HAN’s ability to accurately simulate the spatial distribution of key physical quantities in a fuel cell. The fact that HAN results show a good agreement with both the experimental and the CFD results validates the HAN’s hybrid 2D analytic + 1D numerical approach to modelling the fuel cell governing mechanisms: the diffusive and convective transport of gaseous species; the capillary flow of liquid water in the GDL; the transport of water clinging to the GDL surface dragged by the channel low; and the species production and consumption at the catalysts. HAN models both the two phase and the single phase species transport with a single diffusion equation thus this validation confirms the HAN’s capability to efficiently model both dry and saturated conditions. Very short computational times have been achieved for the co-flow cases with no liquid water on the GDL surface and short computational times for the counter flow cases with liquid water on the GDL surface with good prospects for further improvement. Overall, HAN proves to be very accurate and computationally efficient and as such a very promising standalone fuel cell model for system level simulations. Further challenges in developing HAN for system level simulations remain achieving higher computational efficiency of the cases of counter-flow with liquid water on the GDL surfaces and extending the model to treat also heat transfer and the more complex serpentine fuel cell geometries..

tional models of polymer electrolyte membrane fuel cell stacks. J. Computational Physics. 2, 2007, Vol. 223, 797-821. [10] Tavčar, Gregor and Katrašnik, Tomaž. An Innovative Hybrid 3D Analytic-Numerical Approach for System Level Modelling of PEM Fuel Cells. Energies. 10, 2013, Vol. 6, 5426-5485. [11] Tavčar, Gregor and Katrašnik, Tomaž. A computationally efficient hybrid 3D analytic-numerical approach for modelling species transport in a proton exchange membrane fuel cell. J. Power Sources. 2013, Vol. 236, 321-340. List of publications [1] Tavčar, Gregor and Katrašnik, Tomaž. An Innovative Hybrid 3D Analytic-Numerical Approach for System Level Modelling of PEM Fuel Cells. Energies. 10, 2013, Vol. 6, 5426-5485. [2] Tavčar, Gregor and Katrašnik, Tomaž. A computationally efficient hybrid 3D analytic-numerical approach for modelling species transport in a proton exchange membrane fuel cell. J. Power Sources. 2013, Vol. 236, 321-340. [3] Tavčar, Gregor and Katrašnik, Tomaž. An innovative hybrid 3D analytic-numerical model for air breathing parallel channel counter-flow PEM fuel cells. Acta Chim Slov. 2014, Vol. 61(2), 284-301.

References [1] Springer, T.E., Zawodzinski, T.A. and Gottesfeld, S. Polymer electrolyte fuel cell model. J. Electrochem. Soc. 8, 1991, Vol. 138, 2334-2342. [2] Haraldsson, Kristina and Wipke, Keith. Evaluating PEM fuel cell system models. J. Power Sources. 1-2, 2003, Vol. 126, 88-97. [3] Cheddie, Denver and Munroe, Norman. Review and comparison of approaches to proton exchange membrane fuel cell modeling. J. Power Sources. 1-2, 2005, Vol. 147, 72-84. [4] Gurau, Vladimir and Mann, J. Adin Jr. A Critical Overview of Computational Fluid Dynamics Multiphase Models for Proton Exchange Membrane Fuel Cells. SIAM J. Appl. Math. 2, 2009, Vol. 70, 410-454. [5] Ee, Sher Lin and Birgersson, Erik. Two-Dimensional Approximate Analytical Solutions for the Direct Liquid Fuel Cell. J. Electrochem. Soc. 10, 2011, Vol. 158, 81224-81234. [6] Kim, Gwang-Soo, et al. Reduced-dimensional models for straight-channel proton exchange membrane fuel cells. J. Power Sources. 10, 2010, Vol. 195, 3240-3249. [7] Esmaili, Q., Ranjbar, A. A. and Abdollahzadeh, M. Numerical Simulation of a Direct Methanol Fuel Cell through a 1D+1D Approach. International J. Green Energy. 2, 2013, Vol. 10, 190-204. [8] Ling, Chun Yu, Ee, Sher Lin and Birgersson, Erik. Three-dimensional approximate analytical solutions for direct liquid fuel cells. Electrochimica Acta. 30, 2013, Vol. 109, 305-315. [9] Chang, Paul, et al. Reduced dimensional computa-

Yearbook 2014 Laboratory for Internal Combustion Engines and e-Mobility

Performance of a 6-cylinder turbo charged diesel engine running on tire pyrolysis oil Abstract → The objective of this work was to investigate the performance of a 6-cylinder, compression ignition, turbocharged, 6.9L heavy-duty engine fueled with tire pyrolysis oil (TPO) produced from waste tires. The thermodynamic parameters and engine performance of the engine fuelled with TPO were benchmarked against results of the engine fuelled with commercial diesel fuel (D2). Experiments were conducted in two operating modes, with and without intercooler, at two different engine speeds and at various loads. Impact of thermodynamic and engine performance parameters on combustion process were systematically analyzed in terms of cause and

Rok Vihar Laboratory Laboratory for Internal Combustion Engines E-mail rok.vihar@fs.uni-lj.si Room 311 Phone +386-1-4771-311 Status PhD student (started: December 2012, to be completed: September 2015) Research area Internal combustion engines Mentor Assoc.Prof.Dr. Tomaž Katrašnik

effect phenomena through mechanisms initiated by the fuel properties. The original contribution of this analysis arises from holistic assessment of combustion phenomena and engine performance characterization of a modern turbocharged CI engine fuelled by the pure TPO. Results indicate that TPO can be efficiently used in turbocharged non-intercooled CI engines at high loads, which opens its use in power generation.

Introduction It is generally accepted that internal combustion engines (ICE) will stay in commercial use as a main power source for another 20 years. The reduction of fossil oil reserves is forcing the society to look for alternative sources of energy to power ICE. Additionally, there is a growing problem of waste disposal and thus waste-to-fuel technologies offers a very promising solution approach for both issues. It is estimated that worldwide over one billion waste tires are generated annually, which is considered as a serious pollution problem in terms of waste disposal. Since rubber from tires has a high calorific value (35-40 MJ/kg) as well as considerable amount of carbon black, vehicle tires stand out as being a good feedstock for fuel production. Since most ICEs rely on liquid or gaseous fuels and most waste feedstock is general at least partially in solid state, several conversion processes have been developed for transforming solid wastes into fuels with acceptable chemical and physical properties for the use in the ICEs. These processes mainly rely on thermochemical conversion of input material among which different types

Laboratory for Internal Combustion Engines and e-Mobility Yearbook 2014

of pyrolysis are the most widely used procedures. For conversion of waste tires, the most suitable subtype is vacuum pyrolysis. Depending on process conditions such as residence time, pressure, tire particle size and composition the product come in different ratios. However, process conditions can be optimized to favor either of the products. Compression ignition (CI) engines stand out as being potential power plants for the use of TPO without requiring major engine modifications since distillation properties of TPO fuel are very similar to conventional diesel fuel as presented in Section 2. However some analyses indicate that because of its low cetane number, TPO must be blended with diesel fuel or complemented by a cetane improver, for example diethyl ether, for application in diesel engines. Other approaches addressing this issues, suggest increasing temperature of the air in the intake system. It was also reported that “critical inlet air temperature” at which the TPO could be used in CI engines is 145°C. This temperature was achieved through external heater used for preheating intake air, which lowers total energy efficiency and applicability of the system. It was also

observed that many of the analyses were performed using stationary single cylinder CI engines. Thus only limited number of studies are performed in commercial multicylinder turbocharged (TC) engines but none of them with pure TPO fuel. The objective of this research was to perform an in depth analysis of differences in thermodynamic parameters and engine performance of the same engine operating on TPO and diesel fuel. Original contribution of this study arises from the demonstration of a stable combustion of a 100% TPO in a modern TC multi-cylinder engine operated without intercooler at high loads. Fuel properties The TPO used in this study was produced by vacuum pyrolysis method. Waste tires used in the process were cut into pieces with a mean size of 100 x 100 mm and steel wires and fabrics were removed. The process was carried out at temperature between 600 and 700°C and the retention time in the reactor was 60 min. Fuel tested here consists of TPO fraction between 190 and 350°C. The diesel fuel used in benchmark analyses complies with the specifications of the SIST EN-590 standard. Key properties of the TPO and the D2 are shown in the Table 1. As density of the TPO is higher than density of the D2 and its lower heating value (LHV) is lower, the injection quantity of the TPO must be slightly reduced to achieve the same cyclic energy delivery due to higher volumetric energy density of TPO. Albeit reduced C and H content, the LHV is still relatively high in TPO due to sulfur content which compensates the inert components in the fuel (namely N and O) with its contribution to heating value. The elemental composition of the TPO exhibits also slightly lower H/C ratio than D2, which is the consequence of its molecular composition with higher number of double bonds and polycyclic aromatic hydrocarbons (PAH) than diesel fuel. Table 1. Comparison of TPO with diesel (D2) Property

TPO

Density [kg/L]

0.92

0.83

LHV on mass basis [MJ/kg]

40.6

42.7

LHV on volume basis [MJ/L]

37.4

35.3

~13.8

14.7

39.3

26.0

Stoichiometric ratio Aromatic content [% m/m]

3.22 (20°C)

2.54 (40°C)

C [% m/m]

Viscosity [cSt]

83.45-85.60

87.0

H [% m/m]

9.59-11.73

13.0

N [% m/m]

0.40-1.05

S [% m/m]

0.96

<0.001

O [% m/m]

0.10-3.96

28,6

53,2

Calculated cetane index1 1 Regarding SIST EN ISO 4264

TPO exhibits poor ignition properties, which is reflected in its relatively low Cetane Number (CN). The CN number is an important factor for determining the quality of diesel or diesel like fuel as low CN indicates the fuel requires

higher activation energy and thus higher auto-ignition temperature. It seems generally accepted that the CN of TPO is certainly below 30 whilst regarding SIST EN-590 standard conventional diesel fuels have a CN at least 51. This was confirmed by the calculated value of cetane intex (Table 1) which shows similar difference. One of the measures to achieve diesel like combustion with lower CN fuels is thus to ensure sufficiently high end-of-compression temperature. Experimental setup The experimental work was carried out on a 6-cylinder, 4 stroke, turbocharged, 6.87 liter MAN diesel engine (model D 0826 LOH 15). The main characteristics of the engine are given in Table 2. The engine was coupled with a Zöllner B-350AC eddy-current dynamometer controlled by Kristel, Seibt & Co control system KS ADAC. A Kistler CAM UNIT Type 2613B shaft encoder provided an external trigger and an external clock (0.1-6 deg CA) for data acquisition system. In-cylinder pressure was measured with calibrated piezo-electric pressure transducer AVL GH12D in combination with charge amplifier AVL MICROIFEM, connected to 16 bit, 4 channel National Instruments dataacquisition system with maximum sampling frequency 1 MS/s/ch. Top dead center (TDC) was determined by capacitive sensor COM Type 2653. An AVL 730 gravimetric balance was employed to measure fuel consumption. Intake air flow was measured with Meriam laminar flow meter 50MC2-6F. Additionally, the engine was fitted with instrumentation for measurement and monitoring temperature and pressure (intake air, exhaust gases, lube oil, coolant etc.). Table 2. Engine specifications Engine

MAN D 0826 LOH 15

Cylinders

6, inline

Displacement

6870 cm3

Bore × stroke

108 mm × 125 mm

Compression ratio

18:1

Fuel injection system

Mechanically controlled direct injection

Maximum power

162 kW @ 2400 rpm

Maximum torque

825 Nm @ 1400-1700 rpm

Air intake

Turbocharged

Cooling system

Water cooled

Two separate fuel tanks were used in the system: one for the base D2 fuel and another for the TPO fuel. No modification was done to engine fuel system. Experiments were performed in thermally stabilized steady-state points at two different engine speeds (1500 rpm and 2400 rpm) in two different operational modes with and without intercooler. While LHV and density of tested fuel are different than those from reference fuel, cyclic energy delivery (CED) was selected as the second independent engine operating parameter. Tests were performed for two CED values: 8.5 and 10.5 kJ/cyc in the mode without intercooler denoted by “woIC” and in the mode with intercooler denoted by “wIC”.

Yearbook 2014 Laboratory for Internal Combustion Engines and e-Mobility

Fig. 2. In-cylinder pressure a.), Pressure derivative b.), ROHR c.) and in-cylinder temperature d.); at 1500 rpm and 8,5 kJ/ cyc CED Fig. 3. In-cylinder pressure a.), Pressure derivative b.), ROHR c.) and in-cylinder temperature d.); at 2400 rpm and 10,5 kJ/ cyc CED.

Combustion process analysis Thermodynamic parameters were measured continuously while in-cylinder pressure was recorded over 100 successive cycles at sampling resolution of 0.2Â° CA. Very good repeatability of the individual cycles was proven by overlapping of pressure oscillations that correspond to the lowest excitation frequency of the gas in the combustion chamber over the consecutive cycles. These pressure oscillations that are superimposed on the pressure trace appear after the start of combustion (SOC) as a consequence of the partial auto-ignition of the fuel [1]. Due to these pressure oscillations local pressure at the pressure sensor cannot be considered as a representative in-cylinder pressure of the entire combustion chamber. Therefore, two further steps were done to generate the representative in-cylinder pressure trace. First, a representative in-cylinder pressure trace was generated by averaging 100 cycles of the individual operating point as averaging significantly eliminates point-to-point variations due to signal noise. In spite this fact, it is further necessary to eliminate pressure oscillations in the combustion chamber. This was done applying low pass FIR filter [2] to the averaged pressure trace. Such pressure trace was then 98

further processed by the AVL Burn tool [2] to calculate pressure derivative, rate of heat release (ROHR) and incylinder temperature. Detailed equations for calculating ROHR are presented in [3]. Results and discussion In Fig. 2 and Fig. 3 in-cylinder pressure, pressure derivate, ROHR and average in-cylinder temperature are shown respectively for 1500 rpm and CED 8.5 kJ/cyc (Fig. 2) and 2400 rpm and CED 10.5 kJ/cyc (Fig. 4). These two engine operating points were selected from analyses as they reveal pronounced influence of engine operating conditions on combustion parameters. Selected points show distinctive differences in combustion parameters: â&#x20AC;˘ at 2400 rpm pressure traces of TPO were in both modes (wIC and woIC) similar to those of D2. â&#x20AC;˘ at 1500 rpm (wIC and woIC) large differences could be observed in pressure traces (e.g. peak pressure, high pressure oscillations, high pressure derivate) when comparing TPO to D2. When analyzing the results in Table. 3, one of the most important

Laboratory for Internal Combustion Engines and e-Mobility Yearbook 2014

influencing parameters on traces, presented in Fig.3 is the intake manifold temperature. This temperature is determined by the following: a) Operational mode: In wIC operational mode, the compressor discharge air temperature is reduced towards ambient temperature, whereas in woIC, the intake manifold temperature almost equals compressor discharge temperature. b) CED: CED influences the intake manifold temperature through different turbocharger spin speed. In low engine power operating regions (i.e. low CED), low turbocharger spin speed results in low boost pressures and thus lower temperatures of the compressed air. Contrary, high temperature compressed air is obtained in the area of high CED. c) Engine speed: Increased engine speed also influences the turbocharger spin speed due to increased mass flow over turbine. Thus, higher engine speed results in higher compressor discharge temperatures and higher intake manifold temperatures.

limited operating range is 5.5 bar/° CA. Despite the fact that in some advanced combustion concepts (like HCCI) much higher values of MRPR were reached it was assumed in this study that MRPR values higher than 5.5 bar/° CA could result in undesirably increased ageing and wear. It should be considered that correlation between MRPR and ROHR is valid only around TDC where volume derivative is close to zero [1] and thus high values of MRPR can be used as a reliable indicator for intense combustion only in that region. As a result of intensive ROHR of the premixed combustion phase, temperature and pressure of the TPO in both wIC and woIC mode surpass those of the D2 fuel during the premixed combustion phase. Consequently, higher in-cylinder temperature and pressure further promote high reaction rates and thus intensive ROHR leading to shorter combustion durations of TPO in the woIC mode and to the shortest combustion duration of the TPO in the wIC mode.

Moreover, it is discernable from Fig. 2a and 3d that for both cases, i.e. wIC and woIC, compression pressures and compression temperatures are fuel dependent and differ for TPO and D2. This fuel sensitivity originates from different ROHRs (Fig. 2c) that in turn result in different exhaust gas enthalpies and thus in different powers produced by the turbine and consequently in different TC speeds leading to different pressures and temperatures in the intake manifold. Fig. 2c shows that for D2 fuel temperatures at the end of compression (Fig. 2d) are sufficiently high in wIC and woIC mode that chemical reaction rates do not significantly prolong the ignition delay period. Unlike, this effect is very pronounced for TPO. TPO features longer ignition delay period already in the woIC mode being the consequence of higher aromatic content and thus lower CN of the fuel. Increased ignition delay of the TPO is even more pronounced in the wIC mode as further reduction in end-of-compression temperatures (Fig. 2d) significantly prolongs chemical reaction rate related part of the ignition delay period. Prolongation of ignition delay period is proportional to the amount of fuel-air mixture being prepared during this period and thus to the intensity of the ROHR of the premixed combustion phase (Fig. 2c). Intensive ROHR of the premixed combustion phase result in large value of pressure derivative (Fig. 2b) and thus high gradient of the average in-cylinder temperature (Fig. 2d). High rate of pressure rise (Fig. 2b) can result in objectionable diesel knock. Observing operational point at 1500 rpm and 8.5 CED (denoted as 1500_8.5) in (Fig. 2), maximum rate of pressure rise (MRPR) in wIC mode is much higher while running on TPO (8.29 bar/° CA) compared to running on D2 (2.24 bar/° CA). In woIC mode the difference is even larger (12.36 bar/° CA with TPO fuel). Although the highest allowed MRPR depends on the engine type and operating conditions, it must be controlled especially at higher loads to avoid unacceptable noise levels and structural damage. It was suggested by multiple authors that threshold value of MRPR for knock

Tire pyrolysis oil (TPO) was used to investigate potential possibilities of using it as an alternative fuel in commercial turbocharged diesel engines. This analysis shows that it is possible to use 100% TPO in a modern TC multi-cylinder engine despite its low cetane number while considering a narrowed operating range of the engine. It was namely shown that stable diesel like combustion was achieved without any additional energy supply and any additive added to the fuel at high engine speed and high load without the use of the intercooler. High engine speeds and loads that are required to achieve sufficiently high charge temperature at start of combustion are particularly suitable for stationary power generation, which opens possibilities for energy and cost effective exploitation of the waste derived fuel. If the original engine is equipped with the intercooler, operation without the intercooler increases thermal loadings of the engine components and reduces mass of fresh change. If such engines do not feature additional margins towards increased component temperatures and towards reduces air-fuel ratios omission of the intercooler results in a certain reduction in cyclic energy delivery and thus engine torque and power. However, as reduction in power output is not very pronounced this option is still attractive as no further changes are needed to enable direct utilization of the waste derived fuel.

Conclusions

References [4] Katrasnik T, Trenc F, Rodman Opresnik S. A New Criterion to Determine the Start of Combustion in Diesel Engines. Journal of Enginering for Gas Turbines and Power 2006; 128, 4: 928-933.Cras at elit sed urna aliquam volutpat ut ut nulla. [5] Users Guide AVL BOOST VERSION 2001.1. 12 ed.; 2011. [6] Prah I, Katrasnik T. Application of Optimization Techniques to Determine Parameters of the Vibe Combustion Model. Strojniški vestnik - Journal of Mechanical Engineering 2009; 55, 11: 715-726.

Yearbook 2014 Laboratory for Internal Combustion Engines and e-Mobility

Optimization of the combustion process with innovative fuels and closed-loop combustion control Abstract → Strict European emission standards and rising fuel prices demand engines with cleaner combustion and increased fuel efficiency. Therefore an experimental dual-fuel engine that allows for reactivity controlled compression ignition or homogeneous charge compression ignition is being developed on the basis of the diesel engine from a mid-sized passenger car. To achieve sufficient results, the control of fuel injection has to be capable to adapt the fuel injection depending on the conditions in the combustion

chamber. Up to now, closed-loop combustion control system that shifts the start of fuel injection depending on the position of chosen mass fuel burned percentage has been developed. Next stage of the research is going to be adaptation of the diesel engine for the coincident usage of various liquid and gaseous fuels in order to achieve higher efficiency and lower oxides of nitrogen (NOx) and particulate matter (PM) emissions.

Introduction Urban Žvar Baškovič Laboratory Laboratory for Internal Combustion Engines E-mail urban.zvar-baskovic@ fs.uni-lj.si Room 310 Phone +386-1-4771-310 Status PhD student (started: December 2013, to be completed: September 2016) Research area Internal combustion engines Mentor Assoc.Prof.Dr. Tomaž Katrašnik

100

The growing concern for the environment in the European Union and the measures proposed by the European Commission strive for reducing of gaseous emissions and particulate matter (PM) emissions not only in transport but also in stationary internal combustion engines. Development is focused in several directions, among which the key research areas are optimization of combustion process and adjustment of engines for the use of alternative fuels that are locally available and do not coincide with the alimentary raw materials. The improvement of the combustion process in a piston engine is to a large extent related to the technology of a closed loop combustion control (CLCC), which enables the control of working fluid exchange process and control of parameters that affect the combustion process, on the basis of the parameters, calculated from the pressure trace from the previous cycle. In the engines that operate with non-homogenous charge compression ignition, is the combustion process usually controlled on the basis of parameters, calculated from the previous cycle pressure trace, however the combustion process of the homogenous charge compression ignition engines (HCCI) need to be controlled on the basis of the data,

Laboratory for Internal Combustion Engines and e-Mobility Yearbook 2014

measured in the same cycle. Reliable closed loop combustion control system is a prerequisite for the successful development of the Kinetically Controlled Compression Ignition Combustion (KCCIC) engines. KCCIC technologies can be divided into several subsets, among which are also homogenous charge compression ignition (HCCI), premixed charge compression ignition (PCCI) and reactivity charge compression ignition (RCCI), which combines simultaneous use of two or more fuels. Listed technologies make it possible to achieve lower emissions of NOx and PM, as a homogenous or partially homogenized charge is subjected to the local course of a chemical kinetics, which is defined by the temperature, concentration and pressure distribution. Its dependence on the various parameters is the reason for the requirement of an extremely precise control of the combustion process, since small changes in the temperature, concentration or pressure distribution lead to a very different rate of chemical reactions. Engine speed and load variations are the main reason for pressure, charge composition, charge temperature and combustion chamber walls temperature fluctuations. Therefore successful management of the kinetically controlled combustion process in real conditions is largely dependent on the ability of the combustion process

control, based on the data measured in the same cycle. It is due to the high impact of chemical kinetics on the KCCIC combustion process that properties of fuel are of significant importance. Great potential for development in the field of optimization of the combustion process arise from the simultaneous usage of two different fuels, which have different chemical and physical properties. The potential is even bigger in the case the used fuels are acquired with the lowest possible carbon footprint. Research The research is focused to a development of the KCCIC engine, able to run on various fuels, which are either widely accessible or are being produced locally. Up to now, CLCC algorithm was developed and implemented onto a reconfigurable embedded control system. It was used to control the combustion of the state of the art compression ignition engine, initially placed in a mid-sized passenger car. Start of the fuel injection for the next cycle is controlled depending on the set burn rate position, which can vary from 1% to 99%. In order to achieve reliable injection control, the pressure trace, being the main computational parameter, must be filtered and referenced to an absolute pressure. Filtering of the pressure signal takes place both in the application specific integrated circuit, placed in the pressure sensor, where the FIR digital filer is used, as well as in the pre-processing part of the controller, where running average filter can be used for additional pressure signal smoothing. The pressure data is sampled in the crank angle domain at the rate of up to 0,1°CA, which demands enough processing power. Therefore is the controller implemented onto FPGA integrated circuit, which runs at 40MHz and can calculate 140°CA data in just 2ms. After pressure signal filtering, the rate of heat release is calculated from the first law of thermodynamic, given by: .

(1)

The calculation of the heat loss to the combustion chamber walls is based on the Hohenberg heat transfer correlation and for the rest of the calculations, the real properties of thermodynamic parameters cp and R are used.

Figure 1: Closed loop combustion control topology

From the curve, acquired by integrating Equation (1), position of the manualy set mass fraction burned percentage is determined and the fuel injection is shifted in order

to achieve the desired crank angle position of the mass fraction burned percentage in the next cycle. The topology of the closed loop combustion control system is shown in the Figure 1. Pressure and angle position signals, represented by the red lines, are the input signals of the FPGA controller, which is onward connected to the injector driver module. Calculated thermodynamic parameters are transferred to the PC via Ethernet connection and are being presented on the display. Constants and other parameters, necessary for the calculation, are entered into PC graphic user interface window and sent to the FPGA integrated circuit. Conclusions Development of an experimental dual-fuel engine that allows for reactivity controlled compression ignition or homogenous charge compression ignition is closely related to the development of a closed loop combustion control system, able to control the fuel injection on the basis of the in-cylinder pressure signal. The heart of the CLCC system is FPGA integrated circuit which is capable of performing the calculation of rate of heat release, its integral and other thermodynamic parameters for 140°CA in just 2ms. Depending on the calculated parameters, it calculates the optimal crank angle position for the start of fuel injection in order to meet the pre-set position of mass fraction burned. Next phase of the research is the adaptation of the diesel engine for the simultaneous usage of various fuels in the interest of achieving lower emissions and higher efficiency. References [1] PRAH, Ivo, KATRAŠNIK, Tomaž Application of Optimization Techniques to Determine Parameters of the Vibe Combustion Model, 2009, Journal of Mechanical Engineering 55 11, 715-726 [2] SELJAK, Tine, KUNAVER, Matjaž, KATRAŠNIK, Tomaž. Emission evaluation of different types of liquefied wood. Journal of Mechanical Engineering, ISSN 0039-2480, Apr. 2014, vol. 60, no. 4, str. 221-231, ilustr., doi: 10.5545/sv-jme.2013.1242. [3] RODMAN OPREŠNIK, Samuel, SELJAK, Tine, BIZJAN, Frančišek, KATRAŠNIK, Tomaž. Exhaust emissions and fuel consumption of a triple-fuel spark-ignition engine powered passenger car. Transportation research. Part D, Transport and environment, ISSN 1361-9209. [Print ed.], 2012, vol. 17, iss. 3, str. 221-227. doi: 10.1016/j. trd.2011.08.002. [4] TAVČAR, Gregor, BIZJAN, Frančišek, KATRAŠNIK, Tomaž. Methods for improving transient response of diesel engines – influences of different electrically assisted turbocharging topologies. Proceedings of the Institution of Mechanical Engineers. Part D, Journal of automobile engineering, ISSN 0954-4070, Sep. 2011, vol. 225, iss. 9, str. 1167-1-1167-16. [5] KATRAŠNIK, Tomaž, TRENC, Ferdinand. Innovative approach to air management strategy for turbocharged diesel aircraft engines. Proceedings of the Institution of Mechanical Engineers. Part G, Journal of aerospace engineering, ISSN 0954-4100, 2012, vol. 226, no. 8, str. 966-979, doi: 10.1177/0954410011416750.

Yearbook 2014 Laboratory for Internal Combustion Engines and e-Mobility

101

Ime Priimek

Students’ Expert Activities The Department of Energy Engineering has a sense of responsibility to additionally educate students as part of their courses and lectures. Therefore, additional expert activities are organized periodically, every year, in the form of additional expert laboratory work, workshops, presentations, expert excursions and sightseeing tours to industrial facilities and partners.

Every year several expert excursions are organized to offer students an additional insight into expert working environments in which they will probably work in the future. Usually, students are invited to visit our biggest coal-fired power plant, Thermal power plant Šoštanj; the CHP plant Ljubljana, which also uses biomass as a fuel; the gas power plant at Brestanica, which can also operate as a combined gas-steam power plant; the nuclear power plant at Krško; several hydro power plants on the rivers Drava, Sava and Soča; as well as some small, gas-fired cogeneration units. The goal of all the additional activities is to give students the possibility to see their future work environments, get a feeling for what expert knowledge is needed in these companies and to introduce leading experts to students. Within that cooperation with leading energy companies in Slovenia a very important link to the industrial environment is set up which helps future engineers that finish their studies within the Department of Energy Engineering to be better candidates for employment. With several excursions we try to cover the whole spectrum of energy-production facilities: from big coalfired, cogeneration, combined gas-stem, gas, nuclear, hydro and other power plants. Students get invaluable experience and information regarding the operation, maintenance, investments, research and development in those facilities and processes. At the end of several excursions all the theory, cases and tasks that were presented to students within the study make sense and have both an engineering background and logic.

102

Visit to Nuclear power plant Krško

Exclusive visit to the construction site of Thermal power plant Šoštanj, Unit 6 In 2014 the students had a once-in-a-lifetime opportunity, i.e., to visit the construction site of a new 600-MWe power plant that is currently in the operational testing phase. That was possible because of the long-term cooperation between the Laboratory for Heat and Power of the Department of Energy Engineering and Thermal power plant Šoštanj. All the requirements regarding safety had to be considered and special instructions were presented to everyone before the sightseeing. It was a special experience, because some parts of the power plant were still open for inspection and students were able to see the “intestines” of a big coal-fired power plant: the steam turbine, some heat exchangers, boiler fire place, etc.

Workshops Some students are also the part of department’s team that organize additional laboratory exercises, expert presentation activities of the department and help to present the Faculty of Mechanical Engineering to future students and other members of the technical public.

2x 1100 MW units in Neurath; the same technology as unit 6 in Power Station Šoštanj

Workshop in the Technical Museum of Slovenia in Bistra

Special expert excursion at Siemens in Mülheim and Daimler Benz in Stutgart In 2014, at the beginning of June, in cooperation with Siemens and Daimler-Benz, a special 3-day excursion was organized for students in the final year of their master’s studies. The three days and more than 2500 km were more than worth it as students saw unique manufacturing processes of steam turbines in Siemens with the excellence seen only in that kind of company. Several special pieces of information were passed on from senior managers and researchers with decades of experiences in turbine design. The road took us through the land of the river Ruhr, which is known as Germany’s industrial centre, with 5.5 million habitants. As engineers we took a chance and stopped at one of the biggest power plant in that area, Neurat Power Station, that serves mainly as a base-load power station. It consists of seven units (3 x 300 MW, 2 x 600 MW, and 2 x 1,100 MW nominally). The students were astonished by the look and the size of a power station that has an installed power that is more than 1.4 times bigger than the whole installed power in Slovenia.

At Daimler-Benz we had the opportunity to have a special lecture for the technical public. After that we had 2 hours for the sightseeing of combustion-engine production lines for Mercedes-Benz diesel engines. After that we visited the Mercedes museum, where more than 100 years of history is presented. The name first appeared in 1926 under Daimler-Benz, but traces its origins to Daimler-MotorenGesellschaft’s 1901 Mercedes and to Karl Benz’s 1886 Benz Patent Motorwagen, which is widely regarded as the first gasoline-powered automobile. Mercedes-Benz’s slogan is “Das Beste oder nichts” (English: “The best or nothing”). And this message we want to present to our future engineers in the Department of Energy Engineering.

In front of the Mercedes-Benz Museum in Stuttgart – The best or nothing

Siemens main building in Mülheim 103

Matija Tuma

Short step into the history The Department of Energy Engineering is one of the oldest and biggest departments of the Faculty of Mechanical Engineering of Ljubljana. It has a rich and long lasting tradition. Its existence started in 1946 when “the old building” was built. In 1960 the Department of Power Engines and Thermal Devices was constituted. Later it was renamed to Department of Energy Engineering.

Constitution of the University of Ljubljana University of Ljubljana was constituted by the law issued on 23 July 1919. The law consisted of only four paragraphs. The last paragraph was saying: “This law becomes official after it is undersigned by the King and becomes mandatory after it is declared”. This is an example how uncomplicated the legislation was in the Kingdom of Serbs, Croats and Slovenes. One of the main initiators for the immediate constitution of the national university was Dr. Mihajlo Rostohar an assistant professor at the University of Prague. Nevertheless, he never was teaching at the University of Ljubljana. Dr. Danilo Marjoram was the first president of the University’s Commission. Dr. Josip Plemelj was the first rector and Dr. Richard Zupančič, who was also the corresponding member of the Academy of Sciences in Toulouse, was the first vice-rector. Both of them were mathematicians. The first dean of the Faculty of Engineering became Dr. Karol Hinterlechner, a geologist and an external member of the Czechoslovak Academy of Sciences in Prague. All of them were known as the top experts even outside the borders of the Austro-Hungarian Monarchy. Perhaps they were ranking even higher in the Europe than today’s university professors. They have returned to their native land, the newly formed Kingdom of Serbs, Croats and Slovenes, and took over the teaching at the local university. Upon the establishment the University has combined five faculties: Faculty of Law, Faculty of Philosophy, Faculty of Engineering, Faculty of Medicine (two years of introductory courses) and the Faculty of Theology (seminary). Within the Faculty of Engineering the courses have been divided into sections: mathematics, science, construction and architecture, mechanical and electrical engineering, 104

mining and metallurgy, social science and legislation and two miscellaneous courses. 31 August 1919 the first 18 teachers were appointed for the five faculties. A specialist in electro-mechanical engineering Dr. Milan Vidmar was one of them. The University courses at the future Faculty of Engineering were held even before the formal establishment of the University of Ljubljana. The courses were launched on 19 May 1919. Till the autumn the students have already graduated the first year of their studies. Consequently the courses for the first two years were held simultaneously in the first official academic year 1919/1920. In the first years of its existence Faculty of Engineering did not have suitable facilities neither for holding the lectures nor for the laboratory exercises. Nevertheless the professors were top experts recognized all over the Europe. In the first academic year 286 students, among them two ladies, were enrolled for the section of mechanical and electrical engineering. Each subsequent year 100 to 180 new students were enrolled. The enrollment in the first year was so high due to simultaneous courses for the first two years. The significant drop in enrollment had happened between 1931 and 1938. The economic crisis was the reason for this. The threatening Second World War had led to an increased enrollment because studying at the University was one of the ways to avoid the recruitment into the army. Although electrical engineering was his main interest, Milan Vidmar was a doctor of classical mechanical engineering. The title of his dissertation was “Die Theorie der Kreiselpumpe”, (“The Theory of Centrifugal Pumps”). He was well aware of the importance of mechanical engineering. According to the curriculum of the Faculty of Engineering, the first and the second semester consisted of theoretical courses: mathematics, mechanics, chem-

istry, descriptive geometry and technical documentation. These courses were mandatory for the students of mechanical and electrical engineering. In the third and fourth semester of the academic year 1919/1920 dr. Milan Vidmar started lecturing on electrical engineering and theory of machining. Simultaneously assistant professor Vladimir Stanek lectured on machine elements. After one year Vladimir Stanek left the University of Ljubljana and accepted a job in Czechoslovakia. He was replaced by a part-time lecturer Josip Boncelj. In 1923 he became an assistant professor and got a regular job at the Faculty. In autumn 1928 he moved to University of Zagreb where he was immediately habilitated to a full professorship and also became the head of the newly established Institute for Construction of Machine Components. Soon after he graduated Josip Boncelj was employed by the company Hardware Factories and Foundries of Ljubljana. Under his leadership it evolved from a craft workshop to a modern industrial system. The company became a source of quality engineers. Some of them later became university teachers: Milan Vidmar, Dobromil Uran, Albert Struna and Bojan Kraut. After the departure of Josip Boncelj part-time lecturers Stane Premelč and Romeo Strojnik (employed by the Faculty in the year 1919/1920) continued to teach at the mechanical engineering section. The Arrival of Feliks Lobe Ten years after being established the University of Ljubljana displayed a great progress but the section for the mechanical engineering was stagnating. As a great educator Milan Vidmar noticed this fact. In the beginning of 1929 he issued a tender for the acquiring of new lecturers. Since there were not many candidates the tender had to be repeated. In the response to Vidmar’s personal invitation Feliks Lobe, who later became the honorary doctor of University of Ljubljana and the member of Slovenian Academy of Sciences and Arts, applied for the lecturer position. At that time he was the head of technical section in the company “First Yugoslav Factory of TrainWagons, Machines and Bridges” in Slavonski Brod. Feliks Lobe was immediately given an associate professorship. His arrival proved to be of the great importance for the strengthening of the mechanical-engineering section at the Faculty of Engineering. But it was too late for establishing the complete mechanical-engineering studies in Ljubljana before the Second World War. The students of mechanical engineering were forced to continue and complete their studies outside Ljubljana, mostly in Prague, Brno or Zagreb. The main reason for slow progress and expansion of mechanical engineering was the lack of known and established experts. Above-the-average mechanical engineer was also not interested in extremely low wages at the University of Ljubljana. For example, Feliks Lobe’s income was less than one third of the income that he had received in Slavonski Brod. The second reason for very low interest in teaching at the University may be too slow a career progression. For example, Josip Boncelj as a high-profile expert started at Faculty of Engineering as part-time lecturer and needed two years to become an assistant profes-

sor. After he switched to the University of Zagreb he was immediately habilitated for the full professorship. Stane Premelč, for example, remained part-time lecturer for 17 years. Then he quit lecturing due to the health condition at the age of 49. He died 22 years later. Milan Vidmar can be considered as an initiator of complete mechanical-engineering studies at the University of Ljubljana. He comprehended the importance of mechanical-engineering studies and supported it all the way. He invested many years of personal effort, will and sometimes cunningness to succeed in the realization of his goals. Zoran Rant is also a very important scientist from the Faculty of Mechanical Engineering who made University of Ljubljana known throughout the world. Immediately after his arrival to the Faculty of Engineering, Feliks Lobe started lecturing on the energy related courses “Thermodynamics” and “Power Engines”. This can be considered as an embryo of today’s Department of Energy Engineering. After the year 1937 he lectured also on the course “Manufacturing Technologies”. His first teaching assistants were Leopold Andrée and Boris Černigoj. After the Second World War they both became lecturers at the Department of Energy Engineering. Feliks Lobe had two primary goals: the complete studies of mechanical engineering in Ljubljana and the construction of much-needed facilities. After many complications, that Feliks Lobe ably solved, the construction of the building named Institute of Mechanical Engineering commenced in the autumn 1938 and was stopped at the end of 1939. The construction of the building was completed only after the war and solemnly opened on 15 March 1946. Nowadays this building is known as “the old building”. In the basement and high ground of this building, in the tract along the Aškerčeva Street, there was the Laboratory for Manufacturing Technology”. In the tract along “Murnikova Street” there was “Caloric Laboratory”. Feliks Lobe’s open mind obviously foresaw a simultaneous development of two mechanical-engineering branches: solidbody mechanics and fluid mechanics or, in other words, manufacturing and energy engineering. The “Caloric Laboratory” was connected to the basement containing heating station with two boilers and coal compartment. “The old building” was heated by means of this heating station. The successor of the “Caloric Laboratory” is today’s Laboratory for Heat and Power”. After the employment of new part-time lecturers everything was ready for the complete studies of mechanical engineering in Ljubljana. Finally, according to the Yugoslav-government decree from 5 April 1941, an independent mechanical engineering section of Faculty of Engineering was constituted. Next day the war between Yugoslavia and Germany burst out and the decree was not actually realized. According to the written documents the courses at mechanical-engineering section were held even during the Second World War until the summer of 1943. In July 1944 the supervision over the University of Ljubljana moved from the chief of Ljubljana regional administration directly to the supreme commissioner of the Jadransko Primorje region located in Trieste. Such an arrangement remained valid till the end of the war. The under and post-graduate exams and promotions continued even during this time.

105

The Mechanical-Engineering Section at the Faculty of Engineering After the Second World War the council of the Faculty of Engineering on 23. June 1945 decided to divide the Faculty to seven departments: • • • • • • •

Department of Architecture, Department of Electrical Engineering, Department of Civil and Geodetic Engineering, Department of Chemistry, Department of Mining and Metallurgy, Department of Mechanical Engineering, Department for Supplementary Courses.

The curriculum of mechanical-engineering studies was upgraded and many new lecturers were hired. Feliks Lobe was personally involved in the majority of the selection and appointment process. During the year 1946 the following lecturers were habilitated: assistant professor Leopold Andree, professor Leon Kavčnik (left the Faculty of Engineering after one year), associate professor Franček Kovačec, associate professor Bojan Kraut, professor Zoran Rant, and professor Anton Vakselj. In the year 1948 and 1949 the following lecturers were habilitated: assistant professor Boris Černigoj, associate professor Boleslav Likar, professor Ivo Vuškovič (left the Faculty of Engineering after three years) and professor Dobromil Uran. Nine so called institutes were formed inside the mechanical-engineering section although they cannot be compared to nowadays institutes, chairs or departments. They comprised only specialized mechanical-engineering courses: • Institute for Theory of Machining, head: Zoran Rant, courses: thermodynamics, mechanisms, • Institute for Steam Engines, head Feliks Lobe, courses: technology of metals, manufacturing machines, steam engines, • Institute for Hydraulic Power Engines, head Albert Struna, courses: water turbines and pumps, lubrication and lubricants, internal combustion engines, motor-driven vehicles, • Institute for Aeronautics, head: Anton Kuhelj, courses: aeronautics. Anton Kuhelj was a member of the Department for Supplementary Courses but held lectures at the Department of Mechanical Engineering for about 30 years. • Institute for Measurements in Mechanical Engineering, head: Leopold Andree, courses: steam boilers, technical-measurement techniques, • Institute for Manufacturing Technologies, head: Leon Kavčnik, courses: technology of metals. Leon Kavčnik took over the courses that were previously lectured by a chemist prof.dr. Maks Samec. • Institute for Design of Lokomotives, head: Bojan Kraut, subjects, railroad vehicles, technology of manufacturing, thermodynamics and power engines for electrical engineers, • Institute for Elevators and Transportation Devices, head: Franček Kovačec, subjects: elevators and transportation devices, introduction to mechanical engineering

106

• Institute for Machine Elements, head Romeo Strojnik, courses: machine elements, technical documentation. According to the names of the institutes and the lecturers, approximately half of the institutes were energy and the other half technology related. Today’s Department of Energy Engineering is a direct successor of the Institute for Theory of Machining and partly a successor of the Institute for Steam Engines, Institute for Hydraulic Power Engines and Institute for Measurements in Mechanical Engineering. On 31 July 1948 the first two students graduated the complete mechanical-engineering studies in Ljubljana: Jožef Herman and Josip Kuralt. One of the first doctors of mechanical engineering became Zoran Rant in 1950. He was the professor at the University of Ljubljana and later at the Technical University of Braunschweig, Germany. For years 1919 to 1945 it was characteristic that mechanical engineering in Slovenia had a hard time ploughing its independent route, recruiting quality teaching staff and was facing a lack of suitable facilities. We can say that after the end of the Second World War the pioneering period of Slovene university-level mechanical engineering had ended. During this period professors were mainly occupied with teaching and writing of the textbooks, acquiring equipment for the laboratories and with manufacturing or purchasing various teaching tools. Expert work and scientific research was neglected in this period. The leadership of the University was facing many administrative and financial obstacles caused by the government in Belgrade and the Second World War. In the year 1950 University of Ljubljana, first time after thirty years of its existence, faced a major reform: Faculty of Engineering and Faculty of Medicine separated from the University and constituted their own rectorates and administration. Former departments at the Faculty of Engineering were renamed to faculties except for the Department for Supplementary Courses. After a period of four years Faculty of Engineering and Faculty of Medicine returned into the University of Ljubljana. In the year 1952 Faculty of Theology was removed from the list of national institutions. In 1957 Faculty of Engineering was discontinued. During this reorganization the Departments for Mechanical Engineering and Electrical Engineering merged to form the Faculty of Electrical and Mechanical Engineering which lasted for only three years. Independent Faculty of Mechanical Engineering According to the decision of the University Council from 1 October 1960 during summer vacations the Faculty of Electrical and Mechanical Engineering was divided to two independent units: Faculty of Electrical Engineering and Faculty of Mechanical Engineering. The first dean of the independent Faculty of Mechanical Engineering was Zoran Rant. Along with this reorganization three-stage studies were introduced ending with a diploma, a master thesis and a doctorate dissertation. The Faculty of Mechanical Engineering moved away from a decade-lasting tradition

and significantly lengthened the path to the doctorate. This system had lasted for about 40 years. Lately the old traditional two-stage path was reintroduced: ten-semester undergraduate studies ending with a diploma and postgraduate studies ending with a doctorate dissertation. In 1960 when the Department for Mechanical Engineering was upgraded to the Faculty of Mechanical Engineering, it consisted of four units named “deparments” defined as organizational units where teachers along with assistants and other staff carried out pedagogic, scientific and expert work. This definition of the departments is still valid. From the names of these departments three classical mechanical engineering fields can be sensed: Construction and Mechanics, Manufacturing Technologies, Energy and process engineering: • Department of Working Machines and Transportation • Department of Manufacturing Technology • Department of Power Engines and Thermal Devices • Unit for Mathematics and Other Supplementary Courses In the school year 1962/63 the Institute of Mechanical Engineering was constituted in the scope of the Faculty of Mechanical Engineering. It was a financially independent subject employing the staff from the Faculty of Mechanical Engineering. Its first goal was to establish the cooperation with industry and to carry out scientific-research work. In the year 1966/67 studies were divided into two main branches: technology and energy related. The names of the departments remained unchanged. The number of students rapidly increased after the end of the Second World War and current facilities were not adequate any more. In the year 1958 location documentation for “the new building” was issued. The project documentation was finalized in 1962. In the year 1964 the construction of the building commenced but was suspended due to the financial reasons in 1965. The construction continued in 1968 and ended in 1971. The great role in finalizing the construction of “the new building” played Bojan Kraut who later became the honorary doctor of the University of Maribor and professor emeritus. Years from 1945 to 1971 can be considered a period of classical university-level mechanical engineering in Slovenia. The Faculty of Mechanical Energy gained respect in industry which craved for its expert work and support after the Second World War. Almost all new professors originated from the industry and that is one of the reasons why the teaching staff of the Department for Mechanical Engineering and later Faculty of Mechanical Engineering successfully cooperated with many companies offering their expert and scientific knowledge. University teachers were extensively involved in the design, repair and maintenance work on machines and devices. Namely at that time Yugoslavia was politically and economically isolated from the developed western countries. Modern Slovene terminology was evolving and a big role was played by honorary doctor Albert Struna who promoted and initiated the publishing of the Journal of Mechanical Engineering. New expert societies were emerging and first Mechanical Engineering Handbook Guide was published by Bojan Kraut

(“Kraut’s” Mechanical Engineering Handbook Guide). The conditions for scientific-research work were still unfavorable although Zoran Rant should be mentioned due to his pioneering and world-widely recognized research work dealing with the quality of energy. Introduction of terms “Exergy” (1953) and “Anergy” built him an everlasting monument and placed University of Ljubljana among the established universities of the world. Modern Faculty of Mechanical Engineering In the year 1971, when the construction of “the new building” was finished and populated, a period of classical university-level mechanical engineering came to its end. Characteristic for the first period (1919-1945) was a lack of facilities and teaching staff which prevented the complete studies of mechanical engineering. The Second World War contributed a lot to this unfavorable situation. During the second period (1945-1971) complete studies of mechanical engineering were realized and gained the reputation in Slovenia. Simultaneously cooperation of Faculty of Mechanical Engineering and industry began. After 1971 a period of modern development of the Faculty of Mechanical Engineering began. New square footage in “the new building” enabled the expansion of scientific-research work. Faculty of Mechanical Engineering opened its door to the world. In the school year 1971/1972 the number of departments increased to six and the next year to seven: • • • • • • •

Department of Energy Engineering Department of Heating and Process Engineering Department of Working Machines Department of Cybernetics Department of Manufacturing Technologies Department of Materials Unit for Supplementary Division

The number of departments remained unchanged for 25 years. Two thirds of the departments were dealing with manufacturing technologies and only one third of them with energy: From the day it was constituted the Department of Energy Engineering contains three laboratories: Laboratory for Heat and Power, Laboratory for Internal Combustion Engines and Laboratory for Hydraulic Machines. Nowadays Faculty of Mechanical Engineering comprises 19 departments.

107

1 10 8

16 15

17 19

4 6 5 13

11 12 22

1 Ime Priimek

University of Ljubljana Faculty of Mechanical Engineering Department of Energy Engineering AĹĄkerÄ?eva cesta 6 SI-1000 Ljubljana Slovenia