September 2014 by Energy-Tech Magazine

Avoid Voltage Collapse 12 • Creep Damage 15 • ASME: Correction Curves 18

ENERGY-TECH A WoodwardBizMedia Publication

SEPTEMBER 2014

www.energy-tech.com

Dedicated to the Engineering, Operations & Maintenance of Electric Power Plants In Association with the ASME Power Division

Analyzing boiler tube failure

Cogeneration biomass

waste to watts

Boiler Replacement Parts

toll free 1.877.774.8778 24hr emergency service

229.221.4690

ENERGYT ECH P.O. Box 388 • Dubuque, IA 52004-0388 800.977.0474 • Fax: 563.588.3848 Email: sales@WoodwardBizMedia.com www.energy-tech.com Energy-Tech (ISSN# 2330-0191) is published monthly in print and digital format except in January and July, when it is published in digital format only by WoodwardBizMedia, a division of Woodward Communications, Inc. WoodwardBizMedia assumes no responsibility for inaccuracies, errors or advertising content. Entire contents © 2014 WoodwardBizMedia. All rights reserved; reproduction in whole or in part without permission is prohibited. Printed in the U.S.A. Group Publisher Karen Ruden – kruden@WoodwardBizMedia.com General Manager Randy Rodgers – randy.rodgers@woodwardbizmedia.com Managing Editor Andrea Hauser – ahauser@WoodwardBizMedia.com Editorial Board (editorial@WoodwardBizMedia.com) Kris Brandt – Rockwell Automation Bill Moore – Director, Technical Service, National Electric Coil Ram Madugula – Executive Vice President, Power Engineers Collaborative, LLC Kuda Mutama – Engineering Manager, TS Power Plant Editorial views expressed within do not necessarily reflect those of Energy-Tech magazine or WoodwardBizMedia.

FEAtUrEs

By Wendy Weiss and Terry Totemeier, Ph.D., Structural Integrity Associates Inc.

Address Correction Postmaster: Send address correction to: Energy-Tech, P.O. Box 388, Dubuque, IA 52004-0388 Subscription Information Energy-Tech is mailed free to all qualified requesters. To subscribe, go to www.energy-tech.com or contact Linda Flannery at circulation@WoodwardBizMedia.com Media Information For media kits, contact Energy-Tech at 800.977.0474, www.energy-tech.com or sales@WoodwardBizMedia.com. Editorial Submission Send press releases to: Editorial Dept., Energy-Tech, P.O. Box 388, Dubuque, IA 52004-0388 Ph 563.588.3857 • Fax 563.588.3848 email: editorial@WoodwardBizMedia.com. Advertising Submission Send advertising submissions to: Energy-Tech, 801 Bluff Street, Dubuque, Iowa 52001 E-mail: ETart@WoodwardBizMedia.com.

A division of Woodward Communications, Inc.

September 2014

Avoiding critical voltage collapse in a changing environment By Tony Oruga, P.E., Eaton’s Cooper Power Systems Division

CoLUMNs

Maintenance Matters

Evaluating creep damage in Grade 91 steels By Kent Coleman, Electric Power Research Institute

Machine Doctor

Compressor vibration due to bearing and seal problems By Patrick J. Smith

AsME FEAtUrE

Advertising Sales Executives Tim Koehler – tkoehler@WoodwardBizMedia.com Joan Gross – jgross@WoodwardBizMedia.com Creative/Production Manager Hobie Wood – hwood@WoodwardBizMedia.com Graphic Artist Valerie Vorwald – vvorwald@WoodwardBizMedia.com

You need a boiler tube analysis – Now what?

Quantifying correction curve uncertainty through empirical methods By Christopher R. Bañares, Thomas P. Schmitt, Evan E. Daigle and Thomas P. Winterberger, General Electric Power & Water

iNdUstrY NotEs

4 30 31

Editor’s Note and Calendar Advertisers’ Index Energy Showcase

oN tHE WEB Don’t miss Energy-Tech’s Sept. 16 webinar with Gaumer Process, Electric Process Heating and Demand Fuel Switching, and our Sept. 30 webinar, Power Generating Asset Management, with Komandur Sunder Raj. The live presentations begin at 1 p.m. CST (6 p.m. GMT) and attendees will be eligible to receive 1 PDH credit. Visit www.energy-tech.com for more information.

Cover photo contributed by Structural Integrity Associates.

ENERGY-TECH.com

Editor’s Note

Outage season prep Get details done before the big shutdown For the huge scale of a utility power plant, it is consistently amazing how just one small problem can shut down its power production in a moment. We recently completed Energy-Tech University’s online webinar sessions, Steam and Gas Turbine Fundamentals and Advanced Turbine Fundamentals, with Steve Reid from TG Advisers. As I moderated the presentations and listened to Steve explain the photos in the power point slide, attention to detail was a reoccurring theme. We’ve continued that theme in this issue, although it wasn’t really on purpose. But as many of our readers head into outage season at their plants, it seemed like a good time to talk about non-destructive testing advances, boiler tube inspections and preventing bearing and seal problems. So be sure to read, Evaluating creep damage in Grade 91 steels, by Kent Coleman with the Electric Power Research Institute on page 15. Then turn to page 6 for, You need boiler tube analysis – Now what?, by Wendy Weiss and Terry Totemeier from Structural Integrity Associates Inc. Finally, don’t miss, Compressor vibration due to bearing and seal problems, on page 25. It’s by Patrick J. Smith, who has written as the magazine’s Machine Doctor for several years. If you aren’t a fan of Pat’s, you probably should be. Visit www.energy-tech.com is you want to read more of his work. And if you’re interested in the Energy-Tech University turbine course that just concluded, it’s available for sale on Energy-Tech’s ContentShelf site.Visit www.energy-tech.com to learn more about it. Finally, I hope you can join us for two webinars in September. The first is on Sept. 16 with Gaumer Process, Electric Process Heating and Demand Fuel Switching. The second is on Sept. 30, Power Generating Asset Management, with Komandur Sunder Raj. The presentations will begin at 1 p.m. CST (6 p.m. GMT) and registration is free. Also, attendees to the live presentations will receive a PDH credit for each of them, so sign up today by visiting www.energy-tech.com. I hope outage season goes smoothly for everyone and, in the meantime, thanks for reading.

Andrea Hauser

ENERGY-TECH.com

CALENDAR Sept. 9-11, 2014 2014 Dry Scrubber Users Association Minneapolis, Minn. www.dryscrubberusers.org Sept. 9-11, 2014 Feedwater Heater Operation and Maintenance Seminar Atlantic City, NJ www.powerfect.com Sept. 16, 2014 Webinar: Electric Process Heating and Demand Fuel Switching Presented by Gaumer Process www.energy-tech.com Sept. 16-19, 2014 Machinery Vibration Analysis Salem, Mass. www.vi-institute.org Sept. 30, 2014 Webinar: Power Generating Asset Management Presented by Energy-Tech & Komandur Sunder Raj www.energy-tech.com Oct. 26-28, 2014 Power Plant Management & Generation Summit Atlanta, Ga. www.ppmgsummit.com/mp_et Nov. 3-4, 2014 CCGT 2014: O&M and Lifecycle Management for CCGT Power Plants Houston, Texas www.tacook.com/ccgt-usa Nov. 11-14, 2014 Advanced Vibration Control Syria, Va. www.vi-institute.org

Submit your events by emailing editorial@woodwardbizmedia.com.

September 2014

Enhance plant-wide performance with Rotork actuator and damper-drive solutions Rotork provides high-performance, cost-effective valve actuator and damper drive solutions throughout your entire power plant. Solutions that can help you maximize plant performance, minimize maintenance problems, and help you meet stringent safety and environmental regulations. Hundreds of different valve actuators and damper drives are required throughout a typical power plant facility. Rotork can meet all of your needs by offering motor-operated, fluid power, electro-hydraulic, and manual gear operators. Actuators and drives suited for harsh, rugged operating conditions, inside and out, servicing every size and style of valve from small process control valves to superlarge isolation valves as well as rotary, linear, or quarter-turn combustion-air and flue-gas dampers. Furthermore, the advanced technological capabilities of Rotork smart actuators can help you easily upgrade connectivity and provide a smart window to your process that includes important predictive maintenance data to help you improve productivity and efficiency as well as cut maintenance costs. Whether you are building a new plant or are currently operating a fossil-fuel, nuclear, or green-energy facility, we can help. Rotork actuators and drives are currently servicing hundreds of power plants around the world. Our more-than 50-years experience and proven expertise can help you, too. Contact us today.

IQ3 electric actuator with smart window technology

SM6000 S2 heavy-duty electric damper drive

Rotork Controls, Inc. 675 Mile Crossing Blvd. Rochester, NY 14624 phone: 585 247 2304 email: info@rotork.com

www.rotork.com

Redefining Flow Control

Type K pneumatic rotary damper drive

FEATURES

You need a boiler tube analysis – Now what? By Wendy Weiss and Terry Totemeier, Ph.D., Structural Integrity Associates Inc.

For more than 30 years, boiler tube failures have been the leading cause of lost boiler availability and forced outages. Determining the mode of damage responsible for a failure is important to ensure that the proper corrective actions are taken and similar failures do not occur again. There are more than 30 different damage/failure mechanisms that can affect boiler tubes, some of which superficially appear similar, but can have very different underlying characteristics. An example is a so-called “fish mouth” failure that might be a consequence of several different damage mechanisms. Therefore, understanding the damage mechanism that caused a failure is essential to provide insight into the underlying root cause so that appropriate corrective actions can be implemented.

Figure 1. Failed superheater tube (long-term overheating)

These corrective actions can range from adjustments to operation – such as burner systems, cycle chemistry, steam temperatures and steam pressures – through pre-emptive nondestructive examination to identify other “at risk” tubes, to defining an appropriate repair strategy – such as pad welding, application of overlay or complete tube replacement. Failures are not the only reason a tube sample might be removed for analysis. As part of a proactive approach to life management of boiler tubing, it is prudent to periodically sample “typical” tubes for insights into tube condition (remaining life) that can only be gained through a destructive, metallurgical evaluation of the tube. For example, a high pressure (HP) evaporator tube could be removed from a heat-recovery steam generator (HRSG) for evaluation of the internal deposits to help assess the water treatment program. Superheater and reheater tubes are often removed from boilers for condition assessment to determine the remaining life of those components. A key part of such proactive tube sampling is knowing where to extract representative samples, which requires a thorough understanding of the overall boiler condition and operations. But that would be a whole article in itself, so here we will assume that an engineering-based program is in place for tube sampling, or that samples have been extracted in response to tube failures.

Collecting a tube sample If you have a tube failure, selecting the appropriate sample for analysis is usually straightforward, but in some cases (such as those where a tube failure has resulted in so-called “secondary damage” to other tubes), it might not be so obvious where the failure originated. For these situations, taking multiple samples is advised. Also, removing adjacent samples that might have similar damage that has not resulted in a failure can be beneficial to the root cause determination. Sometimes failures can be violent and dislodge deposits or cause damage that impairs evaluation of the failed area. For proactive condition assessment, samples should generally be taken from the hottest location within a component (or of a particular material within a component). For example, in a superheater or reheater, samples from the “ferritic” (e.g. Grade T22) side of a transition to stainless steel usually represent the “hottest” condition for that “ferritic” material. In the case of a dissimilar metal weld, this also provides an opportunity to sample that DMW to assess its condition. If samples are being extracted from an evaporator for internal

Figure 2. Failed waterwall tube (short-term overheating)

ENERGY-TECH.com

September 2014

FEATURES deposit evaluation, then the “hottest” location generally corresponds to that with the peak heat flux. When removing tube sections, there are some best practices to follow. First, mark the tube before it is cut out. Indicate the tube identification, such as horizontal/vertical, top/bottom, direction of flow, gas or furnace facing (hot) side. If the damaged area is hard to visually locate, such as a pin hole leak, mark it as well. Also, be sure to keep track of the tube number, assembly number and elevation (and keep a record of how this is identified – e.g. left to right or front to back). By marking the tube before it is cut out, there is less of chance of tube identifications getting mixed up or the hot and cold sides being mislabeled. If overheating is suspected, or if a steam-side tube Figure 3. Tube failures can cause a lot of secondary from the hottest part of the component has been sampled, including a section of the damage. same heat of material from the coldest end of the circuit can be very beneficial, if available. A cold end sample allows for ® assessment of the microstructure in the condition that is most similar to the original, and provides reference dimensions (e.g. original wall thickness) with minimal effects of service (wastage or swelling). When cutting samples from tubes in a boiler: • Providing at least 12˝ of tube length for an analysis is generally adequate; if there is a failure or an area of interest, it should be in about the middle of the sample. If a failed tube Bright, B right, High-Res is being removed for analysis, more Video & Photos material could be required dependLarge Range-of-Focus ing on the extent of the damage. 90° Prism & Quality Construction • Prevent contamination, including Close-Focus 4-Way Articulation from cutting debris and cutting tips available! 4 & 6 mm Diameters fluids (it is best not to use cutting Starting at only $8,995 fluids when removing the tube). It is important to ensure that both the OD and ID surfaces are protected from contamination in case deposits on those surfaces prove to be important in identifying the damage VIDEO BORESCOPES mechanism. Obviously, if the internal deposits are being analyzed, they Ideal for cooling tube inspection! need to be contaminant-free. We’ve improved the image quality in the new Hawkeye® • Be careful not to dislodge deposits V2 with a higher resolution, more light sensitive camera, when handling tube. For the same delivering bright, crisp, clear images! The new 5” LCD reason that you don’t want to conMonitor provides comfortable viewing, and intuitive, taminant the deposits, you also want easy-to-use controls, allow photo and video capture to make sure they stay as intact as Quickly inspect cooling tubes at the touch of a button! We’ve increased the 4-way possible. inside heat exchangers, turbine articulation range, and improved the feel. It’s still small, • Ensure cutting techniques do not blades, and much more! lightweight, portable, delivers great image quality, and alter the tube damage or microstrucis priced starting at only $8995. Available in 4 and ture. Torch cutting can heat the tube 6 mm diameters. Optional 90° Prism and Close-Focus metal to the point where the tube adapter tips available. microstructure is altered, which is detrimental to an analysis for which Made in USA assessment of the microstructure is a necessary step (e.g., overheating failgradientlens.com/V2 800.536.0790

The NEW Hawkeye

Video Borescope!

September 2014 ENERGY-TECH.com

FEATURES

Figure 4. A well labeled tube sample.

Figure 5. OD wastage on an economizer tube.

ures or condition assessments). Torch cutting also melts the area that is being cut, so it is very important to ensure that the failure, or area of interest, is not affected. Torch cuts, if made, need to be at least 6˝ from the area of interest. Torch cuts also can leave splatter behind that can alter deposit loading (deposit weight density) measurements and deposit compositional analyses, so splatter should be prevented or minimized to the extent possible. • Abrasive saws can leave behind some heat damage, so make sure cuts are several inches from the affected area. Once the tube section is out of the boiler or HRSG and properly marked and identified, it needs to be prepared for shipping. Proper shipping practices include taping the tube ends to prevent internal contamination, protecting areas of interest from impact damage, making sure markings on the tube will not get rubbed and shipping in a wood crate (this

ENERGY-TECH.com

might seem obvious, but samples have been lost from cardboard boxes during shipping!). Selecting, labeling, removing and shipping the tube sample is just the first step in a good tube evaluation. Equally important is providing thorough background information. The starting point is good information about the tube, including the material’s specification and dimensions. Other helpful information includes: • Drawing to show location of tube • Design and operating parameters (overall boiler and tube) • Operating hours • Total starts • Cycle chemistry • Last chemical clean • Maintenance records • Past tube failure, history/failure analysis reports • Other unit history, information, problems

Metallurgical evaluation Now that a tube sample has been provided for evaluation with good background information, the metallurgical analysis can proceed. While the procedure can be tailored based on particular conditions, in general, the following steps are required for a thorough evaluation. • Visual examination and photo-documentation The tube condition is photographed prior to destructive analysis to record any distinctive features. This visual examination also is used to determine “cut planes” for the sample, which will be analyzed in subsequent steps. • NDE (if appropriate) Before the tube is sectioned, the tube might be examined by various non-destructive techniques (e.g., phased array ultrasonics) to both help identify areas for sectioning or assess the effectiveness of an NDE technique to locate similar damage in other tubes. • Chemical analysis A chemical analysis is performed to determine if the tube metal is within specification, and if any particular additional unspecified or trace elements September 2014

FEATURES

Figure 6. Dye penetrant indications.

might be present that could affect the serviceability of the metal. • Dimensional measurements Tube diameter and wall thickness are measured around the circumference of the tube, possibly at several locations along the sample, to characterize wall loss and swelling. • Hardness evaluation and/or Mechanical property testing Hardness tests can easily be performed on metallurgical sections and provide an indication of the tensile strength of the metal or metallurgical condition. Other mechanical property testing, such as obtaining elevated temperature properties (creep strength), can be performed as part of more detailed investigations, such as remaining life evaluations. • Metallography Tube cross sections (or portions of the cross section) are mounted in a plastic resin and carefully polished to a mirror finish. The sample may be examined in this condition (e.g. to identify holes or cavities) or may be etched with chemicals to reveal the micro-

Restoration of

Gas Turbine

Bearings

A Division of MD&A

Renewal Parts Maintenance 4485 Glenbrook Rd. | Willoughby, OH 44094 ph. 440-946-0082 | www.RenewalParts.com

PARTS | SERVICES | REPAIRS

September 2014 ENERGY-TECH.com

FEATURES

Figure 7. Creep in T22 tube to header HAZ.

Figure 8. Creep in Grade 91 base metal.

Figure 9. Creep in stainless steel reheater tube.

structure (grain size and morphology), which provides additional insight into the condition of the metal and is helpful to identify both damage and the effects of service exposure (such as changes in the microstructure due to high temperature exposure). Such examinations may be performed using optical microscopes; a scanning electron microscope (SEM) may be used to obtain higher resolution images to resolve fine-scale precipitates in the metal. The SEM also can be used to provide 10

ENERGY-TECH.com

local chemical analysis using energy dispersive spectrometry (EDS), including elemental maps, which can be particularly valuable in diagnosis of some damage mechanisms, especially those involving corrosion. • Fractography If the sample includes a fracture (broken) surface, then this is examined to determine its characteristics. The morphology of the fracture surface provides insight into the mode of failure (transgranular/ intergranular) and also might indicate the presence of precipitates, cavities or foreign species that have exacerbated the failure. Fracture surfaces are commonly examined with a stereomicroscope, which provides a large depth of field, or with a scanning electron microscope. • Characterization of internal and/or external oxide/deposits Often, internal or external deposits (oxidation or corrosion products) play a significant role in the damage mechanism, either by directly causing wall loss or internal attack of the metal, or by acting as a secondary contributor to a failure (e.g. internal oxide scale “insulating” a steam touched tube and causing an increase in the tube metal temperature). As a result, the thickness and morphology, and in some cases the chemical composition or crystallographic structure, is mapped to assess the role that these corrosion products play in the failure or overall condition of a tube. In some cases, such as for evaporator tubes, the quantity of deposits (a so-called “deposit loading”) will be measured to determine the need for chemical cleaning. The results of these various analyses and measurements are used to draw a conclusion about the underlying condition of the tube material and the damage mechanism that resulted in failure (or degradation) of the tube. Because of the similarities between a number of damage mechanisms, this requires not only a good metallurgical knowledge, but an understanding of where the tube is located within the boiler, and what operating conditions are possible for that tube (which is why the circumstantial information about the tube sample and its location in the boiler is so critical). To aid in diagnosis of the damage mechanism, stress and temperature calculations also might be performed to determine if it was reasonable to have expected tube failure in the period of service experienced, or if some excursion or other detrimental condition also might have contributed to failure. In the case of tube condition assessments, such calculations are performed to identify the likely remaining life, and this is compared and contrasted against the metallurgical condition of the sample.

From damage mechanism to root cause This systematic approach definitively identifies the damage mechanism (what caused the tube to fail) but further work is often needed to identify the root cause of that damage mechanism (why the damage mechanism occurred) and define corrective actions. This latter step of root cause identification often requires a broader engineering evaluation that September 2014

FEATURES

Figure 10. Beach marks indicating a fatigue crack from a tight bend in a superheater tube.

encompasses the metallurgical work, other engineering evaluations and an understanding of the boiler operation. Only once that root cause is identified can an appropriate set of corrective actions be defined. Often, it is tempting to jump from identification of the damage mechanism to corrective actions, but omitting the step of root cause identification often can lead to misdiagnosis of the underlying reason for failures, and to ineffective corrective actions. Hence involving a team with a broad multidisciplinary understanding of metallurgy, engineering/operating and nondestructive testing is crucial to effective life management of boiler tubing. ~ Wendy Weiss is the FPS Materials Science Center manager. She has a bachelor’s degree from the New Mexico Institute of Mining and Technology and a master’s degree from the University of Texas at Austin. Her primary responsibilities are performing failure analyses and condition assessments of components from fossil power plants at Structural Integrity Associates’ metallurgical laboratory. You may contact her by emailing editorial@woodwardbizmedia.com. Terry Totemeier, Ph.D., is in Fossil Plant Services at Structural Integrity Associates Inc. He has a Ph.D. (Metallurgy), from the University of Cambridge (UK), and a bachelor’s degree (Materials Science and Engineering), from the Massachusetts Institute of Technology. He also has more than 20 years of experience performing research, development and failure analysis of materials used in power generation equipment, both fossil and nuclear, with a primary focus on the physical metallurgy, mechanical behavior and oxidation/corrosion resistance of Fe-base and Ni-base alloys for high-temperature service. You may contact him by emailing editorial@woodwardbizmedia.com.

Applications: • Steam pressure vessels • Hot water heaters • Demineralizers • Steam humidifiers • Water purifiers • Refrigeration units • Liquid treatment vessels

• Compressed air tanks • Filtering units • Dryer cans in paper mills • Water towers • Water softeners • Deaerators • Make-up tanks

The Topog-E Gasket Company produces gaskets to fit every boiler in production today. Topog-E® Series 180 molded rubber gaskets are special for many reasons. Over forty years as the industry standard have given us feedback as to why!

1224 North Utica | Tulsa, OK 74110 T: 918-587-6649 | E: info@topog-e.com W: www.topog-e.com

September 2014 ENERGY-TECH.com

FEATURES

Avoiding critical voltage collapse in a changing environment By Tony Oruga, P.E., Eatonâ&#x20AC;&#x2122;s Cooper Power Systems Division

For years, capacitor banks have solved power quality challenges in transmission and distribution networks, including voltage collapses and phase shifts associated with alternating current (AC) power supply systems. What can power suppliers do to correct the same anomalies caused by an emergency outage or a temporary increase in reactive power load? A mobile capacitor bank is engineered to deliver the reactive power compensation and voltage support needed in temporary situations that often represent a challenge to both utilities and their customers. Consider this: Reactive power is needed to maintain the voltage required to deliver active power through the

ENERGY-TECH.com

transmission and distribution grid. Electric utility suppliers often accomplish this by strategically placing capacitors on the network to maintain the gridâ&#x20AC;&#x2122;s ability to push active power through the transmission and distribution system. This results in a robust, reliable and quality power source for customers. Utilities can apply the same methodology to emergency or temporary situations by deploying mobile capacitor banks.

Challenges facing the power grid One of the major concerns associated with utility power grids is the aging nature of the infrastructure and its tendency to take on a variety of power quality issues. Electric equipment such as generators, transformers, regulators and the distribution and transmission lines are constantly subject to faults, overloading, environmental conditions and vandalism. Additionally, it is not uncommon for the grid to experience unexpected outages that can often lead to voltage collapse or capacity issues cascading to other parts of the system. When the grid appears to demonstrate these types of issues, the challenge is to stabilize the grid quickly in order to get customers back online. If executed successfully, the overall impact to the customer is minimal, with decreased utility downtime and diminished impact on overloaded equipment needed to sustain the grid. This can even lead to a reduction in customer complaints.

September 2014

FEATURES

encounter difficulties within their aging power systems. Capacitor bank applications These utilities are consistently challenged with emergency For decades, capacitor banks have been commonly used outages, unique maintenance situations, the need for peak in power systems to address these issues by supporting the loading support, voltage collapse or the need for reactive system voltage, increasing power flow capability, releasing power support. They also might need a way to delay costly system capacity, improving losses and reducing utility billcapital investments. ing. Capacitors are designed to offer a long-term benefit to The mobile capacitor bank can represent a viable soluthe customer and can last up to 30 years if properly maintion in many of these scenarios. It offers the same features tained. In many instances, the savings during the life of the and benefits of a regular capacitor bank, but with the added product pays for the initial investment several times over, flexibility to be quickly deployed and placed anywhere on depending on the application and its power supply needs. the system to support an immediate need. Capacitor bank designs vary widely in arrangement, from externally or internally fused to fuseless configurations. Capacitor banks also can be connected to the power system in a variety of different configurations depending on the application, such Gaumer has industry leading knowledge in as single or double grounded-wye, single or double ungrounded-wye and delta. fuel gas conditioning including electric heater, The flexibility of these options allows filter/coalescer and control panel design. the use of capacitors to be tailored to Gaumer engineers will work with your meet each customer’s unique application unique operating conditions to provide requirements. Capacitor banks are engineered-to-ora complete, successful solution. der. Though some banks are similar in Call today for: design, there are countless other designs • Fuel Gas Conditioning manufactured that are exclusive to the • Fuel Gas Heaters customer. Expertise in the proper application of capacitor products is crucial, and • Fuel Gas Filters if misapplied can lead to capacitor failure, resonance issues, leading power factor, overvoltages and/or create other system issues.

Fuel Gas Problems?

Rising demand for fast, reactive solutions As the economy grows, the demand for energy also increases. This drives the need for immediate power solutions for applications and environments that

Celebrating 50+ Years

Fuel Gas Conditioning for all Gas Turbines 713.460.5200

www.gaumer.com

September 2014 ENERGY-TECH.com

FEATURES

Latest solutions in reactive power Mobile capacitor banks can be customized to meet each customer’s unique requirements, from a simple capacitor bank to a self-contained substation up to 230 kV with the required controls and protection. Since these banks are able to be transported easily, they are typically constructed to include all the necessary working components typically seen with a substation class capacitor bank, but can be grouped together on one or more trailers in order to meet federal and state transportation requirements. With transportation requirements in mind, switches and breakers containing SF6 gas are subject to Department of Transportation (DOT) and Environmental Protection Agency (EPA) regulations. These situations require reclaiming of the SF6 gas before transporting the switches and breakers and can be accomplished by using a special SF6 gas handling pump, which can be installed on the vehicle’s trailer.

ENERGY-TECH.com

Mobile capacitor banks often come complete with protection and control schemes, capacitor switching and interrupting devices, control sensing and protective fencing. Some of the additional accompanying equipment can include lightning arresters, current-limiting reactors, polymer insulators, safety fencing, grounding mats and storage enclosures. With more than 70 years of experience designing and manufacturing power capacitors, Eaton’s Cooper Power Systems offer a comprehensive portfolio of power distribution products, including mobile capacitor banks. Additionally, Eaton’s Cooper Power Systems support customers globally by answering technical questions, performing system studies and performing onsite commissioning and product maintenance for capacitor banks. This includes the supply of complex engineered-to-order products, such as capacitors, open-rack capacitor banks, metal-enclosed capacitor banks, pole-mounted capacitor racks, capacitor switches and mobile capacitor banks. When applied and working properly, traditional capacitor banks can greatly increase the overall efficiency of the power system for various unique applications. Having the flexibility of a mobile capacitor bank provides utilities with the security of effectively stabilizing the grid by reacting quickly to unexpected contingencies. ~ Tony Oruga is a senior product application engineer with Eaton’s Cooper Power Systems capacitor business. He creates capacitorrelated design solutions to meet customer specifications and necessities globally. Oruga has a bachelor’s degree in Electrical Engineering and is a registered professional engineer with 11 years of power systems experience. You may contact him by emailing editorial@woodwardbizmedia.com.

September 2014

Maintenance Matters

Evaluating creep damage in Grade 91 steels By Kent Coleman, Electric Power Research Institute

EPRI is routinely asked to perform third-party review of failure analysis reports for power plant component failures. These reviews are often requested for components manufactured from new materials, including creep strength enhanced ferritic (CSEF) steels. Several recent reviews have included evaluations by others in the industry using a life prediction technique known as the “EPRI-Neubauer correlation.” This technique, which was developed in the 1980s by EPRI in conjunction with Bernard Neubauer, utilizes surface replication and relates damage development to remaining life in low-alloy steels. The EPRI-Neubauer correlation is appropriate for some low-alloy steels such as Grades 11 and 22. However, recent Figure 1. Development of creep damage in Grade 11 and 22 steels. research indicates that it is not suitable for evaluation of CSEF steels such as Grade high-temperature components to determine the suitability for 91. EPRI is developing new tools and techniques to identify continued service. It also is very useful to determine the matecreep damage and predict remaining life of these steels. rial’s structure. Grades 11 and 22 have a microstructure with distinct grain Creep damage in Grades 11 and 22 steel boundaries that easily lend themselves to the EPRI-Neubauer Low-alloy steels, commonly referred to by their American correlation, and many developments have been made in the Society for Testing and Materials (ASTM) classifications, Grades industry to apply this model to low-alloy steels. However, even 11 and 22, develop creep damage in a well-established order, with these steels, users need to know when the technique starting with isolated creep voids, aligned creep voids, micro should be applied. cracking and finally macro cracking. A diagram commonly used One of the difficulties with the EPRI-Neubauer correlation to illustrate this progression is shown in Figure 1. is that it reveals the damage level, and resultant remaining life, The EPRI-Neubauer correlation has been applied to metonly on the component’s surface. Several types of welds, includallurgical samples removed from high-temperature components ing longitudinal seam welds, develop damage sub-surface. The in power plants, including piping and headers, as a method to damage on the surface might not be representative of the damdetermine remaining life. Additionally, the method has been age throughout the component. This same phenomenon also applied in-situ on the surface of components by the technique has been observed in circumferential welds in very old systems. of replication. Evidence suggests that failures earlier in life might be domiTo apply this method in the field, a small area of the comnated by bending stress, which might be highest on the surface ponent is first polished to a very fine finish. The polished area is of the pipe. However, on welds with lower bending stress, the then etched to reveal the microstructure. Then an acetate film highest level of damage might develop subsurface. More inforis softened and applied to the polished area. The acetate fills mation can be found in the EPRI report Circumferential Seam the profile of the etched surface and makes a “replica” of the Weld Cracking: An Interim Report (1014295), published in microstructure, much like a 3-D picture. The replica can then 2007. be taken to a laboratory microscope, and the microstructure can be evaluated. This technique is often applied to welds in

September 2014 ENERGY-TECH.com

Maintenance Matters

Figure 2. Creep testing of samples instrumented with acoustic emission wave guides.

Creep damage in Grade 91 steel More recently, instances have occurred in the industry of the EPRI-Neubauer correlation being applied to alloys for which it was not developed, including CSEF alloys. Grade 91 steel is one member of the family of CSEF steels. This steel was developed during the 1980s and is increasingly considered the material of choice for boiler, piping and header applications in power plants. This steel is being routinely installed in high-energy components in both fossil-fuel-fired

Like Energy-Tech? Like us on Facebook for exclusive content, conversations and events!

ENERGY-TECH.com

Dedicated to the Engineering, Operations & Maintenance of Electric Power Plants

and combined-cycle power generating units throughout the world. Experience from long-term laboratory testing and in-service behavior suggests that the performance of components manufactured from CSEF steels is likely to depend on creep damage in welds. Grade 91 and other CSEF alloys have a martensitic lath structure that does not lend itself to the EPRI-Neubauer correlation. Of particular concern is the fact that relatively high densities of relatively small creep voids have been shown to develop below the component surface. These voids are, therefore, not easy to detect. Because many voids can be present through much of the component wall thickness, the processes of crack formation and growth can be relatively rapid. Thus, the window when traditional methods of nondestructive examination (NDE) can identify macroscopic defects is relatively small. This time limit means that inspectors need to determine the size of defect using established methods of inspection, to refine existing methods, and if necessary, to consider new approaches to improve the reliability and sensitivity of detection. Additionally, damage in CSEF alloys is almost always greater subsurface than on the surface of components. Because of the differences from low-alloy steels in microstructure and damage development, and the prevalence of subsurface damage, EPRI does not recommend applying the EPRI-Neubauer correlation to this class of materials. To provide a method to predict remaining life in this class of components, EPRI developed a program that includes prediction of remaining life, using highly sensitive NDE, an EPRIdeveloped calculator for material life, and specialized monitoring techniques. The following section describes current research in the first of these areas — NDE tools.

Current research: NDE tools A current three-year EPRI project is evaluating the potential of available NDE tools to determine the damage level in CSEF alloys and to identify the need for future development of new tools. Project sponsors and participants include utilities, NDE vendors and engineering consultants. The project is utilizing current state-of-the-art technologies and working with universities to develop new NDE techniques. To facilitate this evaluation, EPRI prepared large plates using typical submerged arc welding (SAW) and shielded metal arc welding (SMAW) welding techniques. From these plates, the participants prepared very large creep coupons, measuring up to 48˝ x 1.5˝ x 2˝. The coupons were instrumented with acoustic emission (AE) equipment from various vendors and tested to failure while measuring the strain accumulation throughout the test (Figure 2). The continuous strain monitoring allows for better correlation of remaining life to damage during the test. Further coupons were tested, but were interrupted at certain life consumption intervals. The interrupted coupons were then tested with various state-of-the-art NDE processes, including digital radiography, phased array and electromagnetic techniques. Many different transducers and frequencies were

September 2014

Maintenance Matters evaluated to identify the best techniques for detecting damage, which led to many of the project participants modifying their field inspection techniques. Replication was performed on the top and sides of the coupons. Although the replicas did not show any damage on the top surface (what would be the outside surface of a pipe), the replicas did show damage on the sides of the coupons. (However, a lesser amount of damage was found than in the center of the samples during the destructive laboratory analysis.) Results can be found in the EPRI report, Review of Weld Repair Options for Grade 91, Part 2: Damage Development and Distribution (3002000087), published in 2013. Following NDE, the samples were sectioned and prepared for laboratory analysis. Due to the extremely large creep samples, the project team had the opportunity to remove multiple metallurgical samples from each coupon. Guided by the NDE results, the team prepared the coupons to demonstrate damage of interest. This process enabled the project participants to correlate their NDE signals to damage level.

Current research: Fitness for service Another area of interest is determining if new CSEF material is fit for service. CSEF material might be degraded through processes utilized during fabrication, specifically heat treatment and cold forming. EPRI is partnering with a U.K. university in a parallel effort to see if electromagnetic methods might be able to detect microstructural changes in CSEF materials. In phase I of the project, coupons of Grade 91 material were first prepared by heat-treating specimens at various times and temperatures selected to produce different levels of microstructural degradation. The coupons were then submitted to the university to sort them into a ranked damage order. The results were very promising, and the project is continuing with phase II, where more controlled specimens were created, including new tubing that was heat-treated to various damage levels. Normal temper levels and samples that have exceeded the transformation temperature were prepared to expand the research to see if grain structure and precipitate structure information might be determined. Much information has already been developed about the relationship between NDE results, damage development, and remaining life with these alloys. The project is currently in the second of three years and will be completed next year. Look for a final report detailing the results and recommendations in the second half of 2014. ~ Kent Coleman manages EPRI’s Boiler Life and Availability Improvement Program. He has been a member of EPRI’s Generation staff for 15 years after a 17-year utility background and has an extensive background in the materials, life assessment and welding areas, and holds several patents in the areas of boiler materials, welding and repair. He also is a member of several ASME Code committees including SCI, Power Boilers. You may contact him by emailing editorial@woodwardbizmedia.com.

FROM ENGINEERING ANALYSIS TO ADVANCED NDE, LOOK TO STRUCTURAL INTEGRITY

We Connect the Dots Over the past 30 years, Structural Integrity has built a team of over 200 industry experts providing comprehensive solutions to the energy industry. We connect the dots from problem to resolution, from regulations to compliance, from nondestructive examination, to engineered solutions, even custom equipment. You can look to us for our: • Knowledge of power plants, codes, and how things work. • Extensive experience and leadership. • High quality, hard work, and responsiveness.

Call us today and we’ll connect the dots for you.

Scan the QR Code for more information

877-474-7693 (877-4SI-POWER) www.structint.com/energy-tech

September 2014 ENERGY-TECH.com

ASME FEATURE

Quantifying correction curve uncertainty through empirical methods By Christopher R. Bañares, Thomas P. Schmitt, Evan E. Daigle and Thomas P. Winterberger, General Electric Power & Water

Introduction The accuracy of a thermal performance test is typically estimated by performing an uncertainty analysis calculation in accordance with ASME PTC 19.1, or another code equivalent to it. The test uncertainty is a measure of the test quality and, in many circumstances, the test setup must be designed so that the uncertainty remains lower than test code limits and/or commercial tolerances. Traditional uncertainty calculations have only included an estimate of the measurement uncertainties and the propagation of those uncertainties to the test result. However, in addition to addressing measurement uncertainties, ASME PTC 19.1 makes reference to other potential errors of method, such as “the assumptions or constants contained in the calculation routines” and “using an empirically Figure 1. Correction error due to performance variation. derived correlation.” are as close to the specified reference conditions as possible. Performance correction curves are utiFurther, the codes state that there is an acceptable error due lized to correct performance to a specified set of reference to correction methodology of approximately 0.2-0.3 percent. conditions so that the corrected result is independent However the codes do not provide specific guidance on of boundary conditions that persist during the perhow to estimate the incremental uncertainty levels associated formance test. Many of the ASME performance with the correction methodology, and/or how to account for test codes (PTC-22 Sections 3-3.1 and 5-5, them either in the overall test uncertainty or in the corrected PTC-46 Sections 3.4.2.4 and 5.4, ASME results. This is important, since experience shows that there is PTC19.1 Section 5-3.5, ASME PTC6.2 potential for errors in corrections to exceed 0.3 percent. This Section 3-5.3., and ISO2314 Section can happen in cases where insufficient data exist to validate 7.1) recognize the potential for the thermodynamic models used to develop the correction errors in the corrected result curves at extreme off-rated conditions, or as a result of the due to correction methodnormal unit-to-unit variation and its associated impact on ology, and guide the user estimating the response of the equipment to changes in the to make efforts to test boundary conditions. in conditions that Failure to properly For the case where empirical validation via a large data set is either not possible or impractical, then when developing ground rotating the pretest uncertainty analysis, it might be technically valid equipment can result in for the manufacturer to include a line item in the analysis expensive bearing, seal, & that represents an estimate for correction curve uncertainty

Are Shaft Currents Destroying Your Machinery?

gear damage.

SOHRE TURBOMACHINERY® INC. 128 Main Street, P.O. Box 1099 Monson Massachusetts, USA 01057-1099 PH: 413.267.0590 • 800.207.2195 • FX: 413.267.0592 18 ENERGY-TECH.com tsohre@sohreturbo.com • www.sohreturbo.com

ASME Power Division Special Section | September 2014

ASME FEATURE based on past experience and the extent to which the test condition might deviate from the rated condition.

at the reference condition, and the performance calculated by correcting off-design test data back to the reference condition without accounting for the unit-specific component behavior. By using the correction process described previously, error sources due to the use of correction curves were eliminated, with the only remaining effect being the difference between typical unit-to-unit variations in component performance and the model predictions used to generate the correction factor.

Case study 1 – Gas turbine unit-to-unit variation Field test data for new and clean gas turbines (n = 57) of the same hardware configuration were used to calculate key performance metrics for the major components of the units, such as compressor efficiency and nozzle flow characteristics. Multiplicative factors were derived to adjust the unit nominal thermodynamic model to match the observed component performance for each unit, factoring out the effects of boundary conditions like ambient temperature and inlet pressure loss. This results in a set of factors whose distribution approximates the unit-to-unit variation in component performance for the fleet relative to the model, independent of test boundary conditions. Statistical analysis of the variation in these factors resulted in a model of typical fleet component variations. This model of fleet variation was combined with a model of variation of ambient conditions to create a “fleet” Outstanding corrosion protection in of models (n=500) with identical controls a workhorse design settings, each with unique thermodynamic models, test conditions and predicted performance mimicking that of the fleet. A multiplicative correction factor was calculated for each of the model runs by taking the ratio of the predicted performance at the test conditions and the predicted performance at a common reference These scotch yoke actuators provide a condition using the nominal (un-modified) solution for applications where steel model, in much the same way that convenactuators are required, due to corrosion tional corrections are generated and applied. protection requirements that aluminum Model predictions were used to eliminate actuators do not adequately handle . error sources inherent in the usage of Available in both Double Acting and Spring curves, such as curve-fitting error and interReturn, the quarter-turn actuators cover a dependencies between corrections. For each broad range of torques from 800 - 12,000 case, the correction factor was then applied in.-lbs. torque. The actuator shown is to the “observed” model performance, so as featured on the A-T Controls Power Seal to calculate corrected performance at the HPBV, but is also applicable for ball and plug reference condition in a manner consistent valves and dampers. with that used in performance testing. When your application calls for rugged and Each model in the “fleet” was then run reliable valve actuation, rely on the TRIAC SY a second time, at the reference conditions, Series. to determine what the true performance In stock from A-T Controls. of that unit would have been at the reference conditions. The result at the reference condition was then compared against the same model’s performance at the test condition, with the correction factor 9955 International Boulevard applied. The difference between these two Cincinnati, Ohio 45246 (513) 247-5465 results therefore represents the difference FAX (513) 247-5462 between the true performance of the unit e-mail: sales@atcontrols.com

SY Series Steel Scotch Yoke Actuators

www.atcontrols.com

September 2014 | ASME Power Division Special Section

ENERGY-TECH.com

ASME FEATURE

Figure 2. Estimated correction uncertainty as a function of variation in ambient temperature.

Figure 3. Incremental test uncertainty due to correction error.

Figure 1 illustrates the variation between the theoretical unit performance at reference conditions and that calculated using the model-based correction factor. The highlighted section in the center of the graph represents the typical code limit uncertainty band for a PTC-22 test, for reference (approximately +/- 0.5 percent). For cases where the difference between the as-tested ambient temperature and the reference ambient temperature are small, the correction approaches unity and, as one would expect, the error it introduces becomes negligible. However, as the test condition begins to vary from the reference condition, the influence of unit-to-unit variations becomes more pronounced. For variations greater than 30°F from the reference condition, the influence of this effect eclipses that of the combined measurement uncertainty, significantly reducing the accuracy of the test result. In other words, the error illustrated in Figure 1 represents solely the incremental error (i.e. uncertainty) contribution from natural unit-to-unit thermodynamic variations in the equipment components, and does not include the traditional errors associated with measurement uncertainty. The extent to which the uncertainty intervals need to expand in proportion to the extent to which the test conditions deviate from the reference conditions is analogous to the manner in which most instrument manufacturers define the nominal uncertainty (or accuracy) of an instrument to be applicable within a specified range of conditions (such as temperature or humidity), and give a formula the user can use to estimate the incremental additional uncertainty when usage conditions exceed the nominal range. Similarly, in the case of a power plant thermal performance test, the uncertainty (or accuracy) accounting needs to take into consideration the incremental additional uncertainty attributable to testing at off-reference conditions, inclusive of correction curve interdependencies and unit-to-unit component variations. To statistically quantify this effect, the same process was used to correct the “fleet” of 500 units to a reference condition of 59°F from varying ambient temperatures in 10°F increments. The resulting variation in corrected performance from each as-tested ambient temperature was analyzed, and a symmetrical 95 percent tolerance interval was calculated as an estimate of the uncertainty contribution from the correction. Figure 2 illus-

Figure 4. Example of correction error from off-frequency test conditions.

ENERGY-TECH.com

ASME Power Division Special Section | September 2014

ASME FEATURE trates the estimated effect of variation in as-tested ambient temperature on correction uncertainty. By combining this uncertainty estimate with that we obtain from propagation of measurement uncertainty, a combined test uncertainty may be calculated. Figure 3 illustrates the post-test uncertainty for a typical PTC-22 test, 0.4 percent, and the combined uncertainty when the correction uncertainty estimate is incorporated. The shaded area indicates the increase in test uncertainty as a result of the correction curve error. As prior industry experience suggests, the incremental uncertainty is very small within a certain range of test conditions, but increases dramatically as the test condition deviates more significantly from the reference condition. Readers should note that this analysis does not address uncertainty inherent in measuring the component performance factors as described at the beginning of this case study. As a result, this analysis represents an approximation of the magnitude of these effects. It is the authorsâ&#x20AC;&#x2122; assertion that the conclusion of this case study highlights the existence of this error source, its basic characteristics, and a statistically-based method of estimating the incremental uncertainty.

and a second time without the frequency correction (the red data). See Figure 4. As shown in Figure 4, the corrected performance at significant under-frequency test conditions tended to be overstated (the blue data is after using the expected off-frequency correction). The frequency correction was then empirically adjusted with consensus of all parties involved, and the statistical data set was used as validation of the empirical adjustments. This case study exemplified the occasional need of empirically adjusting the test correction curves, when warranted by the data, to avoid an unnecessary avoidable incremental systematic error from correction curves.

Case study #2 â&#x20AC;&#x201C; Gas turbine: Extreme off-rated conditions There can be times when an additional systematic error is introduced into the corrected test results when the actual turbine response to one or more correction variables deviates considerably from the expected response. And similar to the random errors discussed previously, these systematic errors can grow in magnitude in proportion to the extent of off-rated conditions found. Recent GE test experience in Pakistan on several heavy duty gas turbines resulted in a data set that was used to empirically adjust the frequency correction curves. The official tests were run at high ambient and low frequency conditions, and the corrected results were noted to be higher than expected. A database of archived GT performance behavior was then downloaded from the plant historian so that a statistically significant amount of data could be studied across the frequency range. This data was then processed twice, once through the full set of test corrections, including the frequency correction curve (the blue data),

September 2014 | ASME Power Division Special Section

ENERGY-TECH.com

ASME FEATURE

Figure 5. Example of correction error isolated to a specific range of the correction variable.

Case study #3 – Gas turbine: Correction errors in an isolated range Another scenario that can occur is the introduction of a systematic error in the corrected result only in a particular range of a correction variable. In these cases, the turbine’s

What’s the word on the wire?

Follow Energy-Tech and keep up with the industry chatter!

@ETmag

Dedicated to the Engineering, Operations & Maintenance of Electric Power Plants

operating behavior for a given boundary condition might be characterized fairly well by the thermodynamic model in one range of conditions and deviate in other parts of the range. Since the thermodynamic models are used to create the correction curves, errors would be confined only to a particular range of the correction curves. This scenario was experienced during testing of a newer model heavy duty gas turbine. It was observed during preliminary testing of the turbine that the corrected output appeared to be overstated when test runs were made at conditions below the reference value of one of the boundary parameters. This prompted further analysis using the historical archived data system of a unit of similar configuration to gather a statistically significant data set with a larger range of the correction variable of interest. Correction curves were applied to the data to obtain corrected output performance and plotted against the range of the variable of interest. The results can be seen in Figure 5. These data showed that the actual turbine performance was significantly better than expected in the low range of this correction variable. If no error in correction curves existed, the error over the range of the variable would be centered on zero. This can be seen in the range of correction variable above the reference value (approximately 0-2 percent). However, in the correction variable range below the reference, the performance of the turbine appeared to deviate from expected, resulting in significant positive errors in the corrected result when using the theoretical pre-test correction curves. Similar to the previous case studies, this error increased in proportion to deviation from the reference condition.

Case study #4 – Steam turbine: Extreme offrated conditions ASME PTC 6 and PTC 6.2 stress the importance of testing as close to specified conditions as possible to minimize the magnitude of corrections and error introduced by the correction methodology. Table 3-1 of PTC 6 and Table 3-1.3.5 of PTC 6.2 list the allowable deviations between test and rated conditions. One variable with a large sensitivity to output and a greater potential for deviation is exhaust pressure. While PTC 6 and 6.2 list different requirements on exhaust pressure, the allowable deviations range from 0.1˝ to 0.5˝ of mercury absolute. Figure 6 shows an exhaust pressure correction curve for a large steam unit. The solid line is the response predicted by the steam turbine manufacturer’s heat balance modeling program, while the dashed line is based on plant data measured during an operating period with controlled conditions. While the two curves show good agreement in close proximity to rated conditions, the curves differ significantly at the outer boundaries, illustrating the need to adhere to code require-

ENERGY-TECH.com

ASME Power Division Special Section | September 2014

AsME FEAtUrE ments. While these differences could be due to a number of reasons, including plant operation and turbine design, the best course of action is to avoid these regions when testing. Historically, uncertainty estimates have not been increased to account for greater deviations in exhaust pressure. For these situations, consideration should be given to modifying plant operation to change exhaust pressure, waiting until seasonal conditions are more favorable, or verifying the exhaust pressure correction curve through additional testing. When none of these means are practical, PTC 6.2 indicates that testing might be conducted while accounting for the additional uncertainty in the uncertainty analysis.

Conclusion As noted in each major industry test code, efforts should be made by all parties to conduct the performance test at conditions as close as possible to the reference conditions. In practice, owing to natural effects and commercial requirements it is not always possible, nor is it always practical, to conduct the test at conditions that would yield the lowest achievable uncertainty. It is the responsibility of the testing organization to estimate the test uncertainty as accurately as possible, taking into account all known contributing factors. As discussed herein, the test uncertainty should take into account not only the contributions from measurement errors and their propagation to the corrected results, but also any additional uncertainty contributions that might result from the correction curves or calculation methodology. As shown in this paper, these additional incremental uncertainty contributions from correction curves or calculation methodology can be significant and quantifiable. Statistical means can be employed to estimate the uncertainty stemming from the combination of unit-to-unit variations and deviations from the reference conditions. As shown herein, these can easily contribute an additional 0.2 percent to 0.3 percent incremental test uncertainty. As noted previously, additional analyses are warranted to refine these estimates to consider measurement uncertainty effects on the unit-to-unit variations, to ensure contributions are not overstated when applied to the overall test uncertainty. Furthermore, additional test uncertainty can result from limitations in correction curves when considering interaction with control system limits. These errors can be reduced by use of model-based performance corrections (ASME POWER2014-32184), though industry acceptance and proliferation of this methodology has been historically limited. When a statistically significant data set exists (which is now more common with modern plant historians and industrial Internet capabilities) the equipment supplier can empirically adjust correction curves (or the thermodynamic model) to mitigate the impact of off-reference test conditions, which could otherwise contribute >1 percent incremental test uncertainty. While a traditional test uncertainty estimate for a PTC-22 code test that only considers measurement uncertainty might

September 2014 | ASME Power Division Special Section

ENERGY-TECH U NIVERSITY September Webinars Sign up today for these upcoming FREE exclusive webinars from Energy-Tech!

Electric Process Heating & Demand Fuel Switching Tuesday, Sept. 16 • 1 pm CDT Presented by Gaurav Dhingra – VP Engineered Systems, Craig Tiras – VP Technology Sign up at www.energy-tech.com

Power Generating Asset Management Tuesday, Sept. 30 • 1 pm CDT Presented by Komandur Sunder Raj Power generating assets represent sizeable investments for power plant owners. With the power industry grappling with transformative changes relating to people, processes and technology, power generating asset management (PGAM) is becoming vitally important, even critical in some cases. This article/webinar reviews technological advances that may be used in developing and implementing an optimum PGAM program and discusses technological tools to effectively gather and use key data and performance indicators to assess health and optimize the performance of power generating assets on a continuous basis. Sign up at www.energy-tech.com/article.cfm?id=75886

ENERGY-TECH.com

ASME FEATURE the benefit of all test parties to take these incremental uncertainties into consideration when defining the test procedure (including correction methodology) and in selecting test conditions as close as possible to the reference conditions. ~

References 1. ASME PTC 22–2005 Gas Turbines 2. ASME PTC 46–1996 Performance Test Code on Overall Plant Performance 3. ASME PTC 19.1-2005 Test Uncertainty 4. ASME PTC 6.2–2011 Steam Turbines in Combined Cycles 5. ISO 2314 2009 Gas Turbines – Acceptance Tests Figure 6. Steam output sensitivity to exhaust pressure.

yield an uncertainty estimate on the order of +/- 0.5 percent, the actual error of the test result might be well above +/- 1.0 percent due to errors in the corrections. Proactively recognizing and considering these error sources can improve test accuracy, thereby reducing the risk of understating or overstating the true equipment performance. As such, it is to

exclusive

Christopher Bañares is a senior technical manager for GE Power & Water with 18 years of experience in gas turbine design and performance testing and analysis. He has a bachelor’s degree and master’s degree in Mechanical Engineering from Rensselaer Polytechnic Institute. He is responsible for managing the Gas Turbine Performance Test group. You may contact him by emailing editorial@woodwardbizmedia.com. Evan Daigle is a lead methods engineer for GE Power & Water with 8 years of experience in gas turbine testing. In his role he is responsible for development of new testing and analysis methods. He has a master’s degree from the University of Michigan. You may contact him by emailing editorial@woodwardbizmedia.com.

webinars from

Energy-Tech Archived webinars

Our technical webinars are free

Gas Turbine Inlet Cooling

and feature industry experts

Feedwater Heater Level Control Makeup Water

presenting the most relevant

Performance & Reliability

subject matter as it relates to

Condenser Retrofits Hydrogen Safety

electric power generation. Now

Turbine Outage Optimization

archived on www.energy-tech.

Equipment Reliability Programs

contentshelf.com/shop!

Turbine Testing with Cycle Isolation

Thomas P Schmitt is a senior technical manager for GE Power & Water with 29 years of experience in jet engines, gas turbines and combined-cycle power plants. He holds a bachelor’s degree from Michigan State University. He is responsible for power plant performance testing, analysis and methods. You may contact him by emailing editorial@woodwardbizmedia.com. Thomas P Winterberger is a senior technical manager for GE Power & Water with 25 years of experience in steam turbine design, analysis and testing. He has a bachelor’s degree from Clarkson University and an MSME from Union College. He is responsible for managing the Steam Turbine Performance Test group. You may contact him by emailing editorial@woodwardbizmedia.com.

Steam Generation Chemistry Power Generating Assets Steam Turbine Failure Feedwater Heater Troubleshooting Maximizing Condenser Tube Life

ENERGY-TECH.com

ASME Power Division Special Section | September 2014

MACHINE DOCTOR

Compressor vibration due to bearing and seal problems By Patrick J. Smith

The causes of turbomachinery damage are not always obvious. This article presents a case study of a compressor that had a history of minor bearing temperature issues that preceded several bearing failures. The true causes of the bearing damage were not immediately obvious and it turned out to be a problem with tolerances of certain parts and an undetected seal failure. An assessment of the machinery protection system also will be discussed.

Introduction This case study pertains to a 5-stage integrally geared centrifugal compressor driv- Figure 1. Compressor configuration en by a 1,500 rpm, 12,000 hp synchronous motor. The compressor is directly connected to the motor through a flexible disc pack type coupling. This machine compresses dry air from approximately 4.8 barg to 63 barg. The gearbox consists of a bullgear and three pinions. The stage 1/2 and stage 3/4 rotors consist of pinions with overhung impellers mounted at both ends. These rotors are mounted at the horizontal split line. The stage 5 rotor is located in the top of the gearbox cover and consists of a pinion with a single overhung impeller mounted at one end. The stage 1/2 rotor operates at a speed of 16,922 rpm, stage 2/3 at 22,562 rpm, and stage 5 at 25,786 rpm. The gearbox utilizes tilting pad journal bearings for all three pinions. “X” and “Y” non-contacting proximity type shaft vibration probes are adjacent to each bearing, between the bearing and the oil seal. The pinions are fitted with thrust collars, which transmit pinion axial thrust to the bullgear. The bullgear rotor is fitted with a sleeve type journal bearing on the drive end and a combined sleeve type journal bearing and tapered land thrust bearings on the non-drive end. The bearings are all instrumented with temperature probes. The compressor control system includes vibration alarm and shutdown protection, as well as alarm only bearing temperature monitoring. The pinion air seals are a carbon ring type. Each seal assembly consists of multiple self-adjusting one piece carbon rings, which have a close clearance to the shaft to minimize process

gas leakage. The gearbox arrangements for stages 1-4 is shown in Figure 1; stage 5 is omitted for clarity.

History This compressor was commissioned in March 2007. On the very first start, the 4th stage bearing temperature quickly climbed to 149°C before the compressor was shut down. The bearing was inspected and there was no visible damage. The compressor had been started with an oil supply temperature of 33°C vs. a design temperature of 45°C. Although the oil was cooler than design, it was above the recommended minimum permissible temperature. When the oil temperature was increased to the design operating temperature, the 4th stage bearing temperature quickly climbed to about 130°C after start-up, but eventually dropped and settled out at about 120°C. Although the 3rd stage bearing temperature did not rapidly increase in temperature or overshoot immediately after start-up, the steady state temperature during normal operation also settled out around 120°C. Based on discussions with the compressor manufacturer and the author’s experiences with a similar compressor, the oil supply temperature was raised to 55°C to help increase oil flow to the 3rd and 4th stage bearings. When this was done, the 3rd stage steady stage temperature dropped to 113°C and the 4th stage dropped to 102°C. Although the 3rd stage temperature

September 2014 ENERGY-TECH.com

MACHINE DOCTOR

Figures 2a and 2b. September 2013 bearing damage

was above the recommended alarm temperature of 110°C, it was decided that this was acceptable for short term operation. The bearing temperatures on the other stages were acceptable. The vibration levels on all stages also were acceptable. Despite the higher bearing temperatures, both 3rd stage “x” and “y” vibrations were around 0.8 mils peak to peak (p-p) and both 4th stage “x” and “y” vibrations were around 0.4 mils p-p. The compressor manufacturer’s high vibration alarm set point for these stages was 1.1 mils p-p and the trip set point was 1.5 mils p-p. The bearing application, design and assembly records were reviewed with the compressor manufacturer. The bearings were

ENERGY-TECH.com

a 5-pad, tilting pad type with non-aligning, cylindrical, center pivot pads. The pads were made from steel with a Babbitt coating. The 3rd and 4th stage bearings are the same, except that the bearings were orientated differently. Some basic bearing information is shown in Table 1. Although the bearing journal speed and bearing unit load are within typical bearing manufacturer limits, these are at the upper end of the author’s experience for a typical steel backed tilting pad bearing. The author’s experience would suggest that the bearing application is a slightly more aggressive application and the bearings could be less tolerant to certain conditions, which could adversely affect performance. For example, bearing clearance slightly tighter than design is an example of a condition that could adversely affect bearing performance by causing higher bearing temperatures. The compressor manufacturer evaluated various bearing modifications to reduce the bearing temperature and proposed the following changes: • Machine a chamfer on the trailing edge of all the pads. This essentially changes the bearing to an offset pivot type, which increases oil film thickness and should result in lower bearing temperatures. • Increase the diameter of the oil nozzles to increase the oil flow into the bearing. Machine the pinion journal diameter in the bearing area to increase the bearing clearance. In general, higher bearing clearances result in lower bearing temperatures, but also can cause slightly higher rotor vibrations. The compressor manufacturer preferred to modify the journal rather than modifying the bearing to increase bearing clearance.

September 2014

MACHINE DOCTOR Table 1 – Basic Bearing Information Stage

Pinion Speed, RPM

Journal Dia., mm

Brg Axial Length, mm

Journal Speed, m/sec

Bearing Unit Load, barg

3/4

22562

94.5

23.7

It was decided to pursue the first two recommendations, but not the bearing journal modifications because of the additional time to remove, modify and re-install the rotor. It also was decided that only the 3rd stage bearing would be modified. The 4th stage bearing would not be modified because it was already operating at an acceptable temperature. The compressor was operated in continuous service until a planned shutdown in July 2009 provided the opportunity to install the modified 3rd stage bearing. Despite operating at elevated temperatures above the recommended compressor manufacturer’s alarm point, the 3rd stage bearing temperature and vibration had been stable since commissioning. When the bearing was removed there was no visible damage. After restarting the compressor, the 3rd stage bearing temperature was reduced to around 68°C. The “x” and “y” vibrations increased slightly to around 0.9 and 1.0 mils p-p. The 4th stage steady state bearing temperature remained about 100°C and the 4th stage “x” and “y” vibration levels were still both 0.4 mils p-p. In February 2013 the 4th stage “x” and “y” vibrations increased to about 0.7 and 0.5 mils p-p and the bearing temperature increased to about 110°C following a compressor shutdown and restart. The “x” and “y” vibrations slowly trended up to about 1.0 and 0.9 mils p-p and the bearing temperature slowly increased over a matter of months to about 116°C. A sudden spike in 4th stage vibration caused a compressor trip in September 2013. There were no changes in the 3rd stage bearing temperature or 3rd stage vibrations. A process upset, which caused significant changes in compressor flow and discharge pressure preceded the September 2013 trip. It was thought that this upset might have caused an operating instability, which then caused the sudden increase in vibration. However, since there had been an upward trend in vibrations and bearing temperature, it was decided to inspect the 4th stage bearing. When the bearing was removed, there was some bearing Babbitt damage found on

the 4th stage bearing pads. See Figure 2. No other damage was found and the pads were replaced with spares. The measured diametral bearing clearance also was checked and found to be slightly lower than design; 0.150 mm vs. a design of 0.179 to

Celebrating 50 years in North America

September 2014 ENERGY-TECH.com

MACHINE DOCTOR

Figures 3a and 3b. October 2013 bearing damage

• • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • •

• • • SELL • RENT• LEASE • - 24 / 7 • EMERGENCY SERVICE • • • • • • IMMEDIATE DELIVERY • CALL: 800-704-2002 •• 10HP TO 250,000#/hr • • • • • • • • RENTAL FLEET OF MOBILE • TRAILER-MOUNTED BOILERS • • • • • ALL BOILERS ARE COMBINATION GAS/OIL ENGINEERING • START-UP • FULL LINE OF BOILER • AUXILIARY SUPPORT EQUIPMENT. • Electric Generators: 50KW-30,000KW • WEB SITE: www.wabashpower.com • 847-541-5600 • FAX: 847-541-1279 • E-mail: info@wabashpower.com • wabash • •

BOILERS

250,000#/hr Nebraska 750 psig 750OTTF 150,000#/hr Nebraska 1025 psig 900OTTF 150,000#/hr Nebraska 750 psig 750OTTF 150,000#/hr Nebraska 350 psig 115,000#/hr Nebraska 350 psig 80,000#/hr Nebraska 750 psig 75,000#/hr Nebraska 350 psig 60,000#/hr Nebraska 350 psig 40,000#/hr Nebraska 350 psig 20,000#/hr Erie City 200 psig 10-1000HP Firetube 15-600 psig ALL PRESSURE AND TEMPERATURE COMBINATIONS SUPERHEATED AND SATURATED

75,000#/hr 75,000#/hr 60,000#/hr 50,000#/hr 40,000#/hr 30,000#/hr 75-300HP

Optimus Nebraska Nebraska Nebraska Nebraska Nebraska Firetube

750 psig 350 psig 350 psig 500 psig 350 psig 350 psig 15-600 psig

750OTTF

POWER EQUIPMENT CO.

0.211 mm. Based on field measurements, it appeared that the shaft journal was slightly larger than design and this was the cause of the tight bearing clearance. The compressor was restarted and the 4th stage “x” and “y” vibrations stabilized at about 0.6 and 0.5 mils p-p. But the vibrations slowly trended up to 0.9 and 0.8 mils p-p before the compressor tripped again on high 4th stage vibration a month later in October 2013. The bearing temperature after the repair was 100°C and had increased to about 113°C when the compressor tripped in October. Again, bearing pad Babbitt damage was found. See Figure 3. Spare bearing pads were modified by adding the chamfer to the trailing edge, as was previously done to the 3rd stage bearing in 2009. The 4th stage bearing nozzles also were drilled out, as was also done to the 3rd stage bearing in 2009. The bearing clearance was measured again and was once more found to be lower than design. Modifications to increase bearing clearance were not pursued due to time constraints. Although tight bearing clearance might have contributed to the higher bearing temperatures, it was felt that something else was probably causing the poor bearing performance and progressive damage. Although there was no change in the compressor thermodynamic performance since commissioning nor were there any signs of excessive seal leakage, it was decided to inspect the 4th stage impeller and seals for any sign of wear or damage. Upon disassembly, there was no visible impeller wear or damage, but the 4th stage carbon ring air seals were found damaged. It appeared that the several rings were broken and were hung up in the seal carrier. These seal rings are supposed to have some float so that they can tolerate minor contact with the shaft without causing a continuous rub. With some of the seal rings stuck in position, there could have been a continuous seal rub, which then led to higher rotor vibration and subsequent bearing damage. The seal rings were replaced and the modified bearing was installed. The compressor was restarted and the 4th stage bearing temperature settled out at 92°C and the “x” and “y” vibrations settled out at 0.5 and 0.5 mils p-p. There has been no change in this performance since the repair.

444 Carpenter Avenue, Wheeling, IL 60090

ENERGY-TECH.com

September 2014

MACHINE DOCTOR However, some future minor bearing changes are still planned so that the oil supply temperature can be reduced to the design temperature of 45°C without causing any increase in vibration levels.

Bearing performance analysis Detailed bearing analyses performed following the two failures predicted bearing temperatures at the tighter bearing clearances. Predicted bearing performance was better at design bearing clearance. The bearing pad and bearing orifice modifications were successful in lowering the bearing temperature even at the tighter bearing clearance. However, restoring the design bearing clearance and other modifications are currently being pursued with support from the compressor manufacturer. Machinery protection system discussion Looking closely at the September 2013 shutdown trends, the 4th stage “x” vibration gradually increased to about 1.0 mils during several months and then rapidly increased to 1.6 mils p-p in less than a minute, which caused the compressor trip. The “y” vibration also slowly increased to about 0.9 mils during the same time period, and then rapidly increased in vibration to 1.4 mils p-p in less than 30 seconds at the time of the trip. There was no corresponding spike in bearing temperature at the time of the trip. Looking at the Oct. 23, 2012, shutdown, the “x” vibration slowly increased from 0.6 mils p-p to about 0.9 over a month and then rapidly increased to 1.6 mils p-p in less than 30 seconds, which again caused a compressor trip. Similarly, the “y vibration slowly increased from 0.5 mils p-p to 0.8 mils during the month time period and then rapidly increased to 1.4 mils p-p in less than 30 seconds at the time of the trip. And again there was no corresponding spike in bearing temperature at the time of the trip. Having compressor vibration monitoring and protection does not prevent a problem such as internal damage, wear or fouling from occurring. It does provide information of a progressive problem that allows operators to take action before

Table 2 – Compressor Arrangement Vibration Stage

Before Field Balancing

After Field Balancing

MAC Stage 1

0.60

0.62

MAC Stage 2

0.64

0.50

BAC Stage 1

0.49

0.57

BAC Stage 2

0.99

0.74

there is a significant damage or a sudden failure. It also can minimize collateral damage in the event of a sudden failure. Although there were signs of some progressive wear/damage, as seen by gradual increases in vibration and temperature with both failures described in this article, the vibration levels never operated in a sustained period in alarm prior to the high vibration trips. The large vibration increase occurred very quickly and the vibration protection system was effective in minimizing collateral damage. Not all process centrifugal compressors are fitted with “x” and “y” vibration probes adjacent to each compressor bearing. For example, some machines might have only a single vibration probe adjacent to each bearing. Even with machines fitted with “x” and “y” probes adjacent to each bearing; some compressor protection systems are configured such that a high vibration trip on either vibration signal is needed to trip a compressor, while others are configured such that both vibration signals have to reach the trip point before the compressor trips. So, one might ask why this is the case and what is really needed to protect a compressor. Bearing dynamic characteristics such as damping and stiffness can be non-symmetrical, which can cause a rotor response to be more sensitive in one direction than another. In other words, it is possible that a problem might mainly affect the rotor vibration in one plane and have a modest or no affect in

September 2014 ENERGY-TECH.com

MACHiNE doCtor sEPtEMBEr 2014 AdVErtisErs’ iNdEX A-T Controls Inc.

www.a-tcontrols.com

CU Services

www.cuservices.net

Cutsforth, Inc.

www.cutsforth.com

EagleBurgmann

www.eagleburgmann-ej.com

www.ecomusa.com

www.floscan.com

ECOM America Ltd. FloScan Instrument Co. Inc. Gaumer Process Gradient Lens Corp. Hexeco Hurst Boiler Indeck Power Equipment Co.

www.gaumer.com

www.gradientlens.com

www.hexeco.com

www.hurstboiler.com

www.indeck.com

www.miller-stephenson.com

www.mtiinstruments.com

www.renewalparts.com

www.rotork.com

www.schenck-usa.com

www.sohreturbo.com

Structural Integrity Associates, Inc.

www.structinc.com

Topog-E Gasket Co.

www.topog-e.com

www.wabashpower.com

Miller-Stephenson Chemical MTI Instruments Renewal Parts Maintenance Rotork Controls Inc.

Schenck Balancing & Diagnostic Sohre Turbomachinery Inc.

Wabash Power Equipment

Coming in October 2014 Look for the next issue of Energy-Tech magazine and read about: • Outage season • Pumps • Regulations compliance: Ash handling • Turbines: Steam • ASME: Performance

ENERGY-TECH.com

another plane. This is a reason for having vibration measured in two planes and having high vibration trip logic that shuts down a machine when either the “x” or “y” vibration signal exceeds the trip set point. This sounds like a reasonable approach, but it can lead to more nuisance trips due to instrument issues than a machine fitted with a single vibration probe, due to the number of extra instruments. Even with a machine fitted with dual probes adjacent to each bearing, having the control system configured such that the vibration signal from either probe can cause a trip can cause more nuisance trips than a control system configured such that both vibration signals have to reach the trip point before the machine trips. However, having dual probes also can add flexibility. If a problem develops with one probe, having a second probe still provides some protection and monitoring capability, but this also adds cost. The problem described in this article caused a similar vibration response in both planes during both trip events. This is consistent with other incidents the writer has reviewed with similar machinery. It doesn’t mean that this is always the case, but compressor operators need to evaluate the impact of additional instrumentation vs. the benefit of additional diagnostic information.

Conclusions Although tight bearing clearance probably contributed to the higher bearing temperatures on both stages 3 and 4, it is likely that the seal ring damage was the catalyst that led to the bearing failures on stage 4. Although the high bearing temperatures were mitigated by modifications to the pads even with the tight bearing clearances, restoring the bearing clearances and further modifications could make the bearing more tolerant to less than ideal conditions. Operators and compressor manufacturers need to collaborate on the machinery protection system that meets both the compressor manufacture requirements and the end user needs. This includes both the needed hardware and the control system configuration. It might be different for the same machine in different applications, depending on the cost and consequences of a machine trip, relative to the cost of the machine and any safety implications if the machinery shutdown protection is not sufficient to prevent a loss of containment of parts and/or process gas. ~ Patrick J. Smith is lead machinery engineer at Air Products & Chemicals in Allentown, Pa., where he provides technical machinery support to the company’s operating air separation, hydrogen processing and cogeneration plants. You may contact him by e-mailing editorial@woodwardbizmedia.com.

September 2014

Mechanical Engineering Software RENT RENT RENT SALE SALE SALE LEASE LEASE LEASE

RENT RENT RENT SALE SALE SALE LEASE LEASE LEASE

•• •• •• •• •• •• •• ••

•Rental Rental Rental and and and Stock Stock Stock Boilers Boilers Boilers •Generators Generators Generators •Chillers Chillers Chillers •Deaerators Deaerators Deaerators •Boiler Boiler Boiler Parts Parts Parts •Boiler Boiler Boiler Services Services Services •Combustion Combustion Combustion Controls Controls Controls •Solid Solid Solid Fuel Fuel Fuel Applications Applications Applications

• • • • • • • • • • • • • •

Contact Re-Nu® & Lube

•• •• •• ••

Contact Re-Nu®

Boiler Efficiency • Gas Turbines Steam Turbines • Gas Expanders Cooling Towers • Chimneys HRSG • Insulation Condensers • Gas Compressors Cogeneration • Duct Design Heat Exchangers • Restriction Orifice Fanno Flow • Fans Pipe Networks • Flash Tanks Gravity Drain Flow • Pumps • • • Rental Rental Rental and and Stock Stock Stock Boilers Boilers Boilers Steam Heaters • and Psychrometrics Steam • Desuperheaters • •Properties •Generators Generators Generators Space Heating • Deaerators • •Pressure •Chillers Chillers Chillers Piping Loss• • •Deaerators Deaerators Deaerators

MS-738

Vertrel XF Cleaning Agent

MS-730

Precision Cleaning Solvent

MS-780

Acrylic Conformal Coating

m Plastic, Glass & Metal Cleaner

m MS-580

MS-260 MS-465N

•Boiler Boiler Boiler Parts Parts Parts Miller-Stephenson Offers a Wide Range of Chemicals •Boiler Boiler Boiler Services Services Services for the Power Industry! •Combustion Combustion Combustion Controls Controls Controls Aero-Duster Freeze Spray •Solid Solid Solid Fuel Fuel Fuel Applications Applications Applications ®

Contact Cleaners Solvent Cleaners Specialty Cleaners

Conformal Coatings Contact Lubricants Krytox® Lubricants

Low Global Warming Formulations Available

24/7 24/7 24/7 Emergency Emergency Emergency Service Service Service P P847-541-8300 847-541-8300 P 847-541-8300 ••F • F847-541-9984 847-541-9984 F 847-541-9984

info@indeck-power.com info@indeck-power.com info@indeck-power.com www.indeck.com www.indeck.com www.indeck.com

CU Services LLC 24/7 24/7 24/7 Emergency Emergency Emergency Service Service Service

miller-stephenson chemical company, inc. Ph: 847-439-2303 P P847-541-8300 847-541-8300 P 847-541-8300 ••F • F847-541-9984 847-541-9984 F 847-541-9984 California - Illinois - Connecticut - Canada rcronfel@cuservices.net 800-992-2424 203 743.4447 info@indeck-power.com info@indeck-power.com info@indeck-power.com supportET@mschem.com www.indeck.com www.indeck.com www.indeck.com www.cuservices.net miller-stephenson.com

Energy-Tech Showcase

BOILERS BOILERS BOILERS BOILERS BOILERS BOILERS The Energy Analyst