TECHNICAL BRIEF 1.1 PROJECT BRIEF 1.2 HOOKE PARK TIMBER 1.3 BIOMASS FUEL - WOOD CHIP 1.4 BOILER ROOM AND FUEL STORE 1.5 RESEARCH VISITS
CONCEPT DESIGN 2.1 SITE STUDY 2.2 ALTERNATIVE SITES 2.3 CHOSEN SITE 2.4 CHOSEN SITE LAYOUT 2.5 LANDSCAPE INTENTION
MATERIAL EXPLORATION 3.1 MATERIAL SOURCING 3.2 HOOKE PARK BENT TREES 3.3 BENT TREE - HISTORICAL USE 3.3 2D PHOTOGRAPH RECORD 3.4 1:20 MODEL EXPLORATION 3.5 MATERIAL EXPLORATION WORKSHOP 3.6 1:1 MOCK-UP AND TOOL TEST
3D SCANNING 4.1 3D SCANNING IN THE FOREST 4.2 BOUNDARY BOX AND SCANNING REQUIREMENTS 4.3 3D SCANNER CARRIER ITERATION 4.4 ONE-SIDE SCANNING VS 360 DEGREE SCANNING 4.5 3D SCANNING INDOORS 4.6 TREE DATA DIGITAL PROCESS
DIGITAL DESIGN DEVELOPMENT 5.1 TREE DATABASE 5.2 TREE CURVE RANGE 5.3 SORTING - TREE GROUPS 5.4 DIGITAL DESIGN INTRODUCTION 5.5 SCRIPTED METHOD A: BEST-FIT STRATEGY 5.6 SCRIPTED METHOD B: TANGENTIAL STRATEGY 5.7 COMBINED BEST-FIT AND TANGENTIAL STRATEGY 5.8 CONCRETE RETAINING WALL 5.9 ROOF DESIGN
PRE-FABRICATION 6.1 MATERIAL CATALOGING 6.2 PRODUCTION LINE DESIGN 6.3 LOG PROCESSES 6.4 DETAIL TESTING
CONSTRUCTION 7.1 CONCRETE BASE 7.2 TIMBER WALL 7.3 THE ROOF
CONCLUSION 8.1 CONCLUSION 8.2 LOG POSITION: FROM FOREST TO BUILDING PHOTOGRAPHS APPENDIX
THIS IS THE PROJECT DOCUMENTATION FOR THE 2013-15 COHORT OF THE DESIGN & MAKE MASTERS PROGRAMME AT THE ARCHITECTURAL ASSOCIATION’S WOODLAND CAMPUS AT HOOKE PARK. IT IS A COLLABORATION BETWEEN TWO STUDENTS WHO DESIGNED AND BUILD A BIOMASS BOILER HOUSE PROJECT IN THE CAMPUS. THE PROJECT HAS BEEN POSSIBLE WITH THE SUPPORT OF AN INSPIRING GROUP OF TUTORS, TECHNICAL STAFF, PROFESSIONAL BUILDERS AND A HARDWORKING AND ENTHUSIASTIC TEAM OF SUMMER VOLUNTEER, ALL CONTRIBUTING TO MAKE THE PROJECT AN ENRICHED LEARNING EXPERIENCE. THE KEY DESIGN AMBITION FOR THE PROJECT HAS BEEN THE EXPLORATION OF NATURAL FORM AND NATURAL MATERIAL IN CONJUNCTION WITH NEW 3D SCANNING TECHNOLOGY. TWO FUNDAMENTAL INTERESTS PURSUED THROUGH THE PROJECT ARE THE LANDSCAPE ENGAGEMENT OF THE BUILDING ON THE HOOKE PARK TOPOGRAPHY AND THE NOVEL USE OF HOOKE PARK TIMBER. THE PROJECT IS THE STORY OF ADDING VALUE TO A POTENTIALLY WASTE MATERIAL, EXPLORED PHYSICALLY AND DIGITALLY TO MAKE A BUILDING KEEPING TRUE TO THE NATURALLY FOUND FORM.
THIS DOCUMENT PRESENTS THE DEVELOPMENT OF OUR PROJECT OVER ITS ONE-YEAR LIFESPAN. AS A COLLABORATING PAIR OF STUDENTS WE HAVE WORKED VERY CLOSELY TOGETHER, FORMING IDEAS AND AMBITIONS IN MUTUAL DISCUSSION AND DEBATE. WHERE THERE IS IDENTIFIABLE INDIVIDUAL AUTHORSHIP OF A SKETCH, DRAWING OR STATED ARGUMENT, WE HAVE ANNOTATED THIS WITH OUR INITIALS, SS AND YZ. SATTAVEESA SAHU AND YINGZI WANG
MINTERNE HOUSE ST. OSMUNDS SCHOOL RIVER COTTAGE
HOOKE PARK TIMBER
CONCEPT DESIGN - CHOSEN SITE (SITE 1)
1 - 1. YZ
“This initial idea comes from the site location and landscape. It sits in the center of Hooke Park, surrounded by three important buildings, workshop, refectory and Westminster lodge. It is a ‘green island’ between them. The new building, should maintain this character. Also, it is located on a gentle slope, connecting the northern and southern campus. To keep the continuity of view
and space, this new building should not stand out to block this continuity. Then, in this island, we discussed and decided to keep these existing three trees, so the building will sit in the eastern corner. The western part has the potential to become a natural social place for people.” YZ
1 - 2. SS
“To design a non-object building, I imagined the boiler house to be completely hidden in the landscape, almost like a cave. As seen in the sketches, the initial concepts involved expressing the entrance as an opening to a space underneath the natural topography, the ‘building’ almost has no external envelope. The material intention is expressed as a series of cruck
frames enclosing space in the photosketch and also as the wavy layers of the landscape in the second sketch. These initial proposals established the landscape relationship to the building.”
1 - 3. YZ
â€œ If I regard the building as a part of the ground, the form could be generated, as many natural movements, a result of ground force (site1-5). It will be a naturally generated free-form, in which the timber, could form it by controlled angle and length cuts. Different from other buildings on site, it is a complete reflection of the landscape. With proper angle cut, the stagger timber could also be self-supporting against the outside force (site1-6).â€? YZ
1 - 4. YZ
1 - 5. YZ
1 - 6. YZ
1 - 8. SS
1 - 9. SS
1 - 10. SS
1 - 7. SS
While compiling this document I noticed that I have always sketched the roof hatch as a rectilinear aperture. I think we always considered it to be a simple thing in the bigger scheme of things, therefore not explored more.
The project is complete now and this rectilinear hatch really stands out. Amongst all the naturally curving building language, it can be critiqued to be prominently under-designed.
Site 1 - 11. SS
Site 1-7: "Various conceptual massing sketches showing the building placed as the north-east corner of the island. They explored how the building will affect the skyline differently when placed at different levels on the slope" Site 1-8/9: "These sections depict the building as being two different entities, the chip store & the boiler room. The two spaces have different base slab levels, the chip store sunk more. The natural quality of the island can be extended to the roof of the boiler room that can be a shallow mound. Site 1-10:" The two spaces are combined in one building here. But they can be designed to have two distinct roof systems. I imagined the envelope to
be all one system, wherein there is no distinction between the wall and roof, it is all the same thing. " Site 1-11: "These set of sketches explored different layouts contained in northeast corner of the island and keeping clear of the tree canopy. The slope of this site is very interesting. Embedding the building block at different depths would expose varied visible surfaces. The second option would allow a continuous flowing envelope that has no start or finish. The sections explores the roof to be staggered and accessible."
CONCEPT DESIGN - CHOSEN MODEL
Site 1 - Selected Model. YZ
â€œThis design is inspired by the curving and flowing contour lines of the site. Contour lines are the representation of the landscape, but also are man-made lines. It is a medium to communicate between natural and artificial world. As I always consider in this site the building should be a part of the ground, contour lines could be a concept to form the
building. I aimed to design a transition between nature and man-made; the contour lines gradually become the envelope of the building. So the building is not standing out, the interface become vague and ambiguous.â€?
Site 1 - Concept Model. YZ
LANDSCAPE INTENTION- CONTOUR LINES
DIGITAL CONCEPT MODEL Observing the buildings in Hooke Park, all the existing buildings are standing out of the ground, trying to be â€˜objectsâ€™ instead of a part of the ground. Above is the concept design driven from the key contour lines of the site. The
dynamic curves shaped by the ground turned into the outline of the building. The whole building became a part of the landscape and aimed to be harmonious with the natural surroundings.
LANDSCAPE INTENTION - A PART OF GROUND
SITE LONG SECTION The topography of Hooke Park is on a slope. The whole context is very different from the city, it is more natural and vivid. To respect this environment, our concept was inspired by the flowing
organic contour lines of landscape, so we embedded the building humbly into the ground. The horizontal curving log represent and express the landscape contour concept YZ
LANDSCAPE INTENTION - LANDSCAPE CONCEPT MODEL
This is a developed 1 to 50 concept model. We hoped to build a modest building to appreciate the natural landscape. The terrace part is following the original contour lines of the ground, while the buildingâ€™s envelope was driven by the bent tree found in Hooke Park.
This two parts, additionaly with the roof, are flowing in and above the landscape as a whole. The curves in material respond to the lanscape dynamic curves. All the natural curves become harmouniously and the form of the building naturally come from them.
VIEW FROM NORTH
VIEW FROM EAST
VIEW FROM WEST
1 :50 LANDSCAPE MODEL
VIEW FROM NORTH
NATURALLY BENT TIMBER
SAWING MACHINE TEST
2D PHOTOGRAPHY RECORD
1:20 PHYSICAL MODEL EXPLORATION
SAWING TEST LANDSCAPE STEP TEST TRANSITION TEST WALL CONNECTION TEST
In parallel with developing the landscape concept, we were exploring Hooke Parkâ€™s material resource. We found an area of naturally bent trees, which we felt could match our intention of contouring the building. These trees had grown bent
due to the position on a steep slope and had been abandoned from the commercial perspective. Because what the timber market need are straight trees. We took this chance to test how to use this natural form of timber in architecture.
HOOKE PARK BENT TREES
In the 12 compartments of Hooke Park forest, there are more than 500 naturally bent trees growing on a steep slope. They were growing bent because of their location. Due to the intelligent gravity adjustment of trees, they tried to form themselves bent to accommodate the slope. Most of them are Douglas fir, which is
suitable timber construction. However, because of their various bent shapes, they lost their commercial value in the timber market. We think, there are a lot architectural possibilities and potentials for this irregular material. This raw material represents the real beauty of the nature, which could be appreciated and
valued. For organizing the tree information in the future, we divided them into six compartments (A - F), and named the tree according to their location. See appendix 9.1.2
A B D E
HOOKE PARK COMPARTMENT 12
NATURALLY BENT TREE SAMPLES
1986 HOOKE PARK GATEWAY
Because of their distinctive and special form, Andy Goldsworthy used them for building the gateway of Hooke Park in 1986. Those uneconomic bent timber formed these two wooden rings, giving a
signpost, gateway, boundary stone and milestone to Hooke Park. Till today, some parts near the ground of the rings have rotten, but most structure parts still stand stable.
BENT TREE - HISTORICAL USE
Historically curved timbers used to have great value. In medieval times, they were used in timber framing structures in halls or barns. They were used as post, rafter, floor joint or tie beam according to the curve of the trees.
Also, they played an important role in ship building industry. Most of the structures, like ribs, need timber of curved shape. England used to require large amounts of bent timber to build naval ships in the 17th and 18th century.
2D PHOTOGRAPH RECORD
To get the curved timber information, the first method we tried was to take photographs of each tree and traced them in computer to the same scale. By this, we had a 2D image of each curved tree.
At this stage, the digital information was not exactly the same as the real 3D tree data. But this method gave a sense of the general shapes of these tree curves.
2D TREE SAMPLES
1 METER SCALE
1:20 MODEL EXPLORATION
CLEAR AREA AROUND THE TANK
CLEAR AREA AROUND THE BOILER
CURVATURE ENTRANCE CNC CUT TREE PIECES SCALE 1:20 We cut these mock-up tree pieces using CNC router, and to get more pieces for exploring, we prepared five duplicate groups of these tree samples. They came in randomly so that people will not always take the same piece - because in the forest, there are no two trees that look the same! In the model, we manually stacked these
pieces to form an envelope, which tried to express the material character itself as well as giving enough space to meet the boiler house requirements. It was hard to build a long curved wall, so we broke the wall in the middle. This break became the entrance of the building.
1 :20 MODEL USING THE DATA FROM 2D RECORDED TREE PROFILES
This model from FEBRUARY 2014, whilst without completely accurate information, was successful in allowing us to ‘play’ with the pieces - like playing with Lego - and explore the formal potential of the curved timber. This model tried express the initial idea meanwhile get a sense of the general length, diameter, and curvature of these trees. And at this stage there isn’t much limitation of material or regulation, the form could be more freedom, like the big gaps between
to curves to form a skylight. Also, it explored a easy way to connect two curves smoothly: the ends of neighbor curves have some parts overlaying, then mark start point and end point the overlaid part on both curved pieces, and then connect this two points separately on both pieces. These two lines will be the cutting position for two pieces. By this, these two cuts can match and the joint could be smooth.
MATERIAL EXPLORATION WORKSHOP
3.5.1 MATERIAL SOURCING AND PROCESS
In this workshop, with our visiting tutor, we first used real bent trees to make a 1: 3 mock-up to test timber processing, design intention, and details. Choosing scale of 1:3 because the material we used was one third diameter of the desired diameter (The big diameter trees were kept for the real construction).
Meantime, the length and curvature were the similar to the bigger trees. We went to the forest and felled some small curved trees, some have curvature in 2D (single curve) and some in 3D (double curve). We cut these trees to 6 to 7 meters, which two or three people can carry.
For our real construction logs, length and weight are also aspects to be considered. For the full diameter logs the maximum length we could carry was about 4m.
TESTING SAWING BENT TREE WITH WOOD MIZER IN THE SAW MILL
Maximum Length: 4 metres
Maximum Width: 1 metres WOOD-MIZER MILLING MACHINE SIZE
Because we tried to waste less material, all the trees were cut till approx. 30mm diameter, the length of which were varies, from 3 meters to 7 meters. So to process these trees, we first tried woodmizer saw mill, which is used for large size trees. During this process, we realized the wood-mizer had very restricted
size, 4 m. by 1 m., by which means, when treesâ€™ curves expansion were more than 1 m., we had to stop in the middle, adjust and saw it twice. Sometimes, the second cut didnâ€™t match the first one (photo 3).
At this stage, 3D scanning was not yet ex- various organic edges between the bark plored. Our reference untill now was the and sapwood, which made the natural 1:20 model with CNC cut sample trees. curves more readable. Taking the stacking strategy forward, we started exploring the options of how to cut some top and bottom flats of the round logs for stable stacking. The result was interesting; these man-made cuts left
SAWING SAMPLE 1
SAWING SAMPLE 2
SAWING SAMPLE 3
Also, we found that almost all trees were slightly double curved. When we placed them on the saw mill, the operator told us they were hard to see how to place them to get most material. Through iterations of rotating and testing, we got various sawing results. These with double curvature were only touched at different parts along the curve; some
were cut at start and end (sample 1), some in the middle part (sample 2), some were cut too deep so that it lost half of the material (sample 3). But these variations may contribute to a vivid and changing texture to the building.
In retrospect, sometimes there are some unexpected results coming out because of the organic and distinctive natural forms or structure. This is the most exciting and essential part of this natural material, which should be appreciated and expressed in architecture. We tried to keep this character in the design.
LANDSCAPE + TRANSITION + WALL
We were exploring three parts of the building in the workshop; the landscape terrace, transition to the building and the wall details. We intended to use the same material from Hooke Park in these three parts, but by different processes and details, they could be expressed differently. We can change the density of logs from landscape to building part, to create a transition of natural to artificial landscape (sketch 8), accordingly, the log rising from the ground, forming the sitting areas, to the wall and roof, forming the envelope of the building. Also, logs in landscape part were separated to follow the contour line, they could keep the raw form, then when it turned to the building part, they then form a stable wall (sketch 11).
Landscape: The site has trees and small vegetation, Also, we realized the log was taping along and water accumulates to generate a the curve, so we could possible join them small pond in the summer. This microtop to top, and butt to butt (sketch 9). environment could be very comfortable for people chatting and enjoy the view of sunrise and sunset. So we intended to create sitting area in the landscape part, like a log bench. To avoid the wood from rotting, we need to rise the wood upon the ground, possible hold by stones (sketch 6/photo 10) or timber post (sketch 7).
LANDSCAPE + TRANSITION + WALL
Transition: The transition from the landscape to building needs to be smooth and natural. So we aimed to keep the joint tangential between the intersecting log curves. The section of log in landscape part has a round section, whilst in the building part, it was sawn top and bottom. To achieve a continuous joint, the landscape log could be sawn at the end, to subtly insert into the wall.
LANDSCAPE + TRANSITION + WALL ALTERING THE ROUND LOG SECTION We explored cutting the round logs in different ways, top and bottom, inside and outside the curve, and the combination i.e resulting in a rectilinear section. The top and bottom cuts were more easily achieved than the inside and outside
given the tools available. Stacking with top and bottom flats was also simpler than stacking with top and bottom curved which would required complex joining strategies.
TOP AND BOTTOM
INSIDE AND OUTSIDE
Using round logs that transformed to a rectilinear section by cutting facets on the log, we tested a scheme that had organically flowing stacked logs transforming to regular section at the entrance of the building. We imagined that his strategy would give a formal and clear architectural language at the entrance, slowly merging into the landscape as we
move further form the entrance. We did not pursue this scheme more because we wanted to pursue one log section detail that will work for the whole building. Also, the entrance is the only place where this chosen log section will be visible. Therefore it made sense to express that rather than a rectilinear log section.
CONNECTION STRATEGIES These sketches show the different was of connections to stack the logs. Here we also explored cutting the full length of the logs to segments so as to optimise the material most. The different options for connecting were blocks/ packers,
continuous groove along the curve and dowels. The final decision was to use packers as they can be standardised and can be used for connecting any curvature of logs.
NATURAL TAPER DESIGN STRATEGIES Working with natural taper in the logs was tested here. Retaining the taper of the logs resulted in varying gaps between logs when stacked. With the logs bending, it was also difficult to stack them as there was no consistent contact.
Our final decision was to build without optimizing the natural tapering of the logs, extracting the maximum horizontal slab possible for each log. Looking back at our design strategies that optimized using the logs with their taper, the project could have been very different from what it is. There are interesting complexities involved in working with the tapers, for example, the point where a fat end and a thin end meet on a stacked wall will have very varied conditions each
time. The realised design gives priority to contour lines to merge horizontally to the building to depict landscape continuity. Using the taper has the risk of a â€˜messyâ€™ stacking, which still remains unexplored.
SLOT JOINTS Two kinds of slot joints were explored in this workshop, connecting logs with natural taper and without taper. As shown in images X & X, for connecting logs without taper, i.e connection between two flat surfaces, we tested a biscuit joint. This was a hidden joint, stopping the logs from slipping but we were not sure about its structural strength. The second test, as shown in images x to x was to cut a continuous groove along the length of the both connecting logs, and slotting
a bending piece of ply in-between. This option could accommodate any taper, i.e a varying gap along the length of log. This was not a very successful option as it required very precise grooving on both faces connecting that could have been a laborious task for construction and also the bending plywood had insufficient strength to take the load of a stacked log wall..
PACKER JOINTS These prototypes tested stacking with top & bottom sawn surfaces, connected with standard sized packers, allowing for a consistent air gap between logs. As shown in image x, the top connection insets the packer to the log above and the lower connection is placed between them. Both could require screw fixings as nothing is holding it in place as compared to the slot joints. The most useful reasons for using packers on sawn surfaces is that we would not have to
predict their exact position as opposed to precise slot joints, it is highly flexible to position and can transfer the load of a stacked log wall. Image X tests texture change from bark to sawn surface in the elevation. As the bark is a very thin layer, even though we found the texture interesting, the transition from bark to sawn is short.
WORKSHOP LESSONS & DECISIONS This prototype tested packers as connector for logs with natural taper. The strategy of using standard packer sizes fails here as the height of packer will always be varying depending on the gap between the logs. Also, unlike the top and bottoms cut flat logs, to get a proper contact with the packers, slots were required on the tapered logs to grip the packers. Having bespoke individual connections was surely not the way to go.
The workshop was very valuable and set the direction of the project in many ways. Cutting slots or grooves on the logs, be it for biscuit joints, bending plywood joints or even open slots to hold packers, were not a good idea for connections because water would stay in there and would rot the logs. Stacking without any gaps between the logs would also rot the timber. The best solution to keep the timber dry was to pro-
vide gaps between the logs, and standard size packers fixed with mechanical fixings was decided to be the way ahead. All the tapering options tested were complicated and it was difficult to use standard connections. Top and bottom cut flat had a crisper and cleaner architectural language with one regular sized connection as opposed to trying to use the taper which had lots of variations in visual impact and connection detail that
risked weaker architectural language. Further discussions at this stage involved the materiality of the packers and their positioning. The design impact of the packers when randomly placed as against placed in an orderly manner was explored. At this stage, we did not test the packer as a appropriate connection when the logs start stepping in and out from the lower log. This is discussed in section 7.2.3.
1:1 MOCK-UP AND TOOL TEST
During the last workshop, we realized the issues associated with the bent timber, so this time we tried to explore how to solve them through 1:1 mockups. The section of the log above shows the different details; the angle cut is designed to allow water to run from the surface of one log to the next below, which prevents the rain from entering into the building. At the bottom of the section, a small drip cut helps the draining system. It is essential to keep the drip cut beyond the chamfer below. These
two parameters became important in real construction for waterproofing. The other deeper cut is the relief-cut to prevent the log from splitting elsewhere, because we used green timber for our construction. Without the relief cut we risked radial splits occurring elsewhere on the section, which could allow water ingress into the timber. Later on, we realize how difficult to achieve these details on irregular timber.
Developed Tool: Rail Chain Saw
To freely saw these curved timber, we first tested an improvised chainsaw with wheels. It can move along the curve to cut irregular pieces. There is also a pivot to adjust the cutting angle. The problem of this machine is that it is difficult and dangerous to control, especially when we need to follow a guide line to cut the angle. Only the specially trained staff were allowed to operate it.
Set the Angle
OUTSIDE WALL TEXTURE
THREE DIFFERENT ANGLE CUTS
RELIEF CUT AND DRIP CUT
To test the water draining above the surface, we set three different angles, 10, 20 and 30 degree to see how the water would drain. The result was successfully shed in all three of them, but aesthetically we preferred the 30 degree version, in which the proportions worked well and people can see the color and texture difference between the sapwood and bark.
SCANNING IN THE FOREST
3D SCANNER CARRIER
ONE-SIDED SCANNING V.S. 360ยบ SCANNING
TAPING DIAMETER CENTRAL CURVE MAXIMUM DEPTH BEST SAWING POSITION
3D SCANNING IN THE FOREST
All the bent trees have extremely irregular shapes. To use them in the building design we decided to collect the treesâ€™ geometric information and transfer them to digital data. The aim was to build a digital database of all the trees to make this material information accessible for the future design. Having captured a sample of the tree
geometries through 2D photography, we turned to 3D scanning to fully record the curved treesâ€™ information in a three dimension way. Firstly, we marked and named each potential tree. Then we pruned selected trees to clean the trunk surface to scan. After that, we took photographs and 3D scanned them in the forest.
BOUNDARY BOX AND SCANNING REQUIREMENTS
SCANNING BOUNDARY BOX
APPROX. 4 METRES
SCANNING QUALITY - FROM LOW TO HIGH
We used the Kinect sensor, from the Microsoft XBox 360 gaming system, for the 3d scanning. This low-cost approach met our requirements well. The software we used to run the scanner was called Skanect, which provided a user-friendly interface for acquiring the 3d scans. The first interface operation was to set the scanning parameters, in which the boundary box was defined by the users. The box dimension ranges from 0.1 cubic meters to 12 cubic meters and is designed to get rid of irrelevant environ-
ment. This was very usefully scanned in the forest, where there was a lot of vegetation around the target tree. Through the tree scanning process, we found the boundary box was best set to 4 to 5 meter in length, double in height. If we set the box too big, it would lose detection above a height of 4 to 5 meters because it faced direct light as it went higher. The Skanect interface also allowed us to adjust the resolution quality. The highest quality resolution, with a maximum number of points in the point cloud,
described the tree in more detail but took more time to scan and process. We tested different scanning qualities to find the optimum level of detail without making unmanageable large data files. We found the scanner was very sensitive to the light because it uses an optical TOF (Time-Of-Flight) system. This meant that when scanning outdoors it only functioned well in cloudy weather.
3D SCANNER CARRIER ITERATION
FIRST SCANNING TESTING
SCANNER CARRIER SKETCH
We went to the forest to test the scanning process. One of the difficulties was moving around with all the equipment. Initially we needed three people to finish the whole process; one person to carry the scanner, and the other two to carry the laptop and the chair. The weather in April was always raining and the ground was muddy and slippery. It made the moving around even more tough. So we developed a carrier which would allow one person to carry all the equipment.
We developed the prototype carrier, with nylon stripes, plywood and screw. The carrier held the laptop while scanning and moving, and we put the scanner and battery in our pocket.
SCANNER CARRIER PROTOTYPE
We developed the final carrier by modifying a baby carrier. It has a soft and comfortable belt to allow people carry heavy equipment for long time. Also, we tied a back bag to the belt to take the battery and scanner. The weight in front and back were balanced to lighten the burden on the shoulder. In total, we scanned 155 trees in the for-
SCANNER CARRIER - SIDE VIEW
est. It took 5 to 10 minutes for each tree. The scanning process itself was quick, but most of the time was spent waiting for suitable light conditions, moving to the target tree on the slope, or pruning the tree before scanning. With just two people, it took us around two weeks to collect all the information.
SCANNER CARRIER - BACK VIEW
SCANNING IN THE FOREST
These images below present a sample of the trees and the corresponding 3D scanned mesh. See Appendix 9.1.1 and 9.1.2 for the full database of photographs and scan meshes.
TREE SAMPLE: A_02
SCANNED MESH : A_02
TREE SAMPLE: A_07
SCANNED MESH : A_07
TREE SAMPLE: A_09
SCANNED MESH : A_09
TREE SAMPLE: A_11
SCANNED MESH : A_11
ONE-SIDE SCANNING VS 360 DEGREE SCANNING
We experimented with two methods of scanning: to scan just one visible side of the tree from a fixed point, or to move around the tree to produce a full 360-degree scan. The 360-degree scan gave the complete accurate information, but whilst the scanning software was able to compose a full 360 degree scan, this took much more time (about 6 times longer) and was more likely to lose track of the detection during the process. Because our scanning site was outdoors and full of bushes, it was hard to turn the
360 DEGREE SCANNING
scanner around the tree whilst keeping it stable. So to be efficient in scanning process, we decided to derive our geometric information about the tree from just the one-side scan. We compared the results of the two methods and found (as illustrated below) that the central curves were very similar; while the outlines of the trees had some difference.
CENTRAL LINE COMPARISON
OUTLINE COMPARISON - FRONT VIEW
OUTLINE COMPARISON - TOP VIEW
3D SCANNING INDOORS
BENT TREES DELIVERED TO THE BIG SHED
For these trees which cannot be scanned in the forest (due to light or accessibility issues), we considered to scan them after felling. So these felled trees have been delivered by telehandler to the Big Shed. Using pulley and rope tied in flexible and non-flexible nodes, these trees were hung through the roof structure of Big Shed. The result was more ideal than scanning outdoor. First, the scanner can detect till 6 to 7 meters without the light issue. Secondly, the scanning environment is clean so it saved a lot of time of in the later process. In retrospect, for this project, it probably would have been more efficient to have felled all the trees first and then scan them in this controlled way, to produce cleaner and more accurate scan meshes. However, at the start of the scanning phase we were not sure of the number of trees needed and we initially intended to have a full database of standing trees and then to decide which trees to actually fell based on the outcome of the
digital form-finding. In the end we felled all of the trees that the forester was happy for us to use, and utilised nearly all of them in the building. For future projects at Hooke Park using 3d scanning technologies, weâ€™d recommend scanning material in the controlled environment if possible, especially given the Dorset climate.
HANGING LOGS WITH PULLEYS AND SAILING KNOTS
HANGING LOG FOR SCANNING
SCANNING IN THE BIG SHED
TREE DATA DIGITAL PROCESS
A Rhinoscript method was developed to extract the relevant tree data from each of the 255 3D scan meshes. We decided to convert the surface data into a single centre-line and a series of diameters
along that centre-line. This simple data was then stored as text files which were accessed by the curve-optimising scripts. The diagrams below explain this method.
MESH WITH GUID
This is the original mesh exported from the Skanect, which includes elements of the surrounding environment that have been scanned unintentionally.
The first step was to clean the mesh to get rid of the irrelevant objects, like ground, bushes or branches.
A guide line was d trunk to provide a best-fit script.
MESH WITH SECTION CIRCLES
MESH WITH CENTRAL CURVE
drawn along the curved a starting axis for the
Along the guide line, the mesh was divided into 200 segments. A best-fit arc was found through the point-set of each segment, then this arc would be completed to a circle. These circles represent the changing diameter.
The center point of each circle was connected to form the centre line of the tree.
Script-determined planer defining maximum available depth P A_07_114
BEST POSITION AND MAXIMUM DEPTH Following the extraction of the centreline and diameter-circle data, another script was written which was used to determine the maximum available thickness for a parallel-faced sawn slice through each tree, and the corresponding tree orientation. This was found by defining a best-fit plane through the centre line nodes, then offsetting this plane up and down so as to just the intersec-
tion with all of the diameter-circles. The log was then rotated three-dimensionally into this optimum position. In this way, the slice saw cuts would, in theory, just â€˜kissâ€™ the logs at two points (P and Q above). In reality, given the scanning tolerance and thickness lost in de-barking, the assumed maximum depth was reduced by 30 mm.
MAXIMUM DEPTH OF ALL THE TREES
CURVED LENGTH OF ALL THE TREES
MINIMUM DIAMETER OF ALL THE TREES
LOG WALL DESIGN
BEST - FIT STRATEGY
DIGITAL MODEL TESTING
DIGITAL MODEL TESTING
DIGITAL MODEL TESTING
PHYSICAL MODEL TESTING
PHYSICAL MODEL TESTING
PHYSICAL MODEL TESTING
DIGITAL MODEL DEVELOPMENT
PHYSICAL MODEL DEVELOPMENT
CONCRETE RETAINING WALL DESIGN
DIGITAL DESIGN DEVELOPMENT
TREE CURVE RANGE
1000 MM 700 MM
SORTING - TREE GROUPS
GROUP A BEST SAWING DEPTH: 165mm 165 - ABOVE MM - above
NUMBER OF TREES: 27 TREES
GROUP B BEST SAWING DEPTH: 135 -165135 MM- 165mm
NUMBER OF TREES: 51 TREES
GROUP C BEST SAWING DEPTH: 105 -135105 MM- 135mm
NUMBER OF TREES: 33 TREES
GROUP D BEST SAWING DEPTH: 75 - 105mm 75 -105 MM
NUMBER OF TREES: 18 TREES
GROUP E BEST SAWING DEPTH: 40 - 75mm 40 -75 MM
NUMBER OF TREES: 18 TREES
Digital information gave us the ability to sort the data according to different design requirements. The first arrangement was based on the best available sawing depth. We organized all the trees from bigger to thinner depth, which would be correspondingly placed from the bottom to the top of the building in order to be structurally sensible. The logs’ depths were calculated ranging from 40mm to
228mm (from first scanned data), from which, we divided them into five groups with the depth interval of 30mm for each groups. From our timber expert’s perspective, trees with a depth more than 100mm are recommended to be used for building the log wall, the trees with smaller depth could be used in other parts, i.e. balustrade wall.
DIGITAL DESIGN INTRODUCTION
In parallel with collating the database of the tree geometries, we worked on techniques for determining the arrangement of the tree components within the building’s wall. We decided to automate the process using Rhinoscript routines that would iteratively test-fit the treecurves and optimise the wall geometry with respect to criteria that we had set. This approach – described in detail below - meant that we could rapidly test
and refine many variations for the wall, and could continue to adapt the wall in response to the realities found during the fabrication of the actual timber components.
course. In the end, we used a combination of the two methods, which is discussed in the section 5.7. Both methods followed the same basic scripted procedure, in which each course’s target curve was populated Two fundamental methods were develsequentially along its length by the oped – firstly a best-fit-to-target-curve components which best satisfied the test approach, and secondly one in which criterion at that position. The pseudoneighbouring pieces were placed so as to code for this is: maintain tangential curvature along each
User inputs: Select available tree-curves Select target-curve Select point on target-curve to define start position Procedure: Do until the end of the target-curve is reached: Loop through all available tree-curves: Loop through 4 possible orientations of tree-curve (up/down; forward/back): Copy and place tree curve with start at start position and its end on target curve Measure criterion: (misfit distance of mid-point from target curve OR tangency angle relative to previous component) If component is best so far then keep in place, else delete Loop Remove kept component from the remaining set of available curves Find end point of kept component and use for start point for next component Loop Loop End
A video capture of the operation of a typical script is presented in the stills below, and can be seen at http://www.aaschool.ac.uk/PORTFOLIO/MICROSITES/microsite.php?title=Design%20&%20Make%20MArch&url=designandmake. aaschool.ac.uk/&return=graduate.php
SCRIPTED METHOD A: BEST-FIT STRATEGY
ONE TREE WITH FOUR DIRECTIONS
For arranging these trees, we had thousands of possible combinations. The first idea followed the former landscape mode, from which, a set of horizontal target curves was extracted, to represent each course in the wall. The script then worked by testing all the available treecomponents in the sequential positions along each course. It tested each compo-
nent in its four possible orientations. The offset distance from the target curve was measured to decide if one curve was the best-fit curve. All available components were tested and the one which fitted best is kept in position and removed from the available set of components.
LOOKING FOR THE BEST-FIT CURVES
BEST-FIT STRATEGY: DIGITAL PROCESS TARGET SURFACE AND EXTRACTED CURVES
T - FIT CURVES
TAGET CURVE V.S. BEST - FIT CURVE
BEST - FIT CURVES WITH EACH CURVE’S NAME
The diagram on the previous page shows how the script worked to find each tree for each course. The first column shows the different tree groups arranged in different layers in the building. To minimise wastage whilst maintaining parallel courses, we made smaller groups of similar diameter trees, and varied the course heights accordingly. We then ran the scripts for each sub-set of available tree curves. The second column shows the relationship between the number of trees in
each group and number of layers in the building. Groups with bigger quantity of trees would occupy more layers. As the scripting looked for the best-fit trees, it needed a certain number of trees to select from. The third column shows the result of all the best-fit trees. For certain trees and groups, this result would be the only one. The fourth column illustrates the comparison between original curves and best-fit curves. From observing, the
number of trees in one group would affect the fitting quality. The bigger the database is, the better result could be achieved. The first three groups (A-4 to C-4) fitted quite well while the last two (D-4, E-4) didnâ€™t because the database was too small. The last column shows the name of each tree in the forest matched each tree curve in the model. Ideally, we thought we would only fell the trees shown in this model. The name of the tree would give us the information of the treesâ€™ location.
BEST - FIT TREE FORM
Once the trees were located, we extracted the outline of each tree from the database. Then the form can be seen from the real material data.
In parallel to this process, we rethought the relationship between the given form and trees. It seems the given form limited our imagination of tree forms. Even though the generation of the form involved the material information (from 2D photo mock-up tree pieces), we missed the diversity and vividness of these dynamic material.
The illustration shows the fit-in curves of different groups in elevation. From observing, the curves at the bottom were more single curves while at the top there were more double curves.
There were some problems found in this best-fit strategy. We matched the central line of the curves to the target curve, while the outline of logs staggered due to the difference between the trees’ diameter. There are two aspects led to the situation; the taper of the trees and the big difference between the trees. The other problem was seen in the joint of the trees. Since the offset distance
from the target curve is the only criteria to find the best-fit tree, it didn’t consider the joint of two trees. So the result shows some joints were not turning smoothly. Another problem is the best-fit curve the computer selected can’t fit the target curve very well. One reason was that the target curve was too gentle, so there was not any tree curve could match that part of curve; another reason was the data-
base was too small, so there was not any proper tree to match. These problems were addressed by developing of the script that optimised the tree-component placement to best achieve tangency continuity between the components.
BEST-FIT STRATEGY: PHYSICAL MODEL
To further test the â€œbest - fitâ€? strategy, we cut a set of tree pieces by CNC, and built a 1: 10 model of the wall according to the digital model. While we were making this model, no adhesive was used to connect the two layers. The model can stand by itself due to balance between curving out and in.
To imitate the sloping landscape, we staggered some wooden pieces to form several steps, and then placed the wall on them. From this physical model, we realized the mismatch joints were obvious and affected the flowing effect of the whole wall.
To solve the staggering joint, we decided to create two different textures on the outer and inter walls. We matched the outline of the logs outside to create a smooth flowing curve, and left the difference between logsâ€™ diameter inside to create a rough staggered texture internally.
SCRIPTED METHOD B: TANGENTIAL STRATEGY
MINIMUM SPACE BORDER OFFSET BASE LINE
MINIMUM AREA REQUIREMENT
We changed our method to construct a curve. Instead of drawing a curve artificially, we let the tree curves inform the curve. To realize this idea, we started to focus on joining technique of the curve trees. Tangential joint could be a parameter to look for the next curve, in which way, it created thousands of options to form dynamic curves, also the joints
would be smooth and continuous. In this process, we need a set of tree curves to select from, a base line to guide the curvesâ€™ direction, and the script that would calculate the next curve which has the closet tangential joint. To make sure it would fit the buildingsâ€™ envelope, we calculated the minimum space needed for the building (accord-
ing to the space requirement of machine and chip). This border was offset 1000mm outwards so that any result would not affect the minimum space. This offset curve became the basic line to run the tangential script.
TANGENTIAL STRATEGY: TESTED COMBINATIONS
GROUP A POSSIBLE COMBINATION
GROUP B POSSIBLE COMBINATION
GROUP C&D POSSIBLE COMBINATION
GROUP D&E POSSIBLE COMBINATION
The diagrams in the previous pages show the possible curve combinations of different groups. In this process, the script would search in the whole group and construct a tangentially jointing curve. Then these curves used in the constructed curve would be removed from the group and the script would search the rest of the curves to form the next one. Different tree curve base would
result in completely different curves. In this way, we gained 9 curves respectively from group A and B, 7 curves from group C, 5 curves from group D, and 4 curves form group E. If we subtracted some curves from or added some to the group data, there would be completely different results. So the possibilities of combination were infinite.
TWO CURVES POSSIBLE COMBINATION
If we selected more than two curves from different groups, there were some interesting combinations between them.
A2 + B2
A3 + B3
A4 + B4
A5 + B5
A6 + B6
A7 + B7
A8 + B8
A9 + B9
THREE CURVES POSSIBLE COMBINATION
A2 + B2 + C2
A3 + B3 + C3
A4 + B4 + C4
A5 + B5 + C5
A6 + B6 + C6
A7 + B7 + C6
A8 + B8 + D1
A9 + B9 + D2
COMBINED BEST-FIT AND TANGENTIAL STRATEGY
TWO DISTINCT TANGENTIAL CURVES
From tangential strategy, we achieved lots of dynamic curves from the original tree curves. But these curves are separated, not forming any surface. There were big gaps form one curve to the other. So we thought to â€˜fillâ€™ curves between two tangential curves in order to
form a continuous surface. The method we tested was to use two tangential curves as section line, and then lofted a surface between them. After that, new section lines were generated according to this surface. In this way, two tangential curves formed a dynamic flowing
surface. In the next step, we reused the best-fit script, to look for the curves to match the new contour lines. The result was very good because the curvature of the surface fit that of tree curves well.
COMBINED SCRIPTING DIGITAL PROCESS
COMBINED STRATEGY: PHYSICAL MODEL
Again, we built physical model to test this strategy. The result was ideal which could show the lively and diverse tree curves, flowing from one side to the other. The only concern was that the joints came to the similar place, which created a vulnerable part structurally. We though
the reason was because these tree curves were similar length, so the joints appeared at similar place. To solve that, we staggered the starting point for each layer. We found that the joints were more even after that.
1. OUTSIDE WALL WITH CHAMFER
2. CHAMFER BETWEEN BARK AND DEBARK
3. INSIDE WALL WITH STAGGERING JOINTS
4. GAPS FORM HORIZONTAL VIEW
5. GAPS ABOVE HORIZONTAL VIEW
The image on the left shows different textures of the wall. We cut the chamfer on each piece and matched the joint to joint to create a smooth connection (1). We colored the pieces to imitate the difference between bark and debark. The chamfer cut created an organic curve to be read on the surface (2). The inner wall had a rough and organic texture because
of the staggered joints (3). The differences between curves formed some gaps, which would be seen differently from various height and perspective (4&5). We were also concerned with the water issue between the gaps. If the wall was tilting out, then the log above could shade the gap, but if in reverse, the gap would be a problem.
COMBINED STRATEGY: DIGITAL PROCESS
Some overlay points could act as column
Selected curves become key section lines
Selecting one curve from each group (two from group B)
Key section lines construct a continuous surface
New surface generate new contour lines
All contour lines were cut to accommodate the landscape
Different groups were assigned to different section lines
Search for the best-fit curves for the new contour lines
COMBINED STRATEGY: DIGITAL LOG WALL
The tangency factor was particularly important as it allowed us to ensure we created a smooth flow of tree curves along a course. The process was very iterative. For example we identified some key courses in which good tangency and interesting & dynamic free flowing curves had been achieved, and then used them to
recreate the initial target wall surface and start the process again. In this way we were able to create a more natural, smooth flowing progression of curves. Also, when overlapped, these key courses allowed us to control points of connection which contributed to the wallâ€™s structural stability. Again, this was an iterative process, with
many evolving versions of the wall being defined, as we improved our understanding of both our digital strategy and of the trees themselves. The result is an organic geometry derived from this set of bent tree curves. The natural trees found their place again, flowing in the Hooke landscape.
COMBINED STRATEGY: LOG WALL MODEL
Preparation Followed by this tangential and best-fit strategy, we applied them to the whole flowing log wall, and built a 1 : 10 scale model. We prepared a set of tree pieces cut by CNC. They came with five different thicknesses according to their groups. Then we simply built a one-angle slope base to imitate the concrete retaining wall. Extracted from the digital model, we printed out the plan of each course, trees in this course and their names.
CNC CUTTING TREE PIECES
To find the relationship between different courses, we overlaid two courses and printed together. So we could lay one course on the other according to the plan. On the real construction, we didnâ€™t use the relationship between courses due to accumulated difference. Instead, we used constant sections to indicate the location.
CONCRETE BASE MOCK-UP
FIVE GROUPS OF TREE MOCK-UP PIECES
PLAN OF TREES IN EACH COURSE
Details In the model making process, cutting compound angle on the log was the most tricky part. It took a few times to adjust and to level each time. Except the angle piece, the rest of them were placed quite efficiently. The process reflected the time-consumption of real construction.
1. When there was a slight angle difference in the tangential joint, we drew a parallel line and sanded one piece to match the other piece. 2. When there was big gap between two courses, we tried to change the piece from the spare group to fill in the gap. 3. When one piece was not matching at the end, we cut the wrong matching end and sometimes by mirroring it would match. 4. When occasionally some joints came at the same position, we cut the ends of both pieces and fill in with another short piece. These methods allowed us to test techniques that we could apply at full scale in the building.
1: 10 MODEL We aimed to build a log wall showing the possibility and character of the naturally bent trees. Through the digital process, we achieved this dynamic flowing wall communicating between raw material and architectural design. In making this model, what is tricky to make, i.e. compound angle cut at log placing on the concrete, finding the location for each log, also reflected the difficult parts of the real construction.
1. LONG CURVING WALL AND SHORT CURVING WALL 2. OUTSIDE FLOWING WALL 3. INSIDE STAGGERED WALL 4. ENTRANCE 5. TYPICAL LOG JUNCTION 6. LOG WALL SECTION SHOWN AT THE ENTRANCE
CONCRETE RETAINING WALL
ONE-ANGLE CONCRETE WALL
CONCRETE WALL AND TIMBER WALL CONNECTION OPTIONS
Because the building was half-buried under the ground, a concrete retaining wall was needed to support the timber wall above. The form of the retaining wall was defined by the curve wall and straight log wall above. One important concern was the timber shrinkage. At the beginning, the retaining wall top edge just followed the varying slope of the site. However, the roof would twist after the timber shrunk differently at different positions. So we came out with a solution - a retaining wall with the top of the wall defining by “cutting” it with a single sloping plane - which means, if the log wall shrunk, it
would shrink proportionally along the one - angle slope. So the roof would tilt forwards instead of twisting. The sketches on the left page show two iterations of the concrete retaining wall. Initially, we envisioned a stepped slab allowing for a shallow chipstore with two agitator plates. This evolved to a single-level slab with a deeper chipstore and one agitator. We have two options for where the log wall met the concrete wall. The first one was to make steps on the concrete retaining wall so each course would be placed each step. The benefit was that it
was easy to level and on extra angle cut on the log. But the disadvantage was that the step height needed to be fixed. There were still a lot of variations in the logs at this stage and we couldn’t completely predict the height. Another drawback of this detail was, not easy for water draining. The second one was to make a continuous slope, which means we need to cut angle on the log to fit in concrete. But the slope could accommodate any thickness of logs and benefit for water running along the slope.
TIMBER WALL + CONCRETE RETAINING WALL
Finally we chose a continuous one-angle concrete retaining wall because we could not predict the height of each course at this stage. The angle was designed the same from the curved wall to the straight wall. In the construction, the compound angle cut on the log took a lot of time.
ROOF JOIST LAYOUT SKETCH
ROOF KERTO LAYOUT SKETCH
The roof structure consists of wooden beams spanning between the internal blockwork wall and the exterior to log walls. Above the joists are Kerto sheets that form the flat roof (Kerto is a laminated-veneer-lumber product manufactured in Finland).
ROOF DECKING PROPOSAL
We designed decking level above the Kerto to let the roof be accessible to people. Following the landscape idea, the decking pattern would reflect and continue the contour lines of the site.
ROOF STRUCTURE + TIMBER WALL
All the roof structure would sit on the 15th course. Because the roof slopes slightly from back to front to drain the rain water, we had to adjust the hanging brackets to accommodate the changing height.
PRODUCTION LINE DESIGN
IN THE FOREST
IN THE YARD
6.1.1 IN THE FOREST- SELECTION, MARKING & FELLING
The curved trees from the forest were meticulously marked, scanned and re-marked till they were harvested and bought to the yard for processing. The complete sourcing process can be divided into 4 stages. The first stage of marking is shown in the images above. We walked through Compartment 12A of the forest marking all the curved trees that we though we might be able to use for the project. Our standard for selection was the diameter of the tree trunk at about 1m above the
ground level. We selected the trees with a diameter of 150mm or more. We did that by using ropes to measure the standard and then marking them with white and orange tapes. We marked approximately 300 trees at this stage The second stage was going through the complete compartment again, but this time with Chistopher Sadd, who is the Hooke Park forester. Christopher told us which all trees we would allow us to harvest. He decided which trees are alright to take down considering the strand
strength of the forest and considering the quality of each tree. By the end of this stage, we had already ripped the tape off of almost half of previously marked trees and we were left with 156 trees we could use. In the third stage, we divided Compartment 12A into 6 sub-compartments, naming them from A to F for our convenience. We then 3D-scanned and photographed the marked trees, tagging them all with different colored forestry tapes as per different compartments,
naming each tree. We saved the names of the 3D-scanned files and the photographs exactly the same as the name of the trees. The names given to each tree had a format. For example, a tree named A_07_D, meant that it comes from sub-compartment A, 7th tree in our movement from the road to the top of slope, and ‘D’ standing for ‘Douglas Fir’ the specie of the tree. By the end of this stage, we had 156 trees tagged, scanned and photographed, ready to be taken down.
The fourth stage was re-marking all the selected trees again, this time for the hired timber harvesting team. We marked the trees with their names with bright florescent forestry tree marking spray paint for better visibility in the forest. We gave the exact number of trees from each sub-
compartment to the harvesting team, for them to bring back the cut trees to the yard and stack them in their respective groups. We also instructed them to try and cut the trees as close to the ground as possible because we did not want to loose
any of the beautiful curves which starts at the bottom of the trunk. They cut the trees to 4-5m lengths and also cleared the branches in the forest itself.
WINCHING TREES DOWN TO THE ROAD
All this process of sourcing and cataloguing was done in 3 weeks in May. It was a physically exhausting experience to walk up and down the slope of the forest constantly day after day. As discussed in
the earlier chapter on 3D scanning, the scanning itself was ideal on cloudy days. The most efficient days were the gloomy days where we had to walk from tree-totree on wet slippery forest floor, with the
scanner carrier with all scanning equipment and various other marking tools and a camera.
6.1.2 IN THE YARD- CHECKING AND ORGANISING
Once bought to the yard and laid in groups as per the sub-compartments, all the logs had to be re-grouped and organized into groups according to which of the 19 courses they belonged, to be processed together. Until this stage we had the digital model ready from which course definition of each log could be extracted.
While trying to find the logs for each course we discovered few crucial losses of data from the forest to the yard. Some of the logs had lost their names. This was because the forestry tapes were frail and they were torn due to dragging of the logs on the forest floor, and some of the spray painted names were not readable because parts of the bark had fallen off.
We resolved this disconnect of digital data and physical material by two strategies. Firstly, we used the plans (Appendix 9.1.3) derived from the 3D-scanned trees to visually find the appropriate curves from the missing list. We found many trees like this and to find the remaining, lead to the re-scanning process in the Big Shed as mentioned in earlier chapter.
Theo. DIA Comment Course 13 Target Height C_29 A_12 D_06 B_32 A_27 I_53 B_27 C_31 I_34 B_16 ST13L ST13S ST13X ST13Y-1 ST13Z-1
120 125 125 142 126 115 133 137 146 111
Act dia 101
Diff From Piece Diff From C.H.
110 117 109
-10 -8 -16
9 16 8
106 122 131
-20 7 -2
5 21 14
Length 2250 2751 2250 2995 2749 848 2250 3598 1000 2696
We discovered one major issue at this stage of sorting, which was not considered in our design development till then. It was the disconnect between the digital data and the physical material. Being out in the sun for almost 4 weeks before we started the sorting, the logs had lost significant amount of moisture and had shrunk in diameter. Our digital model
had not considered this change in diameter and the courses were determined on the basis of thicker depths possible to extract from each log. This lead to an extensive exercise of checking each and every log diameters to make sure that it will be possible to extract the required course depths. The digital model underwent a few iterations
Theo. DIA Comment Course 10 Target Height 121 I_04 I_03 I_29 E_15 C_06 I_33 J_02 I_36 B_20 E_13 C_42 F_19 ST10S ST10X ST10Y
134 131 131 137 138 138 176 140 135 142 138 131
to accommodate this issue and re-established a more realistic course division. A few logs that were still an issue in the new digital model, had to be replaced manually from the spare set of logs.
Diff From Piece Diff From C.H.
135 130 121 125 145 149
1 -1 -10 -12 7 11
14 9 0 4 24 28
127 142 113
-15 4 -18
6 21 -8
Length 1250 1751 1052 3122 2749 1804 3750 1201.08 2752 2503 1499.08 2749.47
PRODUCTION LINE DESIGN
ef or es
168 170 167
cann 3D s
in fo rm at er
ra te p
Fin al lo
Final processing information
gb yc ou rse
MATERIAL SOURCING COMPARTMENT 12A
DIGITAL DESIGN STUDIO
LOG PROCESSING OPERATIONS BIG SHED & WOOD-MISER
PHYSICAL MATERIAL HANDLING YARD
DESIRED LOG SECTION
The design decision of stacking logs by sawing the top and bottom of the faces horizontally was established in the initial prototypes discussed in the material intention chapter. We also proposed to saw an angled surface or chamfer and cut a drip on the external face of the stacked
wall that would facilitate the flow of rain water to trickle down the wall with ease without travelling to the internal surface of the log wall. The detail proposed in the section design is a radial stress relief-cut on the bottom sawn surface so that there are no cracks developed on the top sawn
surface. Each of these details developed further and are discussed in the following section. We developed a series of tools that helped in achieving the desired log section.
VERSION 1: MODIFIED CHAINSAW
The first tool developed to facilitate the desired log section was a simple but efficient modified version of the chainsaw, designed and made by Edward Coe, workshop assistant at Hooke Park. This tool was designed in response of a requirement to cut horizontal cuts on a curved log. In the material workshop earlier on in the project, we had tried sawing the horizon-
tal surface at the wood-miser and the band-saw both. The wood-miser had the limitation of cross-section which was not sufficient to accommodate the complete log with the curvature. The operator would have to adjust the log half-way through the curve to fit in the fence. The band-saw was fundamentally the wrong tool with the cutting blade at right angles to the desired cutting plane.
The chainsaw on wheels was flexible in its movement to work for any curvature as it does not have any fenced boundaries. The adjustable angle of the cutting plane allows for any angle of the chamfer surface to be achievable.
Cutting Rail Chainsaw
VERSION 2: MODIFIED BANDSAW
This version of the tool designed and made by Charlie Corry-Wright, workshop manager was in response to the chainsaw version being dangerous for students
to use and could only be operated by a chainsaw licensed person. Taking the concept forward of a horizontal cutting plane that can be rolled along any
curvature of the logs, Charlie modified a band-saw to cut horizontally by mounted it on a steel frame on rollers, and can be operated by one person.
There were many details designed for ease of operation at this stage. The welded steel frame was light but strong for minor lateral kick-backs from the band-saw. The height of the cutting blade was adjustable with a screw turner. It could be started with a press button like a usual band-saw and lockable rollers if
required. There were two primary limitations of the band-saw version, the limited fence size being one and the other the power of the band-aw motor. The power of the tool was a major issue because this model of band-saw is not build to saw logs, it is build for smaller workshop timber jobs.
Also, the size of the blade is very thin. All this affected the efficiency of the process because either the motor would heat up quickly and stop or the blade itself would snap if pushed a bit faster. Testing this tool prototype, we realised we will require something more powerful to saw so many logs.
VERSION 3: FRAMED CHAINSAW
The final tool version developed and used to process the logs was a framed electric chainsaw on rollers. Chainsaws are powerful and are the right tool to saw logs. To make it safe for students and volunteers to use, it was mounted on to a metal frame with special foot fencing that keeps the operator away the cutting blade. As the operator is not holding the chain-saw directly, and starting it with a gear start handle on the frame, anyone could use it safely. To add another layer
of safety, the operators were required to wear chain-saw trousers. To speed up the process, Charlie made two of these framed chainsaws on rollers, one fixed for the horizontal sawing and the second fixed at 30 degrees for the chamfer cut. The heights of the chainsaw were still adjustable like the previous version. As the chainsaw blade is essentially a cantilevered blade, the horizontal sawn surface was not accurately horizontal.
This was an issue as we didnâ€™t want to add another planning operation to the process to achieve a good finished surface. To resolve this issue, the farther end of the blade was held in position by rollers attached to the steel frame, guaranteeing a horizontal cut plane. These rollers also acted as kick-back protection so that the end does not damage the logs.
Before starting any of the operations/ processes, each log was carefully checked to be the correct match as per the digital model. Both the ends of the log were marked clearly with the log name, course number and an arrow sym-
bolizing the direction of the chamfer. The digital data provided the maximum slab depth possible from each log. To achieve the optimum depth for each course, we started the operations from the smallest predicted log depth in the
ALL OPERATIONS TO ACHIEVE DESIRABLE LOG SECTION
course, and matched the fatter bigger logs to the real depth achieved after processing the smallest. All operations except Operation 3, i.e the bottom flat cut that happens at the wood miser, everything else is carried out in the Big Shed.
OPERATION 1: TOP FLAT
STATION SET-UP & TOOLS Levelled station deck with 3 raised dogging blocks Metal rods fixed to station Wooden blocks/ wedges to adjust heights on dogging blocks Horizontal Chainsaw on wheels
Hand Drills & Screws 2 Laser Levels Wood ‘ruler’ Tape measure, Pencil, Chalk, Marker Ratchet Straps
There is a natural taper in the logs, from the trunk to the top. To achieve a consistent course depth, two flat surfaces were milled on each log. The clean horizontal elevation lines also give a formality to the language of the building. This decision helps maintain a horizontal datum with each progressing course. The flat surfaces also help better and stable connectivity between courses.
The top flat is the most important operation as it sets the datum for all other operations. As most of the logs are slightly doubly curved and handling round logs can get tricky without any reference planes or points to start with, we extracted the triangulated distances to three points on the desired portion of the log to best achieve the accurate planar orientation.
Each course was drawn up indicating the logs named with triangulated dimensions. This drawing was pinned next to the operation deck. As soon as the accurate achieved course depth was determined from the smallest log, the predicted course depth in the drawing was updated and the remaining logs were worked to achieve this.
METHOD STATEMENT FOR OPERATION After a few trial cuts with the project staff, following is the method statement for this operation: Step 1 - Cross check the orientation marked on the logs from the drawing. Step 2- Place log on the 3 dogs, in correct top-on-top position, do not secure Step 3 - Adjust the log as per triangulated dimensions in the drawing, using the
wood â€˜rulerâ€™, tape measure, pencil and chalk Step 4 - Set up the lasers to find the cut lines in the log as per course depth. Adjust height of packers at the three blocks to best achieve depth. Step 5 - Secure the log by hammering it into the dogs and fixing the dog blocks to the station deck
Step 6 - Secure the ratchet strap around the logs and the fixed metal rods on the station deck. Step 7 - Adjust the height of the chainsaw blade to cut through the top laser line accurately Step 8- Cut
OPERATION 2: CHAMFER CUT
STATION SET-UP & TOOLS Operation 1 station Tape measure, Pencil/ Marker Angled Chainsaw on wheels
The prototypes tested different angles for the chamfer. The final decision was to make all angles constant at 30 dergees. As the primary function of this surface is to shed off water, it was decided to plane this surface smooth on-site. The relationship between ‘a’ & ‘b’ was tried and tested. Making any one of them constant does not work because of the
changing width of the logs in plan. Then the ratio a:b was tested to understand where to make the chamfer cut. The final ratio derived through testing was 50:50 in plan, where in ‘b’ is the chamfer surface. This surface after final planning would result in a 60:40 ratio of a:b. This operation was carried out with the same station as operation 1. The angle of
the chainsaw is fixed at 30 degrees. The line from which the chamfer cut starts is drawn up on the top surface. This is done manually by measuring up the ratio along the length of the log, cross checking the orientation of the log. Finally the chainsaw on wheels is pushed on this line for cut.
OPERATION 3: BOTTOM FLAT
STATION SET-UP & TOOLS Wood Miser Tape Measure
This is the operation that realizes the course depth. The whole course was taken to the woodmizer together and cut upside-down to form the flat. Like in operation 1 of cutting the top flat, this is also started with the smallest log. The top flat provides a stable surface to rest
the log on to cut the bottom surface. After successfully achieving the course depth for the smallest log, the rest of the logs were processed with good speed with the blade set to one depth. Making this second flat cut at the wood miser was a conscious decision to achieve cleaner
cuts than the chainsaw and also to speed up the process. The processed logs were then taken back to the Big Shed for the rest of the operations.
OPERATION 4: RELIEF CUT
STATION SET-UP & TOOLS Modified Small Chainsaw Tape Measure, Pencil/ Marker Construction Horses Clamps
Round timber sections/ logs shrink radially when they dry. Cracks on the top surface would catch water and could potentially rot the timber. To avoid any cracks to develop on the top milled surface of the logs, a 10mm relief groove
is cut all along the log on the bottom surface. This releases the stresses in the log and the crack gets accommodated in the groove. This operation is done by first firmly clamping the log, bottom surface up, to
a set of horses. The line for the cut needs marking manually, center of the log. Then the cut is made with a chainsaw mill.
OPERATION 5: DRIP CUT
STATION SET-UP & TOOLS Operation 4 station Circular Saw
This operation is also done to protect the timber from water. A 6x6mm drip is cut along the whole length of the logs. This is required so that water does not
travel towards the underside of the logs and and find pockets that will keep the moisture in.
OPERATION 6: DE-BARKING
STATION SET-UP & TOOLS Construction Horses Draw Knives
The bark of all the logs had to be removed. This is done so that the bark would not catch water or have any growths that rot the logs. Draw knives were used to remove the bark De-barking was the most physical of all operations. With our design process involving new technologies like 3D scanning, and form exploration in the digital scripts, the physical nature of the task was a completely different spectrum. It
was both hated and enjoyed as an operation by the summer volunteers, Hated when a day long peeling would lead to blisters and back ache, and enjoyed as a purely manual job that can be done while listening to music and building
some muscles. It was also the most time consuming operation. Our concern as designers of the project was to get as all logs peeled while we had the labour force.
With numerous amount of logs, each going through six operations each, keeping track of the progress was very crucial. Above excel sheet is an example of data management where we take as account
of the processes. As each course was processed, we had to record the final course depth achieved by each course. This data was fed back into the digital model as the building height gets affect-
ed by every deviation from the predicted course heights. It was very satisfactory to tick off courses as done and ready to go on site.
TESTING SIZE OF GAP BETWEEN LOGS
Our desired log section prototype and timber knowledge established the requirement of gaps between logs to allow air ventilation and keep the timber dry. Having these gaps in the stacked wall between each layer of log made the wall very porous to external elements of rain
and wind. Driving rain could be a potential issue if the gaps allowed it to ingress. Due to these factors, we tried and tested different sizex of gaps that would be most suitable for this kind of construction. We did this by simulating rain and checking the water ingress for different gaps. The
butt joint between two consecutive logs was also a potential issue area. Our final resolution was to seal the vertical butt joints with expandable water-proof foam and the horizontal gap was decided to be 15mm.
Being a unique design, the building has very few traditional details, We designed and made details as per the requirement on site. The above images show such a
case. The corners of the building were with a strip of expandable waterproofing butt joints. We wanted to keep the joinery foam running diagonally in the middle. simple and waterproof. we proposed a butt joint detail joined with screw fixings,
We tested various ways to connect stacked logs with gaps. The above prototype tested wedges as the connecting detail that would also allow the logs to step in and out .
With the log stacked wall stepping in and out constantly, we were not sure if the wall be waterproof at places where the wall moves inside drastically and the drips no longer fall over chamfer surface of lower logs. Due to the inaccuracy in building stacked wall, we developed several waterproofing details that could be applied locally at the problem areas. The sketches above are a few proposals we tested to make the wall waterproof.
Fortunately, by various strategies like removing thin logs at problem areas and always ensuring that the drip works, we have achieved a stacked wall that is waterproof.
LOG FINDING FOR BUILDING CORNER
We were quite often questioned if we would have been able to achieve the same form of the building without any digital tools. Sculptor Andy Goldsworthy â€˜s work is example of achieving highly complex and beautiful structures or
sculptures that do not use digital form finding tools. We tested manual log finding for the north-west corner of the building. Even though it was a small task involving four logs, it was not very convenient moving
big heavy logs till they fit as we want it to. We were able to achieve smooth curving corner logs but the process was very inefficient as compared to the digital scripting process.
REALIZATION OF PROJECT
LOG POSITIONING STRATEGY
LOG FIXING STRATEGY TANK UPSTAND
This chapter is an account of the on-site construction of the project. The timber wall had been the primary design focus of the project but the other components are detailed out and resolved at this stage. The building as a complete entity takes shape here. Till this stage we had tested a method to place logs at the correct position, tested connection details, designed the concrete base, designed
the roof, designed all possible details we could foresee and prepared the project schedule accordingly. We designed and thought through most issues or situations that we might face during the on-site execution of the project and had some crucial decisions nailed before starting work on-site. A few components of the building, like the door and balustrade on the terrace were designed and
made at a later stage of construction. The connection details for the log wall were constantly improvised on-site with decisions evolving as we learnt more from the actual construction phase. The building can be fundamentally divided into 3 components, the concrete base, the timber wall and the flat roof.
Floor Slab : The floor slab was cast, with the concrete extending a little over the floor area so as provide a footing for the external skin of the formwork to be
fixed on to. As most of periphery apart from the entrance is backfilled, achieving a clean concrete base line was not a priority. We cast the floor slab with the
excavated ground as retaining periphery and a slab formwork for the entrance for which we required a clearly defined profile.
The tender for the concrete works was given to Beacon Foundations who are a local company in Dorset. It was very important to achieve a precise concrete base, as defined by our digital model. We
were not very confident about the level of accuracy possible to achieve for the curved walls by Beacon. It was therefore decided that we will do the formwork for the curved part of the wall in-house,
and that straight lengths that are fairly standard, Beacon will do the formwork for them.
This was also the stage where the time schedule of the project started getting affected by the efficiency or inefficiency of other parties involved. The complete
concrete works got delayed by 2 weeks which affected everything else. In the whole scheme, we had only 2 weeks as contingency which got consumed from
the very beginning of the construction on-site, putting the whole team under a lot of pressure to deliver the remaining components of the building on time.
Formwork Detail Design: This involved extracting all the curved wall profiles, flattening them digitally trimming them to the sloping plane and producing very
precise CNC cut ply-wood profiles. The curved wall formwork was produced in handle-able, smaller lengths that could be put together on-site eventually. The
formwork pre-fabrication was an extensive 3 week job, very carefully made so as to achieve a well finished, smooth concrete wall.
The concrete, 60 cubic meters, was poured in half a day. It was done in two loads which is currently visible as a horizontal line in the interior of the building. The formwork build in-house started
getting some bulges at the back retaining wall due to the concrete load. But the issue was immediately handled withe the on-site team by adding more strengthening struts while the concreting was
carried on. Removal of the formwork after the concrete was set was also done by our on-site team.
The final concrete finish is not to our satisfaction, with a few areas that required immediate repair. The reason behind the bad finish was probably the relationship between the reinforcement and the concrete mix. The steel reinforcements for
the wall was very tight, and the concret mix dry as compared to it. Fortunately, very little of the concrete is visible from the outside and hopefully the timber wall from the interior of the boiler house will keep eyes from going to the concrete
finish. The possitive aspect of this stage od work was that we achieved an accurate concrete base to start building the timber wall on.
Construction of the timber wall was a constantly improvised and intuitive process. because the precise postioning of the connections between the logs had not been predetermined. Many subtle
evolutions to the details occured through the process. There are essentially 3 categories through which we describe the wall construction. First is the log positioning i.e how each log is placed at its
position as defined by the digital model, log placement i.e the fixing strategy of these logs to form the wall and finally the connection details itself with packers and wedges.
One of the most important details of the building, was how the logs get fixed to the concrete base. The 200mm concrete wall top surface has a slight double curve
because of sloping from north to south and curving in plan. This implied that the logs sitting on the concrete had to be shaped to this complex curve, while
This detail sketch was one of the previously pursued proposals. We were concerned about water ingress into the building through the gap between the cill and the log. In this proposal, a metal strip cut to exact size would have closed the gap. But we got a reality check regarding time required to groove concrete and
cut metal to size while prototyping this detail. The final decision was to close the gap with expanding waterproof foam. It has proved to be a simple and successful detail till now.
always making sure that the drip cut is out of the concrete line.
These images show our first attempt to shape the log to fit the concrete base. We used a few different tools, the chainsaw, chisel & planner, to arrive at the desired result.
This detail proved to be way more complicated than we anticipated, with numerous markings, planeing and levelling stages. As the cut was bespoke for each
log fixed to the concrete, it took many dif- mizer, and then it was about finding the ferent approaches and tests to arrive to exact surface required by slowly shaping a standard process. One primary cut was it with a planar. done with the band-saw or at the wood-
LOG POSITIONING ON SITE
The exact positioning of each log was crucial to achieve the desired form. The digital model gave us the position of each log in the building. To construct on site and find the exact position of the log as defined digitally, we provided two kind
of information to site, course plans and vertical building profiles. Firstly, We extracted the building profiles at every 1500mm intervals from the digital model and CNC cut them in plywood. We then fixed these guiding profiles to
the outside periphery of the concrete with posts to the correct heights. The logs were constantly placed with reference to these guides.
Secondly, we also provided the siteteam a set of course plan drawings, that located each log and also marked each
meeting positions of the logs. With this information, it was possible to gauge how much of the over-length logs would
get used for the wall. The plan below is a sample of the drawing provided on site.
The CNC cut profiles were a simple and efficient system, fabricated and installed easily on site. It was important to achieve
accuracy in placement from the very beginning as each subsequent course would get affected by any inaccuracy
in the course below. Each course was constantly checked for level while building up.
Apart from the profile and plan position, it was also important to place the individual logs with the guiding rule that the drip should work i.e the drip should always be above the chamfered surface of the log below. To achieve this, sometimes we had to chamfer the lower log
more, or adjust the upper log to make the drip work. As the log wall started building up, the human adjustments and inaccuracy to the log position started affecting the predicted wall profile. Another factor that affected the accuracy of the system
was the variation in width of the logs itself. This was because of the difference between the digital data of the logs and the real physical log.
LOG COURSE FIXING
All the logs processed were 200-300mm longer than the length required for the wall. This was again a difference between the digital scanned data and the physical logs. We used it in our favour by developing a system of log positioning and fixing that would give us very clean vertical joints between logs in the same course. As seen in the image below, the strategy
was to alternately stagger the logs up and down, with a special temporarily supporting jig of same dimension as the horizontal gap (15mm) in between the logs. The lower level of logs are positioned with the help of the guiding profile and temporarily fixed. The supporting jig is screwed on to the lower logs and is longer in width than all the logs so that
the upper level of logs can be adjusted to find accurate position by sliding on it. Then a single cut position is defined to trim off the ends of the two logs and allow them to meet neatly end-to-end.
METHOD STATEMENT FOR LOG FIXING- USED ON SITE 1. Mark start and end cut on concrete for each log. (According to drawing set 1) 2. Move the next course logs to site. (Check names before moving according to drawing set 2) 3. Put first layer (every alternate log) temporary supports on the course below. (Put in place where is flat) 4. Put first layer logs on (every alternate log, and the log on the concrete should be in second layer), position requirements as below, i. Position along the wall referring to the mark on concrete ii. Position in-out referring to the section profiles iii. Check level both ways iv. Temporarily fix to the course below with timber lock screws
5. Put second layer temporary supports on the course below and put logs on. Position as requirements above in Process 4. 6. Mark positions with pencil on logs. 7. Find the final cutline where the outlines of logs meet in the same place. Mark the line from outside to inside using bending ruler. 8. Find and mark angle cutline for the ends on concrete slope using ply support bracket if needed. Send to wood-mizer. 9. Check stability and ready for chainsaw cut. (Make sure the log is supported for all individual cut. 10. Use chainsaw to cut ends. Trim any splintered edges. 11. Take down the second layer logs and put packers on ( as detail drawing shows
below) Also check the relief cut and grooving. 12. Put second layer logs in position and put foams in between each log and under the log on the concrete. 13. After fix the second layer logs, then fix the first layer in the same process. 14. Fix logs through packers or wedges to the course below. 15. Plane top if needed and plane chamfer to join. 16. Check height of each course and record. 17. Temporary brace/shore any overhanging courses.
The consecutive log courses are fixed to each other with metal screw fixings going through the packers that also maintain a constant gap between each course. There are two kinds of connections, the
oak packers and the wedges. The packers are used for conditions where the logs are almost above each other. This is done by first fixing packers of different lengths to the underside of
the log that needs fixing at an average of 400-450mm centers. The log can then be fixed to the wall below with screws or timber locks.
The wedges come into play when the logs start stepping in and out of the wall below. We used wedges on both inside and outside of the wall, depending upon where the support was needed. The
building has many different sizes and ways of fixing wedges as it was a constantly improvised process. Some of the wedges were added after completion of the stacked wall, where ever it seemed
required intuitively. This wedge detail is one of the special components of the project. It was developed because it was one of the most versatile systems that allows for any deviation in the log stacking.
The number and position of the packers and wedges has no account. We kept on adding more of the connections intuitively where ever the wall was stepping
in and out drastically. An interesting proposition would be to scan the whole building for timber locks and screws. In retrospect, it can be said that the whole
building is held together by these small but numerous amount of metal fixings.
THE ENTRANCE DOOR
The entrance of the building is designed to be a hidden gateway, therefore the door to the boiler room had to be something that become a part of the building
with the same language. Our proposal for the door was a simple pivoted log door with a metal frame. The distance between the short wall and the long wall
is measured for each course and the logs for the door are sourced from off-cuts of the wall. Its a light and simple, but a very efficient design.
The roof is a simple flat roof system, with 200mm X 200mm timber joists spanning the two rooms, sloped towards the south. The joists are spanned with 51mm structural kerto sheets, and then waterproofed with an EPDM membrane. Although the roofing system is quite standard, the
proposed linear arrangement of joists and the curving log wall meet at awkward angles. This is the primary issue and interesting aspect of the roofing system. We proposed bespoke flitch plate joist hangers to support the joists on the wall. The 15th course is intentionally thicker
because of two reasons, firstly to provide a stronger course to hang the joists off and secondly to accommodate the fall of the roof surface to be contained within one course. The hangers were fixed on top of this 15th course.
Even after getting the bracket structurally specified, we had to wait for the wall to be complete till the 15th course. This was because we were doubtful about the accuracy of the wall as it progressed. We wanted to confirm the exact angle the joist will meet the log before ordering the bespoke joist hangers. Surprisingly, the
angles derived from the digital model and eventually from the 15th course were very close. The hangers were fabricate and delivered in a weekâ€™s time for work to continue on the building. The joists were temporarily fixed to one end of the joists to cut the flitch and predrill the bolt positions. Then the hangers
were fitted to the wall. The joists were lowered into position with a tele-handler and fixed to the hangers on the wall. All joists are attached to the hangers on one end and are rested on the middle masonry wall at the extra length end which were cut to size once the joist was in position.
The timber wall till course 13 had open the timber structure very stable. The back ends, and therefore was a bit shaky. wall was the only length which had the Course 14 and 15 form a tie beam and joists meeting at right angles. lock the whole structure together making
Hooke Park had a few sheets of kerto left-over from some previous projects. We wanted to use those sheets for our
roof. They were not all in standard sheet sizes. The image above shows a papercut exercise to figure out how best to use
the available material. Fortunately, we had just the right quantity of sheets for our roof.
The balustrade for the building was discussed from the start of the project but no decisions were taken till the very end. Our design intent always with it was to make it as harmonious as possible with the rest of the log stacked walls and also to make it transparent for visual continuity into the forest from the building terrace. Both intensions were conflicting in nature because making rails with logs wonâ€™t be as transparent as with other
materials. We also considered making the balustrade component something transparent and simple with glass and metal, which reads separate from the building. But this idea was dropped because we wanted the continuity of the logs till the top, bringing in different material might have reduced the aesthetic richness of the log wall. Considering making the balustrade with curved logs, all these sketches show
different proposals. As visible here, with everything else curving and adding to the complexity, we considered making the posts vertical and explored many options of how to connect the curved logs to straight posts of different sections.
The final proposal was to plasma-cut metal posts bespoke to the curve log courses pre-determined in the digital
model and processed during the summer. As show in the sketch above, the post has welded plates at the correct
heights to pick up the curved logs. The posts get fixed to the kerto roof with screws and gets coved with decking.
The discrepancy between the digital and the physical also affected the balustrade. Although at this stage, having finished the log wall with 15 courses already, the
on-site team was more confident about handling the logs for 4 more courses, one fixed to the trimmer 15th course and 3 attached to the bespoke posts. It was
a challenging task on-site to match all these various curving components but the result is harmonious to the initial proposal.
THE CHIP STORE DELIVERY HATCH
The hatch over the chip store for chip delivery is one of the most functionally crucial components of the project. It has been designed for optimum functionality. The primary issue faced for its design and making was to keep the height of hatch as low as possible to comply with our design ambition of visual connectiv-
ity. The size of the hatch opening is 3m X 2.5m, which meant that the spanning cover would have some significant depth. With the engineers, the roof structure is designed to be minimum depth possible. Another issue with chip delivery is that the back wall had to be less than 450mm for the delivery truck back not to hit the
wall. Our back wall was higher than that. To resolve this, we attached the excess height of the wall, i.e 3 log courses to the bottom of the hatch, that can be slid open with the hatch.
THE BUFFER TANK UPSTAND
The upstand on the roof is the cover for the buffer tank in the boiler room which is higher than the rest of the roof level. It essentially is a waterproof cap for the tank. The form of the tank is derived from
the bend curves that were not used for the log wall. The proposal was that these curved logs can be transformed into seating around the upstand, as the only outdoor furniture in the designed social
space of the roof. This proposal is still under consideration. It is proposed that the upstand will also be covered with the same decking as the terrace.
The roof is designed to slope towards the south, i.e the entrance of the building. After a few different design iterations, we decided to make the drainage a simple spout at the lowest point of the roof. The final spout is a black painted metal
square section, sticking 1000mm from the wall, that would take the water sufficiently away from the timber wall. The drained water would fall on a gravel bed and would be taken to the other side of the entrance road from there.
WALL DETAIL SECTION
C17 C16 C15 C14 C13 C12 C11
Timber Stacked Wall
C9 C8 C7 C6 Detail 3 C5 C4 C3 C2 Detail 4
DETAIL 1 C17
shelf plates welded to upright 20mm x 60mm plasma cut steel upright joist hanger fixed with timber locks
base plate 250mm x 200mm fixed to kerto with M10 bolts 27mm x 125mm decking 25mm x 40mm decking battens @ 400mm centers 10mm foam & 2mm EPDM
51mm kerto C14 200mm x 200mm joist C13
flitch plate joist hanger
DETAIL 2 15mm oak packer
C10 timber locks screw fixing C09
15mm oak packer fixed to logs with screw or timber locks sealed with silicon and wooden dowels recess for concrete fixing 10mm relief cut
6mm x 6mm drip cut C06
expanding foam for waterproofing 20mm thick nylon packer
15mm oak packer C05
screw fixing timber lock
log fixed with concrete fixing landscape taken to 150mm below cill level
200mm concrete wall C04 SS
There are six arguments key to our research, following are the conclusions we derived through their exploration: 1. LANDSCAPE IS RICH - CONTOUR THE BUILDING We hoped to build a modest building to appreciate the natural landscape. We were inspired by the site topography which informed the initial form of the building. The contour terrace proposal offered a new social area and it turned gradually and naturally into the envelope of the boiler house. It embedded into the slope and keeps the view communication between the north and south parts of the campus. 2. TREES HAVE FORM- SCAN THEM We chose naturally bent timber as our experimental material. To get the irregular tree-information,we used 3D scanning technology to convert it to digital data.Compared to other 3D scanning equipment used for architecture, the scanner we used was cheap and efficient. It enabled the use of irregular natural form to be recorded and tested, giving new value to this novel material. It increases the scope of formal exploration of any irregular material in architecture. 3. LOTS OF CURVES- LET THEM FLOW
Designing the building was an iterative process. Many evolving versions of the wall were defined, as we improved our understanding of both - our digital strategy and of the trees themselves. The result is an organic geometry,which expressed the natural character of the material. The natural form of the trees found their place again, flowing in the Hooke landscape. 4. DONâ€™T HAVE THE RIGHT TOOLS - MAKE THEM Dealing with irregular material was tricky, and we didnâ€™t have the proper tools to process them. We used existing tool resources to solve our new and specific problems. We learnt tool operation, making skills and material properties from these iterations of testing. 5. TRICKY DETAILS - LET HANDS THINK Details were worked out mainly through the process of modelling and prototyping. We encountered a lot of practical issues during the construction phase of the building, of which many details were resolved on-site. This was a conscious decision to let our hands contribute to the process through tactile learning. 6. MATERIAL IS DEVIANT - NEGOTIATE One of the primary learning from the use of deviant material is that it is dif-
ficult to be precise and determine how the material will behave. It is a special quality of deviant material that can be used to work in a more intuitive process where the material itself determines the direction. Negotiation is an essential part of any project, between team members, with site conditions, with pragmatic requirements and many other things. Designing with deviant material meant that there is a constant negotiation during construction as well. This portfolio is a complete account of realised & unrealised proposals, successful & unsuccessful ambitions and tested & evolved material explorations. The building is the physical manifestation of the project but it is the holistic experience of the programme that is of value. The excitement and frustrations involved in pursuit of an idea, the joy of making and the moments of surprise in the final building, everything combined has contributed to an enriched academic journey. This project has been way for us to pursue individual learning and arguments. The project has been intelligently reflected in our individual theses, the abstract for which is page opposite.
CONTINUUM OF ARTEFACTS: RE-EVALUATING DESIGN PROCESS THROUGH THE TACTILE This thesis proposes that the architect should prioritise the physical and tactile in design. The premise of this paper is based on the value of retaining a sense of the physical qualities of the material throughout the design process. The Hooke Park Biomass Boiler House project is presented as evidence for a potential “continuum of artefact” where the use of natural material prompted the production of physical entities that share an instrumental status in informing design decisions. Properties of materials and their workability manifest themselves in architectural qualities. Similarly, landscape as a physical entity is omnipresent, contributing to design decisions in conjunction with material itself. Thus it is proposed that engagement in the
MATERIAL AS MUSE: NEW APPROACHES TO NATURAL GEOMETRY IN ARCHITECTURE This thesis proposes a new design methodology that is driven by irregular natural materials and form. It proposes that emerging digital 3D scanning technologies will enable a renaissance of the use of non-standard natural material in architecture. It discusses traditional logcabin construction, which was restricted by the standards of materials and methods of construction, before examining other traditional techniques that exploit natural organic forms, including boat building crafts and Medieval cruck timber-framing building. These lead to the central example of this thesis, the Biomass Boiler House(BBH). This is an experimental Design and Make project built at Hooke Park that used 3D-scanning technologies to utilize and process bent timber, and explores new possibilities and the relationship between nature, technology and architecture(Figure 1). Throughout history, architects have been enthused by the use and exploration of natural materials and processes within
physical landscape, process models, prototypes and indeed the building itself during construction, are all valuable as part of a design process. The thesis analyses the decisions made in the Boiler House project to demonstrate the value of making design decisions throughout the execution process, however, the professional framework restricts such explorations from being stretched over the complete time-line of a project. A design and build way of practice is proposed as an alternative that presents an opportunity for in-depth material exploration due to the process of physical engagement that has scope for unpredictability in the construction process. It is therefore important for the sequence of decisions in design and build practice to be sited within professional framework.
Another key issue prompted by this thesis is that in today’s design process, the digital realm often takes precedence over the physical. By proposing to integrate new technologies, such as 3D scanning within the process, a new continuity of engagement within the physical realm is enabled.
architectural design. Even though synthetic materials and industrial technologies emerged throughout the twentieth century that dramatically changed the style and scale of architecture, today there are still advocates of natural materials and natural architecture. Compared to the synthetic material, natural materials– such as wood – are more organic and vivid. However, if we look at today’s timber industry, commercial timber supply are entirely limited to standard rectangular sizes. The processing required in producingt this standardized material is not only contary to the tree’s diverse forms, but also its evolved natural performance. Furthermore, the opportunity to explore these natural creations innovatively in architecture is lost.
looked at traditional mainstream timber structure, log-cabin or timber framing building, we will find that they have restricted standard criteria for trees due to limited technology. We doubt about this standard is still adaptable to today’s timber building. BBH is an exploration to prove, through 3D scanning and other technologies, irregular timber could be easily and creatively utilized. We aim to rejuvenate non-standard and irregular natural material in architecture. These technologies open up new questions. What materials are usable? What is the architectural result of using such natural materials? How does this change existing regulations and definitions, which should be rethought? Considered in these terms, the thesis explores such questions by using a extreme and experimental material–irregularly shaped and previously considered unusable timber–to test the confluence of digital and nature and, though this testing, offer potentially new avenue for architectural design.
Today’s new digital technologies could give new life to timber architecture. In the BBH project, for example, 3D-scanning technology evoke new applications of natural bent material, and, through material-inspired design and process, a new methodology of architectural design has been explored and practiced. If we YINGZI WANG
LOG POSITION: FROM FOREST TO BUILDING
This diagram identifies from which area of the various each of the logs in the final finished building were extracted from. Six colours represent the 6 subcompartments of compartment 12A, one the trees re-scanned in the Big Shed that lost their record of location and lastly the straight logs that could have been harvested from any part of Hooke Park forest. A query: While going up and down the slopes of Compartment 12A for three weeks, hunting for all the bent trees we could harvest, it was a question in my head as to why are these bent trees only in this area. Christopher Sadd, Hooke Parkâ€™s forester, informed us from the beginning that these trees are bent because of bracken growing in this area when the trees were young. Seeing the variety of curves in the forest, I wanted to explore all the factors that might aďŹ€ect this bending. I had a theory that needed testing. I wanted to research and find if there was any co-relation between the slope of the landscape and the curves in the trees. To do so I started recording the GPS location of each tree we scanned and the landscape slope angle at that position with a clinometer. It was an attempt to discover any co-relation between the ground condition and the resultant natural form. Unfortunately, I started the exercise just when we started the harvest of the marked tree because of which the trees started coming down faster that I could record its ground condition. It was also impossible to locate stumps of cut trees in the undergrowth of the forest. I dropped the exercise before any substantial data collection. A query left unanswered: The under valued bent logs found a new place. Removed from their natural landscape, they became part of the built landscape of the campus. From vertically standing trees, they became our horizontal design intent.
Legend Subcompartment A
Rescanned in Big Shed