Design + Make’s project for 2015 was a robotically fabricated Wood Chip Barn. The structure provides long-term storage for wood chip to fuel the Biomass Boiler House. With a storage capacity of 400cu.m, the barn enables the Hooke Park estate to process and use it’s own timber for renewable heat production. The first structure to be erected outside of the educational campus, the barn occupies part of a site which had been used for sawmill operations and is envisioned to facilitate an increase in timber processing activity.
The barn’s arching structure is formed from forked beech-tree components directly sourced from the surrounding woodland. The inherent form and structural capacity of the natural tree is transferred and exploited within the truss structure using 3d-scanning techniques and robotic milling to form the connections.
Having surveyed Hooke Park’s beech compartments, a database of potential forked components was established, and from it, the structural concept was developed. Based on the criteria of this structure, 25 forks were harvested from the forest, brought back to the campus and scanned in 3D. An organization script was used to generate a final arrangement of forks in collaboration with engineers from Arup. This digital model was then translated into fabrication information with which Hooke Park’s new robotic arm transformed each fork into a finished component. After being pre-assembled in Hooke Park’s Big Shed, the building’s pieces were assembled on site.
Students: Mohaimeen Islam, Zachary Mollica, Sahil Shah, Swetha Vegesana, Yung-Chen Yang
Tutors: Toby Burgess, Charley Brentnall, Martin Self, Emmanuel Vercruysse
Workshop manager: Charlie Corry Wright
Robotics developer: Pradeep Devadass
Project coordination / Site management: Jack Draper
Forester: Christopher Sadd
Estate Manager: Jez Ralph
Workshop technician: Edward Coe
Erection team: Timothy Hallbery, James Vooght, Aurimas Bukauskas, Summerbuild Volunteers
Robotics support: Farid Dailami (UWE / Bristol Robotics Lab), Johannes Brauman (Robots in Architecture)
Structural engineers: Arup (Francis Archer, Naotaka Minami, Coco van Egeraat)
Video: Pradeep Devadass and Zachary Mollica
With the last phase of scaffolding removed we can finally experience the space.
With the last of the on site connections made, the truss’ scaffolding has been removed. Now self supporting, the truss spans 25 m x 10 m and rises to 8.5 m at its zenith.
Big day – had a mobile crane on site to lift the two halves of the truss in to place on top of the temporary scaffolding supports. Everything fits! Now we need to complete the bolted connection between the truss segments and the tops of the tripod legs.
The first truss-half has been complete and carefully moved out of the Big Shed, making the assembly jig available for the next half. Video: Zac Mollica.
10 forks are currently placed onto the assembly jig along with 6 straight timbers. With all of the pieces temporarily held into their final positions using straps we are beginning the process of bolting together. In assembling, small errors that have accumulated throughout the process from scanning to milling become evident. As well as developing strategies to correct specific instances we have been able to feed back information to the milling of the forks for the second half of the truss which we hope will smooth its assembly.
With the pre-assembly of the first half of the truss well underway, the robotic arm is in full production mode on the forks and top chords for the second half.
Each fork leaves the robot cell as a finished component with all of the connection geometries required to connect it to its neighbours. Due to their large size we required a method to locate all of the pieces with respect to each other before bolting together. A large assembly jig has been constructed which will be used to construct the truss in two halves (Pink and green holes). The robot mills three reference points into each fork which relate to the tops of the vertical supports shown. With information from a 3D model, their location has been set out in plan using CNC sheets and their heights by laser levelled posts cut to height.
The yard has been full of activity these last few weeks. The eight roof panels are well underway and the finished wall components which will eventually be assembled on site are piling up.
The four tripod units are an important part of the truss. As they are being manually fabricated but must align to the precision of the robotically fabricated pieces of the truss, a complex series of jigs and techniques has been developed.
The connection between the end’s of the fork branches and the top chord members has proved the most complex to detail. In the first prototype we had a rectangular mortice and tenon but this proved tricky visually (a lot of the mortice’s surround flat was exposed) and in engineering terms, as it is difficult to get the needed contact area. Instead we have develop an elliptical-conical tenon which best achieves the compression contact area and fits nicely. To prove this we worked up a version manually (using Charlie’s hand skill!) and then with the robot arm.
We have started machining the first forks with the robot arm. The first step (photo 1) is to register the positions of the support trolley by pointing the robot a the top of each of the three support bolts that correspond to the fork reference positions from the scan. We then (photo 2) bring in the fork and check it is in the expected position, working with Pradeep in the control booth (photo 3). There are three main types of connection: making planar end-faces where chord pieces meet end on (photo 4); tenon ends at the end of the branches; and flats where the secondary web members will join (photo 5).
As well as undergoing robotic machining processes, each fork underwent a series of manual operations including: cleaning and boron treatment, and connection finishing. While the forks were digitally scanned, top chord components were manually surveyed to determine their centrelines.
The connection strategy for the truss relies on the principle that we can precisely machine each connection face so that their relative positions are correct. This means that although our scanned mesh geometries are not perfectly precise (we find errors up to about +/- 20mm) we can use the accuracy and repeat-ability of the robot arm to ensure that connections will match.
As shown in these images, we define the connection machining volumes (that the robot will rout out – in red) in the global geometric frame and then re-orientate a connection set for each fork’s local reference geometry the matches the coordinate system in the robot cell. This local geometry is based on the 3 reference holes that we drilled in the 3d scanning process.
It’s been a busy week, this one! After studying the defects, we selected 20 forks and generated what we hope now is the final truss geometry with the 3D meshes. This generation was carried out by Zac using the Galapagos evolutionary solver in Rhino-3d to optimise the truss arrangement. The summer-builders have been busy debarking, recording and marking centrelines on the Larch poles for the final tripod fabrication.
Bath University has been carrying out load tests on forks as a part of an on-going research by one of their students. They have used Beech forks, to help understand the grain structure and stresses the fork would be under in our structure.
Aurimas, a visiting student from MIT doing his research on roundwood timber columns has been helping us understand, design and fabricate the “tripods” which form the four legs of the truss structure. We built a 2/3rd scale tripod mock up to understand the process and assembly sequence for the final production.
Done processing 3D meshes for all twenty-five of the forks! Each fork was marked with reference points photographed from approximately 20 positions for uploading to an online photogrammetry service that returns a 3d mesh of the photographed surfaces.
Phase 1 Summer Builders arrived this weekend and dived straight into project-mode. They built trestles for the prepping the forks to be photographed.
A week spent in pre-processing of the forks – being debarked, prepped, measured and photographed to get the 3D mesh. Students along with the summer builders tested the robot arm checking for accuracy of the 3D mesh data.
Preparing for the AA’s Projects Review Exhibition 2015. The exhibition displays the Biomass Boiler House book hints of our project including a 1:20 model and the 2/3 mock-up.
We are testing our first run of the whole production sequence to build a 2/3rd scale mockup of a section of the truss this week. Connection details are being worked on – using the 3D scan mesh data. Also, fabrication of a lifting gantry and robot cell trolley began – as part of the robot fabrication strategy.
Once the forks are brought to the yard, they are debarked and prepped for being photographed for 3d-scanning photogrammetry. A drilling template was devised to help locate 3 reference points to be picked up in the 3D meshes, to enable us to correctly located the fork in the robot fabrication cell.
Initiated the pre-fabrication process with foresters Chris and Nick, felling the selected trees after locating them on the map with the acquired GPS data. With some of the forks breaking after being felled we’ve been introduced to the possible defects and the difficulties we might have to prepare for.
We’ve been testing techniques for arranging fork geometries onto the target truss chord curves, using scripts that optimise the placement to minimise the angle change between segments.
We had a quick design charette this the afternoon, to converge onto ideas for the overall form with the revised structure.
Using the data extracted off the photogrammetry 3D mesh, we had a trial run, testing the accuracy of the fabrication process with the robot today. Precision being the key to the whole structure different strategies are currently being explored and developed around the scale of the forks, and for it to be processed by the robot efficiently and precisely.
The scale of the forks being the biggest challenge- different strategies and orientations were experimented with. The strategies revolved around capturing the 3D nature of the forks and the reference points marked on the forks, required for robot fabrication.
The 3D meshes from different processes of 3D scanning with the Xbox Kinect and Photogrammetry using 123DCatch , were compared for detail and accuracy. Photogrammetry seems to be giving better results for our scale.
Extracted our first fork from the survey catalogue today with Chris and Edward. One down, many more to go!
On further structural analysis and the reality of the complications that lie due to the high number of forks required for the arch as compared to the number of good forks in the forest – we had discussions with ARUP on a developing a hybrid structure – including forks and sawn timber.
We split into different groups , covering different compartments of the forest to carry out a fork survey and scout for the forks we need. Each tree was marked,numbered and photographed using theodolite. Which then was catalogued and analysed using different rhino and grasshopper scripts.
Zac developed a pretty impressive grasshopper script using geotagging to locate each and every fork in the forest. Preliminary discussions on project timelines, material quantities and cost estimates began. Got a lot of work to do!
After a long break we are back to the grind with fresh minds! We’ve had a big leap in terms of the structural design today – an arched truss! We all jumped on to the notion and started to work towards design strategies and structural details. Vivian quickly modelled up a Y-shaped version of the truss. Looks exciting!
We presented the scheme design in London to a group of critics including Arthur Manu-Mani, Emmanuel Vercruysse and Peter Thomas.
Using Hooke Park’s Kuka KR150 we worked with Pradeep, who runs the robot, on a first test of router machining of a fork.
Using a cellphone app -Theodolite, we’ve found a quick process to photograph and document the forks in their trees. Which then with the help of a rhino script can be further analyzed by extracting the basic geometry and size of each individual fork.
Following the mock-up, we returned to small models. Vivian made a pair of models that we all liked – in which the cross-brace pieces were staggered between neighbouring frames.
Along with tutors Martin self, Toby Burgess and Charley B we built and tested a 1:2 mock up of the forked-spine roof structure – using forks in the ridge and as rafters.
Realising that to allow a multi-directional moment connection with wood – it is easier to create an in-line moment than an out–of-line moment An arched branching roof structure was developed from the geometry of the fork, where the primary use of the forks is in the spine of the structure. Potential proposal is to erect the structure as a series of rigid tripod frames, resulting in a split-arch structure. The design idea evolved with functional inputs from the chipping company and structural inputs from ARUP.
We presented our different ideas and work-in-process today to a panel of tutors and local architects, including Niall Jacobsen and Keith Brownlie.
In the process of evolving the use of tree forks structurally and more than ‘just’ trusses , we started working on the idea of using them as components used in boat-building . The concept of a large shell structure was derived upon,using grasshopper and kangaroo to digitally develop the form using the principle of a hanging chain model.
We have been studying options for the pushwall.
Drove down to visit the Tithe barn at Bradford-on-Avon, followed up with a visit and a tour around Carpenter Oak with Charley B. Also made a short trip to understand traditional shingle making. In all, a good day!
We have been building full-scale mock-ups to test possibilities for the shingle roof system. Different arrangements of shingles are being tested along with different species and cutting strategies.
To get an a better sense of what it means to handle and machine the forks, we have been testing various fabrication processes in the workshop with medium-sized beech forks.
With advice from Satta and Yingzi (with their experience of 3d scanning for the boilerhouse project) we have be testing the feasibility of 3d scanning the forked components.
A busy week spent in understanding the different ideas for systems and materials through sketches,models and prototypes.
Timber wall and floor ideas were explored. Thatching as an idea was ruled out and shingles were to be explored further.
Tree forks of different sizes were gathered from around the forest, 3D scanned and used to build on the idea of trusses . Met with engineers from ARUP for the first time and discussed our ideas for the design. Shingling mock-ups were made and analysed.
Design research, site survey and siting discussions continued throughout the week. After having researched and spoken to a couple of arboriculturists about properties of tree-forks. We ventured out today into the forest to grade and understand ‘good’ forks and ‘bad’ forks.
Charley and Kate came down today to discuss the ideas for the new project. We all presented our research and set of ambitions. Prototype ideas of using tree-forks were well received by everyone.
We split into two groups- One went out in the forest along with Charley B to scout and get a reality check on tree-forks in the forest. The other, along with Kate and Toby made quick glue-gun concept models using small natural forks. We all presented our ideas and findings of the day and concluded the idea of using tree forks as one of the primary ambitions to follow. The five of us continued our individual research on about systems and techniques for – the floor, the pushwall, the structure and the roof.
After having visited a few woodchip stores, we put presented out thoughts, ideas and concepts for the new project. Ideas were picked from the panels and narrowed down to a list of basic ambitions and ideas to be pursued:
- Using Hooke park timber and resources
- Explore Shingles / Thatching
- Minimal Plastic and Concrete
- Use the new robot arm for fabrication
- Smart and efficient structural systems
- Old agriculture buildings as precedents