A rip cut is a fundamental type of cut in wood parallel to its grain structure. Because the three projects in this series incorporate bending and twisting of their wood members (stressing the wood structure) to give them form, they all share the need for a consistent parallel grain and thus the rip cut was the primary process in producing the components. Without this consistent parallel grain or grain runout (where the grain runs off the edge of a board) the linear wood pieces tend to split and break. Certain measures can be taken to minimize this, such as selecting good boards or steaming the pieces prior to bending, but in the end it is a natural material and so limitations and potential flaws must be embraced. These projects aim to strike a balance between the craft and risk of working with such natural materials and the ‘control’ and precision of digital fabrication methods.
The three projects in this series - the LAMELLAE Screen, the STRAIN Lamp, and the SHEAF Lamp used wood from the three same walnut boards. For the largest - LAMELLAE Screen - slats were cut first, then the remaining stock was used to cut strips for STRAIN Lamp. The SHEAF Lamp was designed afterwards, specifically to make use of the off-cut or scrap pieces from the first two projects.
Step 1: LAMELLAE Screen
The LAMELLAE Screen builds on ideas explored in the LAMELLAE Series and is the first prototype taking these same ideas to a larger more spatial configuration. As a singular piece, it was designed as a small room divider/visual screen, leveraging the twisting capacity of the custom sliced walnut slats to modulate visibility, shadow, and depth. As a potentially larger system, the screen can be modified (through the parametric model) to accommodate any number of ‘local’ requirements such as level of visibility, density of slats, degree of twist, or amount of undulation. For this first prototype, the slats were formed as-is. By incorporating steaming to the process a higher degree of bending and forming could be achieved.
1/2” plywood / walnut slats
TOOLS/SUPPLIES: Table Saw Metabeam Laser Cutter ½” Plywood 5/4 Walnut boards Wood glue Epoxy Clamps Zip ties
Step 2: Design Process
Prior to developing the digital model, a number of tests were done with proportionally varied sections of walnut to experiment with flexibility - using both dry and steamed pieces. This gave a better, more intuitive sense of the material limits when designing digitally (this is essential as material resistance is conveniently absent from the digital realm!)
A digital model was developed and tuned parametrically to accommodate specifics of the wood I was to later rip down. Parameters allowed customization of the # of slats (spacing), depth, width, bunching, degree of twisting, and amount of displacement or translation out from the face of the screen. These parameters were adjusted at 4 datums (the levels which were to become the formwork or ribbing. Once all of this was set, the ribs could be ‘extracted’ as flat .dxf files with slat notches and numbering for precise alignment.
Step 3: Fabrication
With these .dxf files, the ribs were cut from ½” plywood on the Metabeam laser cutter at Pier 9. A few test pieces cut from ¾” proved too crude in cut quality, so thickness was dropped down to the ½” size. This ribs were also notched on the backside to key into the two vertical rails. This framework ‘assembly was laid out flat, glued together and clamped in place.
Next (3) 5/4 Walnut boards were crosscut to length and then ripped down into XX 3/16” wide strips (painstakingly chosen for the least grain runout I could find at the time - chosing boards is always an organic process and sometimes you have to go with what you have or otherwise wait for another shipment!)
With the frame set and the slats or louvers all ripped, layout and gluing began. * It's worth noting that through the entire process, ripped pieces were kept in strict order so that any variation in graining/features read as a subtle change across the screen and not randomly selected slats. For the most part this was a relatively straightforward process: put down wood glue into the notches for about 2-3 slats at a time, lay them in place, and then zip-tie at any connections where twisting (out of the notch) wanted to occur. A few connections required additional clamping/ties (more about this in a moment). After allowing ample time for the glue to cure, the screen was set vertically (the entire system is incredibly lightweight) and any zip-ties were removed. The connections where much stress was present began to pull away from their notches over the period a several days. Any small gap between the walnut slats and the plywood notches was a poor condition for wood glue (which necessitates a tight fit up). Those small gaps were filled with epoxy and the joints re-clamped. This provided enough hold for those difficult moments.
Step 4: Conclusions / Speculations
For a ‘prototype’ system this turned out quite nice and was great to see this technique (previously used at the scale of lamps) at an interior/furniture scale. The density used for this piece was surprisingly transparent from some angles - the spacing between slats could likely be even tighter. As for the highly stressed pieces which wanted to pull away from the notches - notching could be re-designed for a tighter hold, the overall twisting patterning could be closer scrutinized in the digital model (some of the twisting worked to the joints advantage) and/or the pieces could be steamed (requires an 8’ steam tube).
Step 5: STRAIN Lamp
The STRAIN Lamp is a prototype exploring the steam bending process, material capacities (for bending, twisting, and translation) and their relationship to digital design. The name refers to both the stranded units of fabrication and the stress the material undergoes during the forming process. Early tests with the process provided feedback which informed a parametric model used to generate the formwork (armature ribs) for the walnut strands. Because the digital model does not fully simulate the material behaviors, it is only an assumption and a starting point. Parametric design is much about controlling specific variables and so the project begs the question - how can the material potential be fully exploited with the minimum amount of ‘control’ or ‘how much control is too much’?
laser cut 1/2” plywood / steam bent walnut
TOOLS/SUPPLIES: Table Saw Metabeam Laser Cutter Steaming chamber / steamer ½” Plywood 5/4 Walnut boards Wood glue Clamps Zip ties
Step 6: Design Process
As with the LAMELLAE Screen, this exploration began with some testing of steam bending of sliced hardwoods, though unlike the more conservative screen, this piece attempted to push the limits of bending and twisting with the linear members. A study model was also produced from thinner basswood strands to get an overall understanding of the form/aesthetic. This was helpful in evaluating density and overall composition, symmetry, etc. Initial tests with the actual material gave an idea as to the amount of bending and twisting that could happen with sections of this size (⅜” x ½”). From there the parametric digital model allowed for adjustment of the # of slats (spacing), depth, width, bunching, degree of twisting, and amount of displacement or translation of the ‘slats’. Best judgement was used on limiting how much bending and twisting the pieces could take on, though it was difficult to fully coordinate this from rib-to-rib across the entire lamp. This is where digital design falls short and intuition/experience takes over. Once these decisions were made, the rib rings were extracted as .dxf files for laser cutting. Additionally a support armature or fixture was designed and cut to temporarily hold the rings in place while the strands were being attached.
Step 7: Fabrication
The rib rings and support fixture were quickly cut on the Metabeam at Pier 9 from ½” Walnut Ply and ¼” ply respectively. The strands were cut from the same boards as the much longer pieces for the LAMELLAE Screen. Once the fixture and rings were assembled and fixed in place, steaming of the strands could begin. Strands were presoaked in water overnight prior to steaming to add to their pliability. With a PVC steam tube approximately 6-8 strands were steamed at a time. Within a few minutes, the pieces would reach temperature and were pulled from the tube and fixed in place one at a time. This was done symmetrically from side to side to balance any stresses building up and keep the assembly aligned. Notches were pre-glued and the slats were slid/popped into place. Quick clamps were used to hold the strands at each of the seven connection points across the length as the working time with the steamed wood was only a minute or two. The clamps were replaced with temporary zip ties soon after and the process repeated on the other side, alternating until the lamp had to be flipped and stranding began on the other half.
Step 8: Conclusions / Speculations
When starting this piece it was unclear whether it would work at all - some of the bending and twisting of the pieces was quite radical and indeed some split during the process and had to be replaced. This was a good thing as it was intended to explore the material capabilities and limits. Unanticipated alignment issues came up with some of the pieces which had large variation in twisting from one rib to another - the piece would simply not sit nicely in its notch. This seemed to be OK in the overall composition, but a ‘looser’ method to control or fix the strands could be investigated. Lastly, the lengths of strands needed varied quite a bit and some of the areas where there was more bending/displacement required longer strands! 48” was the max length of the steam tube and so some of the strands came up short….in a future version, ideally longer strands would be used and cut down after being assembled!
Step 9: SHEAF Lamp
Unlike the first two projects in this series, the design of the SHEAF Lamp emerged not as a formal or material exploration for its own sake, but from a desire to incorporate ‘waste’ or offcut pieces into an object, leveraging similar techniques as the other projects. As the name implies, it references a vertical bundle of strands (the offcut ¼” x ¼” walnut pieces were bound together in the shop to store and became the conceptual starting point). A simple strategy to ‘pinch’ and ‘pull’ the density of the bundle drove the design.
laser cut 1/4” plywood / steam bent walnut
TOOLS/SUPPLIES: Table Saw Metabeam Laser Cutter Steaming chamber / steamer ½” Plywood 5/4 Walnut boards Wood glue Clamps Zip ties
Step 10: Design Process
With a 48” steamer and a 48” and 96” bundles of walnut strands, the decision was made to design around those limitations. The 96” bundle was cut in half - after that roughly 90 strands were available. A parametric model was then designed to leverage this ‘bundling’ as much as possible. Moments where the strands could be placed as densely as possible were balanced out by moments where they were pulled apart and even opened up to create an ‘inside’ view. A subtle twist in the model created modulation between the peaks and valleys of the ‘surface’. Using 81 strands (never assume all the pieces will be usable!) governed the overall general dimensions and various parameters were tweaked to accommodate the material parameters and desired effects.
Step 11: Fabrication
As with the other two projects, work began first by cutting the support ribs/rings on the Metabeam laser at Pier 9. A temporary support fixture was also cut from ¼” plywood. All the rings were numbered at each notch to keep the index of strands in order as each strand had a very particular path. Again, once the fixture and rings were assembled and fixed in place, steaming of the strands could begin. Strands were presoaked in water prior to steaming to add to their pliability. With a PVC steam tube approximately 12 strands were steamed at a time. Within a few minutes, the pieces would reach temperature and were pulled from the tube and fixed in place one at a time. This was done evenly around the lamp to balance any stresses building up and keep the assembly aligned.
*Actually, one of the bundles of strands was about ½” shorter than the rest. To make sure these pieces would fit and not come up short, a script was added to the parametric model which sorted through the index of strands to order them from shortest to longest. These short strands were then glued into the correspondingly shorter positions and used before filling in the remaining positions with longer strands. Notches were pre-glued and the strands were slid/popped into place. Quick clamps were used to hold the strands temporarily as the working time with the steamed wood was less than a minute.. The clamps were replaced with temporary zip ties soon after. This process was repeated until most of the strands were in place. After the fixturing was removed, the remaining strands could be fixed into place. After being allowed time to dry, the zip ties were cut away and a finish sanding with flexible foam sanding blocks cleaned up the rougher grain which had been raised by the steaming process
Step 12: Conclusions / Speculations
With each of these projects, the process becomes a bit more refined and technique improves too with continued iterative testing. The importance to think through the temporary support fixturing cannot be stated enough. Another aspect worth noting is (having done several stranded/louvered lamps) that in general the denser the slender components are, the finer more subtle the piece reads. This might seem obvious, but there is a big difference visually between the same scale piece using 60 vs 80 strands. The process of combining digital (formwork) and analog processes (steam bending) has been quite productive and I look forward to continuing these explorations.