INHERENT PARJAREE MANUCH FORM Chalmers School of Architecture Department of Architecture and Civil Engineering Master’s Thesis Spring 2024 MATTER & MEDIA JONAS LUNDBERG Architectural Experimentation Examiner/Supervisor INHERENT FORM PARJAREE MANUCH Master’s program in Architecture and Urban Design Chalmers School of Architecture Department of Architecture and Civil Engineering Master’s Thesis Spring 2024 Architectural Experimentation MATTER & MEDIA Examiner/Supervisor: JONAS LUNDBERG 5 ABSTRACT 6 INTRODUCTION 20 PROTOTYPE 54 TIMBER DETAIL DESIGN 68 71 8 BACKGROUND 14 DIGITAL FABRICATION METHODS OF RAW TIMBER 60 10 STRUCTURAL DESIGN METHODS USING RAW TIMBER 70 BIBLIOGRAPHY VISUAL REFERENCES ARCHITECTURAL APPLICATION CONCLUSION Rather than viewing digital tools as incongruous with craftsmanship, ‘Inherent Form’ suggests that they reinforce certain principles characteristic of craftsmanship that help elevate fabrication with benefits for architecture. The thesis aims to reveal the potential of architecture shaped by the interplay between natural materials, craftsmanship, and the capabilities of digital tools. It explores the design of medium span structures entirely composed of non-standard timber components derived from the inherent forms of local wood and demonstrates the potential of a raw timber structure for an architectural application for public use. The development of ‘Inherent Form’ involves a combination of design-driven research and material experiments. The thesis methodology offers a holistic approach to the design and construction process, rooted in material exploration, data acquisition, and structural solutions. It focuses on hands-on making as a means to explore and develop ideas, encompassing three key stages: Prototype, Timber Detail Design, and Architectural Application. These stages overlap during the development of the thesis, formalizing knowledge of design tools and techniques. Prototype involves utilizing digital tools for structure design, as well as point-cloud processing and photogrammetry to compile data from physical raw wood. Timber Detail Design is the process of crafting wood joinery for irregular tree bits, serving as a pathway to deeply understand and appreciate the material qualities of raw wood in structural way. Architectural Application presents a raw timber structure for a greenhouse situated within the Gothenburg Botanical Garden. ABSTRACT The thesis makes a contribution to the field of sustainable architecture by demonstrating an alternative to conventional timber construction that relies on the use of standardized materials produced through an industrialized process. Using non-standard found material to create architecture provides complexity and constraint to the design and fabrication process, but at the same time, it enables more sustainable material practices. By incorporating raw wood components into design, it is possible to not only diversify the design and construction methods according to locality but also create a new form of architectural expression. Keywords: Tree; Raw timber; Wood connection; Natural material Fig. 1. Inherent Form. Structure Prototype. 54 INTRODUCTION PURPOSE ‘Inherent Form’ refers to the external geometric form of trees. This thesis endeavors to propose a method rooted in natural material exploration for approaching the design and fabrication processes. It aims to explore building systems for a medium- span structure, utilizing non-standard timber components derived from the natural forms of local wood. The development of ‘Inherent Form’ involves a combination of design-driven research and material experiments. The thesis’ main emphasis is on exploring the application of digital workflows combined with non-digital fabrication processes to materialize architecture, showcasing the potential of a blend of natural materials, traditional craftsmanship, and new technology in shaping sustainable architectural design. THESIS QUESTIONS “How can digital design and fabrication tools be used to develop structures from non-standard raw wood components?” “What advantages do raw wood components offer in designing structures?” “How can raw wood be leveraged through innovative architectural design and construction methods to promote sustainability?” Timber Detail Design is the process of crafting wood connections for irregular tree bits, serving as a pathway to deeply understand and appreciate the material qualities of raw wood. The study of collected physical raw timber pieces was integral to the investigation of wood-to-wood connections, aimed at learning joinery techniques. Architectural Application demonstrates the potential of raw wood as a durable and functional material for architecture. The design proposal centers on a raw timber greenhouse structure within the Gothenburg Botanical Garden. METHOD The thesis methodology focuses on hands-on making as a means to explore and develop ideas, encompassing three key stages: Prototype, Timber Detail Design, and Architectural Application. These stages overlap in the development of the thesis, formalizing knowledge of design tools and techniques. The purpose of these three phases is not to depict the process in a direct chronological order, as the developments occurred concurrently. Prototype involves using digital tools for structure design and also the use of point-cloud processing and photogrammetry to compile data from collected physical raw wood. Throughout the stage, a series of four prototypes were developed, each varying in complexity from simple to intricate designs. Fig. 3. Prototype 4. Structural members. 76 Fig. 2. Collected raw wood, which could be divided into 4 types by topological differences in shape: bifurcated, curved, straight, and bent (crooked). Wood has stood as a fundamental structural material, valued for its versatility and strength throughout history. In ancient times, when human energy was the primary shaping force, builders devised intricate jointing systems that skillfully utilized the inherent diversity of wood. However, the industrial age brought about a transformation in the production of building materials. With the development of machinery, industrialization allowed for the application of increased physical energy to shape building materials. Variation, once valued, became an obstacle to overcome, with machinery enabling repetition and standardization over customization [8]. A reliance on standardized building components or blocks of material persists in many contemporary timber construction projects. The current industry tries to reduce irregularly grown wood into standardized straight timber. During such processes, internal fibres of the wood are repeatedly cut, which leads to the sacrifice of structural strength [1; 2]. Currently, only around 35% of harvested timber is used in construction worldwide [7], when considering only straight tree trunks. Small-radii, curved, or forked wood is discarded. A significant percentage of premium wood is still thrown away as pulp or firewood, despite the timber industry’s use of extremely advanced geometrical analysis technology to minimize waste. Architects became aware that high performance comes from the ability to adapt to local conditions [5]. Research on raw wood for architectural applications has been a well-studied field in the last decade. The research seeks to provide alternative approaches to the use of raw   wood in construction. Digital design and fabrication tools offer architects and designers unprecedented capabilities for developing structures from non-standard raw wood components. Another way to use computation in rationalizing design is to incorporate non- uniform natural materials into the design. Several technologies, such as laser scanning and increased computation power, have developed sufficiently to allow this to be feasible [9]. BACKGROUND Using non-standard timber components to create architecture provides complexity and constraint to the design and fabrication process, but at the same time, it enables more sustainable material practices. There are a variety of projects that focus on new adaptable design-to-fabrication workflows, which demonstrate that there are possibilities for making better use of the capacities found in natural wood by employing technologies to gain control over complexity. The processes include creating a digital material inventory, finding the best match between the initial design and the material inventory, creating a balance in structural performance, generating machine code for fabrication, and assembling the required raw wood elements to build the structure. Some architectural research institutes, such as the Architectural Association postgraduate program Design + Make, Cornell Robotic Construction Lab, and Taubman College of Architecture and Urban Planning, have investigated different prospects of wood natural forms as structural elements. They integrate emerging tools, such as 3D scanning, generative modeling, and robotic fabrication, enabling feedback between the designer and the material properties of raw timber. Through the rethinking of architectural design methods for using non-standard materials, this thesis tries to identify approaches that can contribute to the reduction of waste production in the current industry. The process, at the same time, discovers the unique architectural expression and aesthetic  and structural qualities of this natural material. With this investigative approach, material capacities and fabrication methods are explored towards new workflows and architectural applications, where material, craftsmanship, digital tools, and design are closely interlinked. Fig. 4. Limb. Diagramming CNC-milling potential on historical prints. 98 STRUCTURAL DESIGN METHODS USING RAW TIMBER Tree forks are naturally engineered structural connections that work as cantilevers in trees, which means that they have the potential to transfer force very efficiently thanks to their internal fiber structure. While not typical in conventional construction, tree-forks offer distinctive properties as connection elements. The research projects that follow make use of crooked trees and tree bifurcations to harness the structural advantages of a joint using a single piece of wood that was intentionally grown by natural forces. While a tree fork has a structural advantage over beams connected with external fasteners, it imposes a structural topology limited by an angle between two branches. Consequently, tree forks’ applications have a specific structural and architectural vocabulary based on the available stock of tree crotches [11]. Applications of raw wood can be categorized into three main types. First, straight wood is utilized in forms such as slabs, frames, trusses, grid-shells, and nexorades. Second, bifurcated wood, which includes tree forks, necessitates novel digital surveying, design, fabrication tools, and reinterpretation of traditional carpentry methods. Third, crooked curved wood is employed in experiments aiming to fit curved beams to larger surfaces, used in grid-shell systems. There are at least a dozen recent examples of research in the field of raw wood. One of the first prominent examples of truss structures was built in Hooke Park, UK. Other examples investigate wood joinery and assemblies of tree forks, branches, or just the crotch part, as well as reciprocal linkage of sawn timber and round straight logs [4]. Fig. 5. Tree Fork Truss. Fig. 7. Tree Fork Truss. Fig. 6. Tree Fork Truss. Final truss geometry exported from organization script. 1110 The tree fork truss, designed by the AA, London’s Design & Make program, exemplifies contemporary tree fork utilization. The design approach incorporates four core strategies: a precise referencing system for consistent component placement, photographic and photogrammetry techniques for creating a database of tree geometries, evolutionary optimization for component placement, and automated tool-path generation for connection fabrication. This innovative process minimally alters the natural tree form, focusing fabrication at the connection nodes, ensuring dimensional precision and structural integrity while embracing the inherent uniqueness of each tree component. The design method of Tree Fork Truss is demonstrated by fitting the inherent forms within a system which has been designed with them in mind - developing a set of variable control modeling tools which would allow the rationalization of complex forms into a rather simple geometric organization [8]. Photogrammetry was used to 3D scan 25 forks from the forest on campus based on the structure’s parameters. Together with Design & Make students and engineers, a Rhino/Grasshopper organization script generated a final fork component arrangement. This digital model was then translated into fabrication information with which the 6-axis robotic arm transformed each fork into a finished component. These huge components were pre-assembled in the workshop using a precision assembly jig before being moved to site for final installation. Fig. 9. Wasp aggregations of tree forks with same count of elements but varying number of proto-parts. Fig. 10. Structural exploration based on tetrahedral cell aggregation in a 3D gridFig. 8. Limb. Full-scale installation of one leg of a three-legged reticulated shell structure. Fig. 11. Log Knot. Robotically Fabricated Roundwood Timber Structure. 1312 The LIMB project was realized to explore the potential use of natural tree bifurcations as a new joinery method in a heavy timber construction. The placement of forks was based on the angular dimensions and dynamic inventory-constrained form- finding. The process selects the available crotch geometries intothe design geometry through optimization to minimize the geometric discrepancies of the intended design [11]. Design explorations of tree forks can be seen in cellular formations [Fig. 9 & 10]. Another approach, put out by the Conceptual Joining team, would study the design and performative potential of a particular set of unique elements and the relative emergent formations, rather than relying on a predetermined design, for which elements have to be manipulate. With the aim to creating unique, non-standard spatial structures at an architectural scale, the proposed workflow assumes a collection of physical branch parts and their distinct properties as the starting point and design driver. With the aim of activating the structural potential embedded in the material logic of bifurcating strands of fibers, the branching parts are joined following the extension of the wood grain orientation. The design and manufacturing process involves several key steps. First, digitization through 3D scanning of branches and automated feature extraction. Next, categorization clusters scan geometries and generate “proto-parts” as averages. Discrete element aggregation then creates design structures by recursively aggregating proto-branches. Population replaces “proto-branches” with real branch geometries based on design and performance parameters, such as length and cross-section. Relaxation compensates for gaps using force-based relaxation of the model. Finally, detailing for fabrication involves placing original meshes and post-processing for fabrication. Experiments with crooked wood often aim at a curved beam fitting to a larger curve or surface. Log knot [Fig. 11] is a robotically fabricated architectural installation that creates variable compound timber curvature utilizing both regular and irregular roundwood geometries [12]. The project also provides minimal-formwork assembly, bending, and moment force optimization of bespoke mortise and tenon joints. The project builds an infinite loop of roundwood that curves three- dimensionally using figure-8 knot logic. The new digital design and fabrication tools help develop design methods with irregular elements, whereas industrial applications only focus on the use of straight timber sections. These projects manifest the idea of exploring timber of minimal value with a particular architectural language coming from the appearance of elements. The overall workflow needed to be developed for the tree forks, such as scanning and robotic cuttings, demonstrates the diverse use of non-standard materials (not necessary timber) that do not need to be unified into equal shapes to have value in construction. The 3D scanning is necessary because of three points: a) each tree trunk is different (evenstraight), b) the design space has to consider these differences, and c) the timber fabrication requires to know the most accurate tree trunk position within the machining space. Therefore, it is necessary to collect the data about the real-world object (the tree) and possibly its appearance (the color). The collected data could reconstruct a 3D model or the low-level 3D representations, such as the central axis and radial parameters [11]. Scanning DIGITAL FABRICATION METHODS OF RAW TIMBER Fig. 12. Mother Maple. 3D scanning of a well known tree at Asitu’lisk. Fig. 14. A 3D model of a log with isocurves (Left). Physical log with milled isocurves (Right). 1514 3D scanning procedures have been used by the forestry industry to maximize tree growth and sawing procedures for straight wood. The use of cutting-edge technologies, including CT scanning, 3D analysis, LIDAR and RADAR scanning, and customized sawing, is highly developed in this industry. Nevertheless, these methods are limited to sawmill or forest growth statistics. Since 2010, an increasing number of studies in the field of raw wood research have used scanning techniques to shape structural forms using raw timber. The techniques rely on the topology of a tree log, including tiny and large radii, straight bending, and bifurcated trees, as well as economic reasoning, SDK availability, and scalability [11].  Utilizing an irreguular piece of wood necessitates a wide range of intricate details. Generating a digital representation of the wood that can be utilized to align it with a particular component design, place it during manufacturing, and create machining toolpaths is the first challenge. The machining setup is a prevalent concern in raw wood fabrication workflows due to the unique shapes of each raw wood piece. Moreover, unlike conventional rectangular oak beams, the round surfaces lack any reference points. Numerically-controlled machines (CNC and Robots) are capable of milling every conceivable angle from a wooden element to a certain extent. Therefore, the development of CAD-to-CAM applications has already made it possible to design and fabricate irregular forms of raw wood. Fig. 13. Point-cloud processing of a log laser scan. Scanned data are transformed into NURBS geometry, which functions as a lightweight representation of the digital stockpile. The translation is done using a custom script optimised for crooked log: The overall shape of the log is precisely described using simply geometry. Specialized software for point clouds can facilitate the creation of mesh geometry from the point cloud data. Nonetheless, the objective was to obtain a NURBS geometry representation, aiming for a lightweight data format and consistent construction of sawlog geometries. [See Fig. 13]. Newly developed digital tools are expanding the possibilities for working with this material, which previously required intricate manual skills and was often replaced by more straightforward production methods. These tools facilitate the drafting of geometrically complex wooden joints that no longer depend on elaborate manual techniques. Numerically- controlled machines (CNC and robotic systems) are capable of milling almost any angle from a wooden element, making the design and fabrication of irregular timber shapes more feasible through CAD-to-CAM applications. According to scanning methodologies, raw wood can be represented minimally in 3D using two parameters: a central axis and radial parameters along this axis. Tool-path Generation for Connections Connection geometries are then positioned based on these parameters and subtracted using polygonal primitives. This approach, referred to as a Minimal Model, allows for faster digital representation and tool-path generation for wood-wood connections. This method streamlines the process, ensuring that digital tools can efficiently handle the irregular shapes and unique characteristics of raw wood. Each cut in digital fabrication is governed by tool-head dimensions and machine reach-ability, as well as the size of components and joints, while a digital fabrication tool, e.g. CNC or robot, allows each saw to approach the work-piece in ways a human operator may not manage. Fig. 17. Tenon-mortise joint from raw wood. Fig. 16. Mesh Boolean operations on tree meshes, where a collection of outlines are constructed around the connection nodes (Top). Assembled prototype (Bottom). Fig. 15. Multiple tool-paths for a timber joint fabrication. The connections are defined as pair of polylines for cutting and lines for drilling. 1716 Fig. 18. Chainsaw robotic arm machining a surface on to a tree’s stem. Industrial Robot Arm Fig. 20. Log geometries and resulting wall surfaces. Fig. 19. A robotic arm with a custom band saw end effector is used for sawing irregular tree logs into naturally curved boards of varying thicknesses. 1918 Robotic fabrication usage for raw wood is distinguished by two primary motives: to develop complex timber-to-timber connections rooted in tradition and to explore performative manufacturing. [10]. Industrial robot arms have become indispensable tools for cutting raw wood, enabling customized workflows that integrate various cutting and vision tools. Raw wood cutting may require a set of customized tools atypical of traditional CNC machining, for example, bandsaws and chainsaws. Equipping a robotic arm with an analogy tool means implementing the potential offered by traditional techniques. It helps to materialize the complexity of digital space derived from the lack of homogeneity of the material and its tolerances [11].  Ashen Cabin [Fig. 19 & 20] is an experimental prototype for 3D printing and robotic architecture.   Using a KUKA KR200/2 with a custom 5hp band saw end effector, the designers can saw irregular tree logs into naturally curved boards of various and varying thicknesses (down to 2 mm thin). To integrate the non-standardized material, the sliced boards are arrayed into interlocking SIP facade panels. By adjusting the thickness of the bandsaw cut, the robotically carved timber boards can be assembled as complex single curvature surfaces or double-curvature surfaces. The SIPs are insulated using a two component closed-cell foam for which a fully biodegradable option is available [6]. The Prototype stage focuses on utilizing both raw materials and digital tools for structure design and development. This involved conceptualizing the overall geometry of the structure and also meticulously arranging and controlling each individual component. Digital exploration informed by raw material played a crucial role in refining these prototypes, allowing for the precise arrangement and control of each structural part. Throughout the stage, a series of 4 prototypes were developed, each varying in complexity from simple to intricate designs. The first prototype served as a foundational model, informing the design and construction of the second prototype. Insights and improvements gained from the second prototype were then applied to the creation of the third prototype, and the third prototype subsequently informed the development of the fourth. This iterative process ensured a progressive enhancement in the design and versatility of each subsequent prototype. These prototypes served as tangible representations of the exploration into structural possibilities. “Prototype” explores the solution for both a feasible structural arrangement and a structural design method. The process began with the design of structures, explored digitally to conceptualize the global geometry of a structure and spatial configuration that it could form. Following this initial phase of digital exploration, the project moved to the collection and preparation of raw materials. Tree bits, selected for their manageable dimensions suitable for woodworking hand tools, were carefully cut and collected from discarded piles and fallen trees around the campus. These locally sourced raw wood materials were intended for various types and stages of experimentation, ensuring they were manageable in both analog and digital contexts. The found materials are cataloged, and digital representations of each are created by photogrammetry operations. PROTOTYPE During the prototype stage, a collection of physical models was made using both analogue and digital fabrication techniques. The means of producing the physical models mainly involved the use of 3D printers, which enabled rapid testing of ideas. These physical 3D-printed models played a crucial role in transitioning to the timber detail design phase, where the raw material informed the specifics of the design. The creation of these models aimed to enhance comprehension of structural assembly sequences. The outcome of the Structure Prototyping was the creation of a final structure assembled from 140 distinct tree bits. Each component, characterized by its unique typological shape, was connected directly to one another, resulting in a cohesive whole that showcased the culmination of the prototyping process. PROTOTYPE 2120 Fig. 21. Structure Development 2322 Fig. 22. Prototype 1. Side view. Fig. 23. Prototype 2. Side view. Fig. 24. Prototype 3. Side view. Fig. 25. Prototype 4. Side view. 30 m 15 m PROTOTYPE 1 In the early stage of structure design, significant emphasis was placed on how the natural forms of trees could loosely inform the overall geometry. The objective was to create a design that not only reflected the organic shapes found in nature but also functioned effectively within an architectural context. Prototype 1 was primarily concerned with realizing this global geometry, focusing on scale and structural design optimized for an open plan layout. The concept revolved around utilizing natural tree shapes for structural components, ensuring that the design could be practically applied in a local context. To achieve this, iterations of the structure were explored during the development of Prototype 1. These iterations aimed to investigate different geometric configurations of the structure, all comprising relatively similar framing members. Each iteration targeted a uniform structure size of 15 x 30 x 8 meters. Through this iterative exploration, the arch frame structure emerged as the most promising design for further development into Prototype 2. This particular structure was selected due to its ability to emulate the natural parts of a tree, with members that closely resemble branches and trunks. The arch truss frame not only provided a robust and stable configuration but also maintained the aesthetic integrity of natural forms, which was a crucial aspect of the design ideology. The insights gained from Prototype 1 were instrumental in refining the design and ensuring that the structural and architectural goals were met. The selection of the arch frame structure aligned with the intention of local application, demonstrating how the design could be effectively utilized within the specific context of the Gothenburg Botanical Garden. This decision was based on the structure’s ability to integrate with the natural environment, providing an ideal solution for a botanical greenhouse that requires both functionality and aesthetic harmony with its surroundings. 2524 Fig. 26. Early structure design exploration. Fig. 27. Prototype 1. Structure Plan. Fig. 28. Prototype 1. 3D printed model in 1:100 scale.Fig. 28. Prototype 1. 3D printed model in 1:100 scale. Fig. 29. Prototype 1. 3D printed model in 1:100 scale. Prototype 1. Structure Model 2726 Fig. 30. Prototype 1. 3D printed model in 1:100 scale. Fig. 32. Prototype 1. 3D printed model in 1:100 scale. Fig. 31. Prototype 1. 3D printed model in 1:100 scale. PROTOTYPE 2 The units were composed of a variety of parts in an effort to optimize structural efficiency and enhance overall system performance. In Prototype 1, the framing members exhibited uniformity in size. The transition from Prototype 1 to Prototype 2 is characterized by a shift from uniformity to variability in both shape and size of the framing members. They’re diverse in shape and size based on position and the way they carry the loads. The development from Prototype 1 to Prototype 2 involved a systematic approach aimed at refining the design and improving its structural integrity. Initially, considerations were made for the augmentation of framing members and the detailed 3D modeling to closely replicate the shape and size of the tree’s trunk and branches. Prototype 2 retained the curvatures and overall geometry established in Prototype 1. However, a deeper level of systems thinking was applied to the structural components. This involved the division of components into cohesive units, each strategically designed with the sequences of assembly in mind. 2928 Fig. 33. Prototype 2. Digital model. Fig. 35. Prototype 2. Digital model. Fig. 34. Prototype 2. Digital model. 3130 Fig. 36. Prototype 2. Three groups of structural members. Fig. 37. Prototype 2. Structural members within one unit group. 3332 Fig. 38. Prototype 2. Structural members within one unit group. Fig. 39. Prototype 2. Structural members within one unit group. PROTOTYPE 3 In the beginning of the development of Prototype 3, a dataset of found forms was generated. A detailed dataset was made by 3D scanning 23 distinct raw wood pieces via photogrammetry operations. Raw wood was intentionally collected for digital and analog experiments. They could be divided into a few types by topological differences in shape: straight, bent, curved, and bifurcated bits. This operation leads to an opportunity to build a relationship between the physical raw materials and digital exploration that would be crucial for the rest of the design process. With the goal of applying inherent forms of tree parts to create a non-standard structure at an architectural scale, the workflow of the subsequent prototypes (Prototype 3 and 4) assumes a collection of physical local raw wood as the departure point and structural design agents. This aims at exploring the unique design potential of a specific set of elements, testing both their ability to adapt a predetermined structural geometry realized in the previous stages of prototyping and their potential to freely self-organize to directly inform the refinement of a structure. 3534 Fig. 40. Prototype 3. Digital model. Fig. 41. Prototype 3. 3D printed model in 1:50 scale. S02 S03 S04 S05 C01 C02 C03 C04 C05 C06 C07 S01 C08 C09 C10 F01 F02 F03 F04 F05 F06 F07 F08 Fig. 42. Found form inventory. 3736 In prototype 3, the process was time-invested in digitizing real branches to create a set of manageable mesh models as their digital twins. The scanned branches were categorized according to their geometric attributes, including length and diameter, arranged from largest to smallest. The scanned meshes were scaled up, and some of them were multiplied and manipulated to ensure adequate structural components to enable their arrangements within the desired target curvatures, ready to be adapted to the specific design and performative parameters. The placement of the straight and curved tree bits in these arrangements plays a significant structural role and acts as a form generator, while the forked bits are secondary to the entire structure. 3938 Fig. 43. Prototype 3. A digital model showing a polygon mesh of structural components. Fig. 44. Prototype 3. A digital model showing a polygon mesh of structural components. 4140 Fig. 45. Prototype 3. Bottom view. Fig. 47. Prototype 3. 3D printed model in 1:50 scale. Fig. 46. Prototype 3. Structural members are categorized by raw wood topology. PROTOTYPE 4 This variation in Prototype 4 is intentional, designed to emphasize the unique geometrical and morphological characteristics of each piece of raw timber. The structural units in Prototype 4 are carefully differentiated, leveraging the natural diversity of the wood to create a more dynamic and visually compelling structure. Prototype 4 demonstrates a higher level of readiness and consideration for practical architectural applications. One significant aspect of this readiness is the thoughtful increase in the number of components to accommodate an additional membrane, enabling the structure to function effectively as a greenhouse. This adaptation reflects a deeper integration of environmental considerations and functional requirements into the design process. Additionally, the development of Prototype 4 incorporates a more refined approach to structural integration with its base. The design includes detailed provisions for attaching the timber structure to a concrete foundation, recognizing the necessity of such support in real-world construction scenarios. This consideration ensures that the raw timber structure can be securely anchored, providing the necessary stability and durability for practical use. The development of Prototype 4 represents a significant evolution from Prototype 3, characterized by a notable shift in the approach to structural member orientations. In Prototype 3, the structural units maintained a high degree of uniformity, reflecting a preliminary exploration of form and function. However, Prototype 4 diverges from this uniformity, embracing a varied and distinct approach to highlight the wild and organic forms inherent in the raw wood components. The Prototype stage results in the final development of a structure built from 140 distinct raw wood components connecting directly to each other by way of milled connections. The final structure is made up of seven arched frame units that land at fourteen points. To exploit the structural capacity of the fork junction, Prototype 4 uses more forked bits to compose the structure supporting the main inclined arched frames that are composed of bent bits. The structure is an interplay of many curvatures, consisting of forked pieces weaving together. The strategic use of forked bits allows for innovative structural solutions, reducing the need for additional support elements. 4342 Fig. 48. Prototype 4. A digital model shows timber structure on concrete bases. Fig. 50. Prototype 4. 3D printed model with organic support unremoved in 1:200 scale. Fig. 49. Prototype 4. Structure plan. Fig. 52. Prototype 4. Structure model in 1:25 scale. 4544 Fig. 53. Prototype 4. Bottom view. Fig. 51. Prototype 4. Structure Model. Physical raw wood pieces were 3D, and the resulting models were 3D printed in 1:25 scale. Fig. 54. Prototype 4. Structure model in 1:25 scale. 4746 Fig. 55. Prototype 4. Structure plan showing original and simplified component meshes. Fig. 56. Prototype 4. Section. Fig. 59. Prototype 4. Pre-assembly of structure model. The evolution of the structure from Prototype 1 to Prototype 4 demonstrates a progression in design and application. Prototype 1 focuses on how the natural forms of trees could loosely inform the overall geometry and architectural scale, leading to the selection of an arched frame structure. In Prototype 2, this design is refined with a deeper level of systems thinking, introducing variability in the size and shape of the framing members to optimize structural efficiency. Prototype 3 introduces digitization, creating manageable mesh models of real branches and scaling them for desired target curvatures, significantly increasing structural complexity. Finally, Prototype 4 incorporates more structural components, enhancing both structural integrity and visual complexity, while also accommodating an additional membrane for practical application as a greenhouse. This prototype also included provisions for attaching the structure to a concrete foundation, ensuring stability and readiness for architectural application. 4948 Fig. 57. Prototype 4. Structure Unit. Fig. 58. Prototype 4. Structure model in 1:25 scale. 5150 Fig. 60. Prototype 4. Structure plan. Fig. 61. Prototype 4. Structure model in 1:25 scale. Fig. 62. Prototype 4. Structure model in 1:25 scale. 5352 Fig. 63. Prototype 4. Drawing of 7 structure units. Fig. 64. Prototype 4. Structure physical model in 1:25 scale. TIMBER DETAIL DESIGN A common technique for defining joinery for irregular timbers is to create solid connection geometries — a pair of corresponding’subtraction volumes’ defined by each of the pair elements meeting at a given connection zone. In short, a joint is composed of a female and a male element that must be crafted from raw wood. These subtraction volumes consist of geometric rimitives, such as cuboids, cylinders, and truncated cones, and represent the wood material’s volume needed to be removed to obtain the connection surface [8]. The outcomes of these experiments culminated in the creation of milled connections done by woodworking hand tools, where each timber element was precisely shaped to fit together seamlessly.  The connection topology of raw wood connections can be categorized as follows: a) side-to-side, b) top-to-side, c) top- to-top, and d) cross-halving. Additionally, they frequently implement additional fasteners. While mechanical fasteners have become a standard practice for jointing timber assemblies due to their predictable performance and simplicity of use, research in robotic fabrication could investigate the development of intricate timber-to-timber connections that are inspired by traditional wood joinery. Timber Detail Design is an experimental stage that involves a detailed examination of how raw wood components could be effectively joined, considering the natural irregularities inherent in the material. Through a series of experiments, a range of joinery methods were developed, each requiring both planar and non-planar cuts to accommodate the unique geometries of the raw wood. In the full-scale realization of the structure, these joinery studies will inform the robotic fabrication process, utilizing cutting and milling techniques with the precision afforded by industrial robotic arms. This precision ensures that the joining of parts can be achieved without the need for additional extension elements. The joint studies conducted can be digitized using scanned branches, facilitating the production of robotic simulations and full-scale prototyping. For the final development of the structural prototype, multiple types of joints will be required. The spatial complexity introduced by the geometry of the parts, along with their specific locations, necessitates the use of an organization script and joinery solver for the assembly process. 5554 Fig. 65. Top-to-Side connection of a forked piece and a bent piece. Fig. 66. Close-up of milled connection. Fig. 67. Milled piece of tree fork attached to base. 5756 Fig. 68. Series of milled connections. Fig. 69. Final joinery. 5958 Fig. 70. Top-side joint. Fig. 71. Top-side joint. 1 The Raw-Timber Greenhouse 2 Herb Garden 3 Alpine House 4 Horticultural Gardens 5 Växthusen 5 3 4 1 2 Fanny G réns Väg 10m0 ARCHITECTURAL APPLICATION Fig. 72. Exterior view of the raw-timber greenhouse. This final design proposal incorporates the use of two layer inflated ETFE membrane cushion, highlighting the raw timber structure’s capability to support such membranes. The integration of these panels presents a significant challenge, as it involves designing the membrane to complement the structure aesthetically and functionally. The ETFE outer layer is a significant benefit of the membrane design because it protects the raw timber structure from the weather. The approach to membrane design begins with the strategic placement of anchor points for individual cushions to be fixed, which are distributed across the structure in accordance with the arrangement of the raw wood pieces. Following this, the outlines for the membrane panels are designed to ensure a cohesive and visually appealing integration with the underlying timber structure. The knowledge gained from the prototype and timber detail design stages is integrated into the development of an architectural application. The objective of this stage is to demonstrate the potential of raw wood as a durable and functional material for architecture. The design proposal centers on a novel greenhouse structure within the Gothenburg Botanical Garden. The proposed raw-timber greenhouse will be situated adjacent to the newly built greenhouse facilities at the Gothenburg Botanical Garden. This design aims to showcase the feasibility of constructing sustainable structures using locally sourced materials. The project illustrates how local resources can be effectively utilized for sustainable architectural applications, aligning with contemporary environmental and sustainability goals. 6160 Fig. 73. Gothenburg Botanical Garden site plan. 6362 Fig. 74. Greenhouse’s components. Main structure with anchor points and ETFE panels as outer membrane. Fig. 75. Exterior view of the raw-timber greenhouse. 6564 Fig. 77. Section. Fig. 76. Aerial view of the raw-timber greenhouse. Fig. 78. Greenhouse floor plan. 6766 Fig. 79. Section. Fig. 80. Interior view of the raw-timber greenhouse. Fig. 81. Top view of greenhouse’s outer skin. CONCLUSION To advance this design research, the focus will primarily be on refining the methodologies to ultimately achieve optimal arrangment when the original material exhibited complex irregularities. This can be achieved through the further incorporation of the following key points: - Development in the procedure of the inherent form organization: It is beneficial to utilize a set of variable control modeling tools that can simplify complex forms into more straightforward geometric organizations. This rationalization will enhance the efficiency and accuracy of the structure, which could inform the joinery generation. - Comprehensive Branch and Structure Analysis: Capturing a branch’s form involves more than merely recording its shape with tools such as photogrammetric software. It also necessitates non-destructive techniques to formalize the branch’s specific physical, mechanical, and inherent behaviors. This holistic approach ensures a thorough understanding of the material properties, which is crucial for structure simulation. - Joinery Solver: To integrate the use of an algorithm aiming at geometry generation for pair-wise wood-wood connections, it will help facilitate the connection method, which in the context of this study depends on the complex element orientation.  By following these stages, the development of research aims to enhance raw wood structure design methods, leveraging advanced digital tools and a deep understanding of the geometric complexity of raw wood.  6968 Fig. 82. Details of a final structure model. Bianconi, F., Filippucci, M., (2019). Digital wood design: innovative techniques of representation in architectural design. Hoadley, R.B., (2000). Understanding Wood: A Craftsman’s Guide to Wood Technology. 1st edition. ed. The Taunton Press. L. Allner, D. Kroehnert, & A. Rossi, (2020). Mediating Irregularity: Towards a Design Method for Spatial Structures Utilizing Naturally Grown Forked Branches. in Impact: Design With All Senses, C. Gengnagel, O. Baverel, J. Burry, M. Ramsgaard Thomsen, and S. Weinzierl, Eds., Cham: Springer International Publishing. Kerezov, A.D., Koshihara, M., Tachi, T. (2023). From Natural Tree Forks to Grid Shells: Towards a Self- forming Geometry. In: Cheng, LY. (eds) ICGG 2022 - Proceedings of the 20th International Conference on Geometry and Graphics. ICGG 2022. Lecture Notes on Data Engineering and Communications Technologies, vol 146. Springer, Cham. Kolarevic, B. (2005). Performative Architecture - Beyond Instrumentality. Spoon Press. Lok, L., & Zivkovic, S., (2019). Ashen Cabin. Open. Michael H. Ramage, Henry Burridge, Marta Busse-Wicher, George Fereday, Thomas Reynolds, Darshil U. Shah, Guanglu Wu, Li Yu, Patrick Fleming, Danielle Densley-Tingley, Julian Allwood, Paul Dupree, P.F. Linden, Oren Scherman. (2017).The wood from the trees: The use of timber in construction. Renewable and Sustainable Energy Reviews. Mollica, Z. (2016). Tree Fork Truss: An Architecture of Inherent Forms. Design and Make. URL: http://zacharymolli.ca/assets/download/ Mollica-DMThesis-Tree-Fork-Truss-an-Architecture- of-Inherent-Forms.pdf. BIBLIOGRAPHYVISUAL REFERENCES Parjaree Manuch holds a Bachelor of Science degree in Architectural Design from the International Program in Design and Architecture at Chulalongkorn University. With a background in practicing architecture in Thailand, she has gained hands-on experience in hospitality and residential projects, where her focus extends beyond design to encompass construction aspects as well. Parjaree also has worked in interior design, further enhancing her understanding of the intimate scale of architectural work. She is interested in craftsmanship and the utilization of digital tools in architectural design, which she has investigated in her Master Thesis at Chalmers University of Technology in Sweden. 7170 All other drawings and images by the author. Fig. 4. Fig. 5. Fig. 6. Fig. 7. Fig. 8. Fig. 11. Kasey Vliet, Peter von Bülow, & Steven Mankouche. “Limb”. 2018. Zachary Mollica. “Tree Fork Truss”. 2015. Zachary Mollica. “Tree Fork Truss”. 2015. Zachary Mollica. “Tree Fork Truss”. 2015. Steven Mankouche. “Limb”. 2018. Lukas Allner, Daniela Kroehnert, & Andrea Rossi. 2020. Lukas Allner, Daniela Kroehnert, & Andrea Rossi. 2020. Jeremy Bilotti. “Log Knot”. 2018. Zachary Mollica. “Mother Maple”. 2021. Anders Aagaard, & Niels Martin Larsen. 2020. Anders Aagaard, & Niels Martin Larsen. 2020. Petras Vestartas. 2021. Petras Vestartas. 2021. Petras Vestartas. 2021. Emmanuel Vercruysse, Zachary Mollica, & Pradeep Devadass. 2019. John Munson. 2019. Leslie Lok, & Sasa Zivkovic. 2019. Fig. 12. Fig. 13. Fig. 15. Fig. 16. Fig. 17. Fig. 18. Fig. 19. Fig. 9. Fig. 10. Fig. 14. Fig. 20. Stanton, C. (2010). Digitally Mediated Use of Localized Material in Architecture. Vercruysse, E., Mollica Z., & Pradeep, D. (2018). Al- tered Behaviour: The Performative Nature of Manufac- ture Chainsaw Choreographies + BandsawManoeuvres. In Robotic Fabrication in Architecture, Art and Design. Vestartas, P. (2021). Design-to-Fabrication Workflow for Raw-Sawn-Timber using Joinery Solver [Doctoral dissertation, Swiss Federal Institute of Technology Lau- sanne]. EPFL scientific publications Živković, S., Havener, B., & Battaglia, C.A. (2020). LOG KNOT: Robotically Fabricated Roundwood Timber Structure. Open. [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] My family in Thailand for always rooting for me. Jonas Lundberg for instructive guidance. Bongkodpast Tantipisanu for making impossible tasks possible. Erica Hörteborn for sharing valuable ideas. Tabita Nilsson for showing me the right techniques. Chalmers Fuse for model-making tools and materials. Eylül, Jiaming, Xiaofei, and August for help. My friends in Thailand for constant encouragement. I dedicate this thesis to my father, who was my tree. I want to thank the following people who have been important to this work: Parjaree Manuch parjaree.manuch@gmail.com