Research of Theories
How are we to understand the precision outside of its duty to exclude error, and potentially uncoupled from its contract with truthfulness? And, given this, what is the newly reconfigured space of error?
Bibliography
_Francesca
House Rachel Whiteread 1993
System Crash Courtesy of Gergely Kovács Digital Error
Computer Sketchpad The MIT Museum 1963
At high magnification, multiple lines that were originally precisely intersecting at a point do not actually intersect perfectly. This kind of flaw appears particularly prominent under the magnifying glass function, but cannot be observed in a normal view.
Gallaudet DB-1B Detail of Construction U.S. Air Force
In the 1920s, there was intense competition between wood and metal in the field of aviation materials. The transformation from unreliable wood to future material metal reveals the ideological struggle in material selection, rejection of organic materials, transition to metallic materials, and the conflicting demands of modern culture for persistence and stability.
"Blebs" Idea Greg Lynn 1998 Digital Error
"Blebs are pockets of space formed when a surface intersects itself making a captured space." Greg Lynn mentioned in "Greg Lynn Form" the generation of blebs. As an architect and designer, Greg Lynn was a pioneer in using computers systematically to create "methodical" deviations from traditional patterns and stereotypes.
Splitting Gordon Matta-Clark 1974
Matta-Clark selected a suburban house and used a chainsaw to cut along its central axis, splitting it in half. His work demonstrated how direct intervention and improvisational adjustments could address unpredictability and errors within materials and structures.
Regulatory Network La Polygraphie Du Cavalier Jonathan Meyer
Sculpture
The Carver Barbara Hepworth 1950s
Precise surface effects created by geometric optics simulationin in digital rendering are a form of "false precision". Fabricating errors refer to deviations caused by craft, technique, misoperation, or material properties in the production and prefabrication process of building components.
Inflexible 2100HP Roll Royce Condor 1917-1921
After the World War I, metal gradually replaced wood as the main material for the aircraft construction. However, there is a generation of airplanes that have never flown on abandoned airports in the United States during the interwar period. These planes were bound to the ground due to the weight of the new metal wings, becoming a batch of relics of material errors.
Figures I to XVII "Metamorphosis Successivam" Wellcome Library
In the medical field, there is also the phenomenon of eliminating individual differences and errors through standardized image, which ignores the inevitable mutations and variations in the actual process. Sommering presented illustrations of fetuses that are not drawn entirely based on actual observations, but are "standardized" to showcase the ideal form of fetal development.
Mario Carpo further developed the viewpoint of Alexandre Koyré in a famous essay first published in 1948, shifting from "precision is the hallmark of modernity in all aspects of life and science", to "the effectiveness of the result in the digital age is achieved not by dint of authorial precision, but through approximation, redundancy, and endless participatory revisions.
Micrographia Robert Hooke 1665
Gene Network Courtesy of Gergely Kovács (Typical)
Programmatic modeling of the entire process allows designers to create complex geometric shapes through node networks and encourages the introduction of the randomness and variation in the design process. The core of Houdini's programmatic modeling is generated through parameterized control, with each design step adjustable and repeatable. This flexibility provides ample space for errors to occur.
Unlike printed drawings, digital notations seem to be in a state of permanent drift. However, any changes may randomly introduce some errors, faults or glitches.
Splitting Frame Enlargements 1974
The observations of "Splitting"
indicate that any tool and design, when magnified, will expose its imperfect side, and absolute accuracy does not exist in practical applications. Redundant accuracy in digital design may also lead to design complexity.
Randomness Error & Layering
Randomness error refers to: when it is necessary to divide a larger scale object into many smallscale components that can be fabricated in the process of using Houdini for programmatic modeling, a "cluster" node can be introduced to randomly group the internal lattice of the object. By adjusting parameters such as the number of clusters and the seeds, the size of the buildable components, as well as the shape of gaps and interfaces between components, can be changed.
In Summary, due to the introduction of these programmable generation nodes with random attributes in algorithms, the design results cannot be precisely controlled to some extent, and this phenomenon is defined as the randomness errors. Therefore, it can be hypothesized that the randomness errors can be transformed into the strategy of presetting ageing paths, as well as replaceable and maintainable design.
Bio-Corbelling
Using computational design, rethinking how architecture layers work. Instead of enclosing them in separate boxes, we suggest they intersect, in a bio-corbelling way. We designed a houdini script for spatial generation and hierarchical layer division, and tested it with different massing inputs. By defining the materials, we aim to ensure that, their durability and replaceability align with the hierarchical shearing layers.
Exploring the possibility of transforming the generated structure into a habitable space. The prototype 01 includes mycelium as a non-load-bearing partition wall, supported by a sturdy structure of concrete and stone. Through the particle simulation, attempting to replicate on digital platforms the effects of mycelium degradation and the bio-receptivity of materials that have observed in the real world.
Hybrid Material Facade
Firstly, an algorithm is designed in Houdini for spatial generation and division of structures and material layers. Based on the durability and replaceability of different materials, they are applied to structural layers with different life span. For example, lightweight concrete and mycelium materials are used for the outer layer that protects the main structure, reducing the overall load of the building while also lowering the difficulty and cost of replacing and maintaining components.
Randomness errors are introduced following the "cluster" node, and by adjusting the parameters iterations and seed, the components exhibit a stepped interlocking connection. This connection method enhances the overall strength and stability of the structure and eliminates the need for additional connecting elements.
The Voxelization Error & The Resolution
In summary, due to repeated voxel processing, objects exhibit unexpected morphological variations or surface resolution deviations compared to the original object, which is defined as the voxelization errors. If the new geometry is further voxelated, the higher the density of the internal lattice of the geometry, the richer the details of its surface texture after voxelization.
For example: In Houdini, voxelizing a double-layer geometry with a voxel resolution of 0.5 and voxel size of 0.2 to generate a three-layer structure. Further voxelizing the three-layer geometry with a voxel resolution of 1 and voxel size of 0.1 to create a fivelayer structure. At this point, the new geometric surface has changed compared to the original state.
Mycelium Window Frame Design
In the process of modeling window frames using Houdini, the multi-layered, stepped, and concave geometric forms are generated by introducing multiple voxelization errors.
The outer layer of mycelium should have excellent weather resistance and water repellency, the middle layer, serving as structural support, must possess sufficient material strength and stability, while the inner layer needs to meet certain thermal and acoustic insulation requirements, as well as environmentally friendly, mold-resistant, and antibacterial.
Error Effect on Ageing
Simulating design errors and their impact on building aging through 4 sets of comparative experiments, and demonstrate the feasibility of transforming errors in design process into strategies that are beneficial for building aging.
Experimental results demonstrate that introducing voxelization error in the design process can be transformed into the strategy of pre-setting ageing texture and multi-layer structure. While introducing randomness error in the design process can be transformed into the strategy of presetting ageing paths, as well as replaceable and maintainable design.
Experiment 1
Before the ageing simulation, the surface of the "error 4" geometry presents the highest resolution and most layered stepped shape.
After the ageing simulation, the three-layer skin: resist weathering erosion, prolong service life. The five-layer skin structure: biological acceptance and biodiversity.
Experiment 2
Before the ageing simulation, the aging path of "error 1": more tortuous step shape. "error 2": tecture density increase, more complex and variable.
After ageing simulation, the surface of "error 2" is significantly more conducive to plant growth and ecological balance.
Before ageing simulation, on the basis of "error 3", there are more interlocking connections between the components of "error 4".
After the ageing simulation, the stepped interlocking connection method between components is beneficial for improving structural strength and stability, without the need of additional connectors.
Experiment 4
Before ageing simulation, the texture of "error 2" is relatively clear, and uneven details are more abundant, presenting a much more natural and organic state.
After ageing simulation, "error 2" shows visibly stronger durability and adaptability during weathering and wear,and a better visial effect.
Ageing Performance and Patterns
Response Spectrum Analysis
Stress concentrations are presented at points of the church where the structure has been cracked. Most stresses on the dome, as shown in a set of following figure, are tensile stresses.
Seismic Excitation
For the evaluation of the structural behaviour under seismic excitations, the church were analysed to study the influence of masonry buttresses on dynamic behaviour of the structure.
The various materials on the building facade exhibi different aging conditions after undergoing varying degrees of environmental degradation. Record and analyze the aging conditions on the east and west facades of the Venetian church, and categorize them into the following three types: Biological Colonization on Finishing, Discoloration of Limestone, and Cracking. Understanding the performance of different materials in real environmental conditions, as well as their aging patterns and causes, helps to identify potential issues early and develop targeted design and maintenance strategies.
Biological Colonization on Finishing Discoloration of Limestone Cracking
03-1
Prototype 03-2
03-3
Prototype 03-3
Ageing Distribution
Simulating aging distribution can provide strategic recommendations for selecting appropriate materials and optimizing material distribution during the early stage of the facade design.
For example, to optimize material distribution, different materials can be strategically placed on the facade based on ageing patterns. In areas prone to water accumulation or with poor drainage, materials resistant to water erosion should be used; in regions with significant temperature fluctuations, materials with high crack resistance should be selected, which is valuable for improving the scientific and sustainable nature of architectural design.
Prototype 03-2
Prototype 03-3
Machine Learning
Using voxel diffusion skills to complete a successful training of a custom 3d model in python in the google colab environment and testing to bring generated index level data back into Houdini as a new geometry with identical features and entirely generated by AI.
Training Procedure
The dataset composed of 2D images is more conducive to AI recognition and analysis compared to 3D, to improve its accuracy, (a. Creating a custom dataset in houdini with at least 500 sliced 3d models. (b. Preprocessing the data with python that it is suitable for training the given neural network. (c. Training the diffusion model with the custom dataset and testing it by generating synthetic data.
Training Results
The learning model underwent 21 iterations of training, with the output from each iteration serving as the training data for the next round. The sliced images generated became progressively clearer, ultimately leading to the precise output of geometrically accurate sliced images.
Composite Material
Outer layer: short-lifecycle, replaceable biomaterial mycelium. Inner layer: high-strength and stable concrete material.
Multi-layer Material
Multi-layered materials will enable the facade to exhibit a more richly layered dynamic variation effect during the ageing process.
The introduction of secondary voxelization errors generates a double-layered composite texture on the wall surface. By completing the training of the learning model, hundreds or thousands of surface textures with similar features and styles can be quickly generated.
The composite texture on external wall has a positive impact on aging. In addition to reducing maintenance cost and improving ecological benefits, it can also much more effectively disperse and absorb external stress, thereby reducing the crack propagation caused by local stress concentration.
Discussion & Conclusion
It was not until postmodern architects questioned the obsession with precision, emphasizing that architecture should reflect diversity and complexity rather than singular rationality and order, that the necessity of errors began to be acknowledged. "To ask: what is the newly reconfigured space of error?" The core issue of this study is how to transform design errors into targeted strategies that beneficial to the aging process of buildings. Unlike modernist architecture's fascination with sterile environments and anxiety about natural decay, this study intends to use design errors to strategically plan the aging process, showcasing a unique beauty that emerges over time. This beauty originates from the traces and changes left by time on materials, endowing buildings with a profound sense of history and vitality.
This project proposes an experimental method to simulate aging by conducting dynamic simulations primarily focused on wind particle erosion in Houdini. This approach aims to verify the positive impacts o two types of design errorsvoxelization errors and randomness errors - on the aging process of buildings. Furthermore, this study develops a systematic methodology that integrates the introduction of errors, transformation strategies, and aging simulations, demonstrating how this methodology can be applied in actual design projects.
The project also provides a detailed description of the error transformation strategies, including preset aging textures and paths, multi-layered facade structures, and replaceable and maintainable design These strategies play a significant role in extending building lifespan, enhancing durability, reducing maintenance costs, and improving ecological benefits. However, there are still many areas worth further exploration. Firstly, expand the sources of design errors by continuing to explore the potential for innovative design brought about by computational errors in digital software. Secondly, introduce diverse dynamic simulation algorithms and develop a tool capable of simulating the combined effects of multiple environmental factors on ageing. Finally, incorporate machine learning tools such as Neural Networks and GANs to establish a database of different materials at various stages of ageing.
Limitations
The phenomenon of ageing is the result of the combined effects of natural environmental factors. This study is currently limited to using wind simulations in Houdini to obtain experimental data and conduct analysis. It is not possible to simultaneously observe the impact of other external environmental factors such as sunlight, rain, and temperature changes on structural ageing.
Furthermore, in addition to the erosion from the external environment, the internal material characteristics of the building are also closely related to the aging phenomenon. At present, this study cannot assign the corresponding material properties to the research objects during the computational simulation process. Therefore, the experimental results need to be further validated.
Pattern Studies
The project started with series of experiments on adding and subtracting mass from basic, hollow cube by manipulating deposition speed of a 3D printer. The faster printer moves the less matter it deposits. To design the relationship between the speed of deposition and a material is to design time contingent, controlled variation of semiornamental mass distribution at the surface scale and beyond.
Fabricating Error
Additive manufacturing technology provides the ability to accurately control materials and shapes during the fabricating process, but it also comes with potential machine parameter settings errors.
Parameter Setting
The
Physical Model
Using variable speed settings in 3D printing to produce different textural effects on Bacterial Cellulose, which are not just for aesthetic, but improve structural integrity and performance.
Prototype
Gap Between Digital & Physical
This project emphasizes the interaction between digital design parameters (such as speed and movement patterns of the printer) and the physical properties of the materials being used. Different materials respond n unique ways to changes in printing speed, affecting how textures form and behave structurally. Based on this, the research develops methods that allow to anticipate and utilize these behaviors, integrating them into design process to enhance both function and form.
Digital Fabrication
"False Precision", which obscures potential material and technical uncertainties in the actual fabricating process, namely fabricating errors. This project challenges the conventional linear thinking in digital fabrication that physical fabricating is only an engineering simulation that follows digital design, proposing instead a model that values unpredictability and the intrinsic properties of materials as essential elements of the design process.
Discussion & Conclusion
In the architectural design cycle, fabricating errors refer to deviations caused by craft, technique, misoperation, or material properties in the production and prefabrication process of building components. These errors, although initially negative, can actually provide designers with unique opportunities. Different from "Accident" implies that the deviation is a one-time, non-replicable event, the "errors" discussed in this project represent a form of predictability that can be intentionally introduced, offering the potential to optimize design and achieve innovation.
Erwin Schrödinger used the concept of "entropy" to explore life and gene replication, which is metaphorically applied to architectural errors, suggesting that errors much like genes, are not only responsible for information transmission but also play a role in the development of form. This metaphor highlights the viewpoint that by guiding errors, one can achieve optimization of both form and function.
