Martin Tamke

Assessing implicit knowledge in BIM models with machine learning

The promise, which comes along with Building Information Models, is that they are information rich, machine readable and represent the insights of multiple building disciplines within single or linked models. However, this knowledge has to be stated explicitly in order to be understood. Trained architects and engineers are able to deduce non-explicitly explicitly stated information, which is often the core of the transported architectural information. This paper investigates how machine learning approaches allow a computational system to deduce implicit knowledge from a set of BIM models.

Design Environments for Material Performance

The research project that induced the Dermoid (Fig.1) installation investigates the making of digital tools by which architects and engineers can work intelligently with material performance. Working with wood as a material, we were especially interested how the bend and flex of wood, can become an active parameter in the digital design process. Traditional building structures facilitate load bearing through a correlation of compressive and tensile forces passing loads linearly through the building envelope. However, materials hold internal forces that can be incorporated into structural systems thereby reducing material use and leading to a more intelligent and potentially sustainable building practice. [2] [6]

Hybrid Tower, Designing Soft Structures

This paper presents the research project Hybrid Tower, an interdisciplinary collaboration between CITA—Centre for IT and Architecture, KET—Department for Structural Design and Technology, Fibrenamics, Universidade do Minho Guimaraes, AFF a. ferreira & filhos, sa, Caldas de Vizela, Portugal and Essener Labor fur Leichte Flachentragwerke, Universitat Duisburg-Essen. Hybrid Tower is a hybrid structural system combining bending active compression members and tensile members for architectural design. The paper presents two central investigations: (1) the creation of new design methods that embed predictions about the inherent interdependency and material dependent performance of the hybrid structure and (2) the inter-scalar design strategies for specification and fabrication. The first investigation focuses on the design pipelines developed between the implementation of realtime physics and constraint solvers and more rigorous Finite Element methods supporting respectively design analysis and form finding and performance evaluation and verification. The second investigation describes the inter-scalar feedback loops between design at the macro scale (overall structural behaviour), meso scale (membrane reinforcement strategy) and micro scale (design of bespoke textile membrane). The paper concludes with a post construction analysis. Comparing structural and environmental data, the predicted and the actual performance of tower are evaluated and discussed.

Processing Branches: Reactivating the performativity of natural wooden form with contemporary information technology

Angled and forked wood – a desired material until 19th century, was swept away by industrialization and its standardization of processes and materials. Contemporary information technology has the potential for the capturing and recognition of individual geometries through laser scanning and computation and subsequently design and bespoke CNC fabrication. The question whether this allows for a new approach to the uniqueness that is offered to us by nature is discussed in a series of workshops and projects, which explore the performative potential of naturally grown materials.

Enrichment and Preservation of Architectural Knowledge

In the context of the EU FP7 DURAARK project (2013–2016), inter-disciplinary methods, technologies and tools have been researched and developed, that support the Long Term Preservation of semantically enriched digital representations of built structures. The results of the research efforts include approaches of semi-automatically deriving building models from point cloud data sets acquired from laser scans and the integration and overlay of such representations with explicit Building Information Models (BIM). We introduce novel ways for the further semantic enrichment of such hybrid building models with contextual data and vocabularies from external resources using Linked Data (LD) and the recognition relevant features and building components. A special focus of the research reported here lies on strategies and policies for their long term archival, information retrieval based on rich semantic metadata and the use of such archival systems in research and commercial scenarios. We introduce a set of prototypical, open-source tools implementing these features that have been integrated into a modular preservation framework called the “DURAARK Workbench”.

Generating a scalar logic: producing the "it's a SMALL world" exhibition

This paper presents the design project “its a SMALL world”, an exhibition design developed for the Danish Design Centre in 2009. The project investigates the making of a generative design environment by which multiple design parameters as from program, site or the subsequent digital fabrication and assembly process can be negotiated. In this paper we discuss methods for understanding the emergent interrelationships between encoded parameters, how to manage these and their impact on design. The implementation of the design necessitated a novel design method that allowed to blend the qualities of a generative design approach, that can adapt through recursion gradually to local requirements, with explicit definitions. The project showcases with its new developed manufacturing system for nonstandard element how customized digital design and production tools allow for a novel nearness to material and new ways of production and collaboration of architects, engineers and the crafts.

Computational Strategies for the Architectural Design of Bending Active Structures

Active bending introduces a new level of integration into the design of architectural structures, and opens up new complexities for the architectural design process. In particular, the introduction of material variation reconfigures the design space. Through the precise specification of their stiffness, it is possible to control and pre-calibrate the bending behaviour of a composite element. This material capacity challenges architectures existing methods for design, specification and prediction. In this paper, we demonstrate how architects might connect the designed nature of composites with the design of bending-active structures, through computational strategies. We report three built structures that develop architecturally oriented design methods for bending-active systems using composite materials. These projects demonstrate the application and limits of the introduction of advanced engineering simulation into an architectural workflow, and the extension of architectures existing physics-based approaches.

Multiscale modeling frameworks for architecture: Designing the unseen and invisible with phase change materials

Multiscale design and analysis models promise a robust, multimethod, multidisciplinary approach, but at present have limited application during the architectural design process. To explore the use of multiscale models in architecture, we develop a calibrated modeling and simulation platform for the design and analysis of a prototypical envelope made of phase change materials. The model is mechanistic in nature, incorporates material-scale and precinct scale-attributes, and supports the design of two- and three-dimensional phase change material geometries informed by heat transfer phenomena. Phase change material behavior, in solid and liquid states, dominates the visual and numerical evaluation of the multiscale model. Model calibration is demonstrated using real-time data gathered from the prototype. Model extensibility is demonstrated when it is used by designers to predict the behavior of alternate envelope options. Given the challenges of modeling phase change material behavior in this multiscale model, an additional multiple linear regression model is applied to data collected from the physical prototype in order to demonstrate an alternate method for predicting the melting and solidification of phase change materials.

Machine learning for architectural design: Practices and infrastructure:

In this article, we propose that new architectural design practices might be based on machine learning approaches to better leverage data-rich environments and workflows. Through reference to recent architectural research, we describe how the application of machine learning can occur throughout the design and fabrication process, to develop varied relations between design, performance and learning. The impact of machine learning on architectural practices with performance-based design and fabrication is assessed in two cases by the authors. We then summarise what we perceive as current limits to a more widespread application and conclude by providing an outlook and direction for future research for machine learning in architectural design practice.

Shaping hybrids – Form finding of new material systems

Form-finding processes are an integral part of structural design. Because of their limitations, the classic approaches to finding a form – such as hanging models and the soap-film analogy – play only a minor role. The various possibilities of digital experimentation in the context of structural optimisation create new options for the designer generating forms, while enabling control over a wide variety of parameters. A complete mapping of the mechanical properties of a structure in a continuum mechanics model is possible but so are simplified modelling strategies which take into account only the most important properties of the structure, such as iteratively approximating to a solution via representations of kinematic states. Form finding is thus an extremely complex process, determined both by the freely selected parameters and by design decisions.