AUGMENTED PARAMETRICS - Simulation and Expert Knowledge in Parametric Design

This research project is about parametric design in architecture. More specifically, it focuses on the integration of expert information and digital simulation into a parametric modelling system for digitally fabricated wood constructions. While its practical applications are very specific, the underlying issues it proposes to work on are of a fundamental nature and have larger implications. This research project takes on the main shortcomings of today`s parametric design systems, which can be summarized as a general lack of successful methods for design decision- support. These shortcomings have become especially pertinent in light of the recent implementation of the directive 2010/31/EU, which was adopted in order to strengthen the energy performance requirements on buildings. Tools that would allow designers to deal creatively with these more complex requirements as part of a normal design process are simply not available today. Current Building Information Modeling (BIM) systems not only use simple physical simulation models, they also lack some fundamental capabilities for design decision support. One main unsolved problem is that the integration of simulation systems is so far done in a purely one-directional way. We propose to develop an augmented parametric modelling environment with an open, modular structure that can provide real-time design feedback and suggestions for improvements and optimizations based on a variety of design criteria that define a building`s performance. To that end we will also explore ways in which more accurate simulation methods can be integrated in a bidirectional way into the parametric system. To investigate these issues in depth we will concentrate on just one particularly interesting type of wood construction: cross-laminated timber (CLT). CLT is a material with very positive characteristics in terms of energy-efficiency and sustainability and - as we have already seen in an earlier FWF project conducted at our institute (FWF grant L695), - it has large potential for customization through digital fabrication. The project proposes to develop what we refer to as "Augmented Parametrics": a new type of parametric program that supports performance-based modelling based on detailed physical simulation modules. By focusing just on one single construction material we can also take construction issues such as jointing into account, while also providing easy access to high level optimization approaches (e.g. genetic algorithms). The project`s ambition is to lay the conceptual and methodological groundwork for such augmented parametric systems and to test it with practical applications for a particular type of construction and simulation in a proof-of-concept manner.

The goal of this research project was to support architects in the design of energy efficient buildings. The EU Directive 2010/31/EU, which is currently going into effect across Europe, leads to stricter rules for the energy-performance of buildings and makes architectural design more demanding and complex than ever. Currently there are no tools available that allow architects to take the new demands into account already in an early design stage. In the project augmented parametrics so-called parametric design programs were extended to provide this support. Parametric CAD programs allow designers to quickly create variations of their designs because the form-defining variables, the geometric parameters of the design, are linked. The project extends this functionality with optimization-procedures, by which various relevant properties of the design can be calculated and optimized within the constraints of the parametric definition. With the help of these optimization procedures, a bi-directional parametric functionality was realized as part of the project. This means that a parametrically defined design can be controlled by changing its form-parameters as well as, so to speak from the other end, by setting performance goals for the optimization procedures which can be derived from the design and by which the program can calculate sets of parameters that will allow these criteria to be met. As architecture can never be reduced to a single criterion and the different relevant criteria (such as) are often contradicting each other, a multi-criteria-optimization process was developed as part of the project, which can be used with a number of different optimization modules. In order to be able to use these optimization procedures already during the early design stages it is essential that they can be executed as part of the design process, that is in real time. Typically optimization processes require thousands of iterations in which various combinations of parameters are being tested and compared. Simulation procedures that are carried out in every one of these iterations have to be extremely efficient, otherwise such optimizations will run for hours if not days and thus make the direct feedback during the design process impossible. Therefore simplified calculation methods were developed by which the performance-simulations could be integrated into such real-time optimization procedures. Comparing our calculation of energy demand for heating and cooling with the leading simulation programs, in our tests the results were within a 5% margin of deviation, but calculation speed was improved by a factor of 1000 or more.