Conventionally, concepts and/or options are modelled suo moto, and analysis tools are used for validation in the later stages of design. Architects design parallel facade concepts at the Concept Design stage. Concept in this context means a design system using which different options can be generated – for example, horizontal fins or vertical fins or a combination of horizontal and vertical fins.
Upon consultation with the client, one of the concepts is selected to be developed further. The process of further development, typically involves manual modelling of 3-5 options based on the selected concept. Options in this context mean variations of the same design system- for example, for a design system of vertical fins, two options can be, 300mm deep fins at a distance of 750mm c/c, and 600mm deep fins at a distance of 1200mm c/c.
One of these options is selected with the client, which is developed in detail at the Design Development stage. Sustainability or Façade consultants are usually hired at the Design Development stage, whose input remain absent at the Concept Design stage where most of the key design decisions are taken. The drawbacks of the conventional workflow are as following:
- The initial design concepts are not compared quantitatively.
- Each initial design concept is not judged to its full potential, because the options of each design concept are modelled manually.
- Consequently, the options modelled are not exhaustive.
- Because the options modelled are not exhaustive, the final selected option is the local optimised option.
Need of a New Workflow
If the manually modelled options are plotted on a Cost vs Performance graph, they are noted to be randomly distributed. These options which are usually designed using thumb-rules, and compared within themselves ought to lie inside the shaded circle in the graph.
In the example of a design system of vertical fins, among the two, the option with 600mm deep fins will be better for daylight; but the global optimum might lie between 300-600mm depth and 750-1200mm c/c distance. Even for such a simple system, finding the best combination of depth and c/c distance is impossible without generating and simulating all the possible combinations.
Such an exploration is not feasible when performed manually because of the time required to do so. Consequently, there is a need of a system and a workflow which can take care of the following:
- Establishing a common quantitative quotient to compare parallel design concepts and/or options.
- Each design concept and/or option needs to be exhaustively explored to generate different possibilities and achieve full creative potential.
- The exploration needs to be automated to save time.
- With a common quantitative quotient, the auto-generated and exhaustive options can be optimised globally.
The requirements of a new workflow are met by the use of computation which exploits the speed, memory and calculative power of computer and applies them strategically in the design process. The parametric workflow has three key parts described as following:
- Parametric Representation– the parameters (numbers) chosen to represent the facade for optimisation. Every geometrical dimension of the facade and material specifications are kept flexible within permissible domains to generate the options.
- Optimisation Algorithm– the form-generating rule(s) that coverts and assigns the parameters into geometry and material specifications, and connects the parent software to an environment analysis tool. This process is set on loop by varying the parameters to automate exhaustive exploration. When the looping is concluded, the result is sorted according to their performance indices (or fitness) to filter out the best performing options. For more on optimisation, click here. <i eat bugs>
- Simulation– for each run of the loop in the algorithm, the option generated in the current loop is automatically exported to an environment analysis tool to use its existing simulation engine. The result based on one or more modules of optimisation (solar gain, daylight, glare, view etc) is imported back into the parent software and stored against the option number. This concludes one loop.
In the example of a design system of vertical fins, the depth and c/c distance are the two parameters. Assuming the domains to be 300-600mm and 750-1200mm respectively, and discretized at an interval of 10mm, every possible combination ((300, 750), (300, 1200), (600, 750), (600, 1200), (450, 900), (550, 930) etc) of the two parameters are generated and simulated for daylight. For each of the 1462 combinations, the result is stored in an array which is sorted to give the best performing options.
Parameters can be divided into the following two types:
- Intrinsic parameters – a value which represents a property of the façade which does not change with quantity, eg, visible light transmittance of glass, u-factor of aluminium etc.
- Extrinsic parameters – a value which represents a dimension or value of the façade which changes with quantity, eg, depth of fin, quantity of glass etc.
Both kinds of parameters can be variable or constant. In the example of a design system of vertical fins which is optimised for daylight, the depth and c/c distance are extrinsic and variable parameters. The other kinds of parameters which can constitute a complete parametric representation for this example are as following:
- Extrinsic and Constant parameters – width of fin, sill and lintel level of fenestration, length and height of facade.
- Intrinsic and Constant parameters – reflectance of fin, visible light transmittance of glass, reflectance of walls, ceiling and floor.
Ideally, all the above mentioned parameters excluding width of fin, and length and height of façade should be taken as variables for optimization. However, in most cases due to aesthetic preference or budget constraints, some or most of the parameters are kept constant. The selection of parameters to represent the design, the variable or constant nature of every parameter, the domain of every parameter and the discretization of every domain become the most important decisions in the process.
These decisions consequently affect the range of options that are generated. As a common measure, only extrinsic parameters are taken as variables with high discretization of domains when comparing parallel concepts, to understand the mutual differences. When exploring the global optimum within a design concept, as many parameters as possible are taken as variables with low discretization of domains.
The technique of parametric representation opens new ways of interpreting geometries. It marks a paradigm shift where relationships between elements are used to manipulate and inform the design of geometries and structures. The focus shifts from the process of modelling geometry, to the process of defining geometry. The designer’s attitude thus transforms from developing a pre-conceived idea of a form and directs towards the form-generating rules and the dependencies between the parts.