In the creation of any type of structure, there is an energy cost required to produce and maintain the materials that make up that structure. In the natural world, energy costs can mean the difference between surviving and not surviving for a given biology. Therefore, it is advantageous for a given biology to reduce the amount of material used in its structure. In order to accomplish this reduction of material, structures in nature tend to be evolved to grow into forms that distribute stresses across their shape as uniformly as possible. Not only does this optimization strategy drive the location/distribution of material in order to create an efficient structure, but it also drives changes in a materials’ properties across a structure.1 According to AskNature.org, trees and bones achieve an even distribution of mechanical tension through the efficient use of material and adaptive structural design, optimizing strength, resilience, and material for a wide variety of load conditions. For example, to distribute stress uniformly, trees add wood to points of greatest mechanical load, while bones go a step further, removing material where it is not needed, lightweighting their structure for their dynamic workloads. At the scale of the cell, trees arrange fibers in the direction of the flow of force, or principal stress trajectories, to minimize shear stress.2
In 1992, Claus Mattheck at the Karsruhe Research Centre in Germany published a paper proposing a means of simulating the manner in which nature optimizes structure. Starting with an initial conditions defining object shape and material, support points and forces are then applied to the object. Over the course of multiple evolutions, finite element analysis (FEA) is used to calcuate young’s modulus of elasticity at each element within the object, the method removes elements that are under little or no strain until the remaining material left is under load. Termed “Soft Kill Option” by Mattheck, this method of using FEA to optimize a given shape is more generally referred to as Evolutionary Structural Optimization (ESO) and has been used in several industries over the last twenty years. Opel car company used the method to create a lighter, more stable engine mount and car body.3 Designer Joris Laarman created a line of furniture using ESO.4 Not only do these structural designs reduce material but one could argue there is a beauty in the elegantly curved shapes that result from the process.
The idea of evolutionary optimization in architectural structures has been made available by many different software programs. Most of the experimentation so far has been in the realm of truss or floor/column optimization. Recently, several designers have began to rethink how structure might create enclosures for humans. In a recently published project, Soft Kill Design’s protohouse explores how a house may one day be 3D-printed.5 There design uses evolutionary structural optimization and structure is printed along the force flow lines in the structure. The designers have stated that the design is meant as a provocation but its formal qualities do bring forth some important issues about ESO.
It is important for designers to understand that there is no such thing as “optimal”. The term relies on defining what parameters one is measuring. Using a measure of young’s modulus at each element, as is common in ESO programs, will result in structures that are generally lighter per their ability to withstand forces. However, replicating the range of all forces that may apply to an object is not how ESO methods are designed. There may be certain loads or forces that cannot be anticipated in the optimization process. In the case of the Protohouse, Soft Kill designers applied ESO methods to a starting object that represented an entire house. The result is a series of ‘living’ spaces that are very cave like in nature. As designers continue to explore the possibilities of ESO, it will be critical that they remain designers in their use of ESO methods, using it when and where needed and being able to add in the design vision not supplied by structural optimization.