Architecture as an emergent property of a Complex Adaptive System (CAS) with self-organizational properties.
Abstract
We are starting to realize that the earth’s capacity is not infinite. Space, air, water, energy and resources are limited. This spurs the need for more sustainable ways of architectural design e.g., by incorporating cyclic processes of low-scaled interrelated actors. Ironically, the increasing global awareness of the importance of sustainability, has led to an ever-increasing number of local development projects. Citizens living in huge-scale housing areas are starting to lose their sense of particularity of place. They are looking for a way to increase self-reliance. Self-reliance increases citizen-initiated building projects and collective private commissioning. As a result, local contractors propel low-scale building activities and resources are retrieved from their close-by neighbourhood. Reduction of global planning induces self-organization with a massive diversity in forms of living spaces as a spin-off.
Nature is not familiar with the concept of waste: one’s output is another’s input. Organisms extract raw materials from their immediate environment and surroundings to construct their body materials and live together in a complex, tight knotted web of interrelated dependencies in which each organism has its own properties and function. The varieties of organismal forms are enormous. Once more, nature acts as a source of inspiration for small-scale, diverse, and cyclic design processes. Biomimicry, for instance, is an emerging field of research that applies natural (biological) principles and solutions into modern techniques and organizational structures.
Apart from showing us a lot of examples of cyclic interrelated processes, nature teaches us how efficient and adaptive biological processes at different scale levels can be: from highly efficient chemical processes in bacteria, to adaptive microstructure of bones in vertebrates, and the climatological properties of termite hills and social behaviour of bees and ant colonies. The well-known Darwinian principle of survival of the fittest postulates that only the most adaptive species survive, forcing species to react and adapt to their environment adequately. However, two more – lesser known – natural mechanisms increase efficiency and adaptivity of biological processes. These do not object the theory of evolution but are acting together in an interwoven way.
In the book On Growth and form, D’Arcy Wentworth Thompson describes the development and transformation of form of organisms. He shows that exoskeletons of algae show remarkable resemblances with crystallized soap bubbles. Soap bubbles reveal their form by decreasing their surface tension. Therefore their form is a result of finding equilibrium in a field of physical forces and form so-called minimal surfaces. Various architects examined these surfaces. In the 1970s and 1980s German architect Frei Otto (known by his Olympic stadium in Munich) studied the architectonical possibilities of these structures. In 2008, PTW architects and Arup engineers used clustered soap bubbles (i.e. Weaire-Phelan structure) as inspiration when constructing the Beijing National Aquatics center. Formfinding includes all the different ways shapes are generated by finding an equilibrium state in some physical force field. Parametrical design is a subset of formfinding. The research in the field of formfinding is propelled by increasing computationpower and rising quality and affordability of 3D printing. This technique makes it possible to produce real physical objects out of virtual models, which can then be studied more thoroughly.
A second natural mechanism that increases adaptivity and efficiency of biological processes are complex adaptive systems (CAS). In the 1990s, scientists studied the behaviour of non-linear systems, developing new scientific vocabulary like order, chaos, emergent properties and complexity. It was found that, despite their complexity, these systems are controlled by a small set of behavioural rules and interactions between relatively naive agents that lack a hardcoded blueprint or global system knowledge. Yet, these overall systems employ an efficient strategy, and follow adaptive behavioural patterns, and shapes. What is more, these emergent properties have a reciprocal influence on the interacting agents, sometimes causing the system to be self-organizing (i.e. autopoiesis). Examples of strategy include ants finding the shortest way to a food source, and the development of learning by machines and animals. Some self-organizing systems show properties of adaptive shapes. For instance termite hills, flocks of birds, fishes, and packs of cyclists. Yet, these typical shapes still are studied poorly by architects, one of the few exceptions being the Hyperbody Department at the Technical University in Delft, led by Kas Oosterhuis.
To bridge this research gap, I’ll describe the basic principles of self-organizing complex adaptive systems (CAS). And finally I’ll explore several complex adaptive systems in depth, and assess whether found properties can be valuable to architectural shapes and design.
Ants. Ants find their way to a food source by exploring en groupe each possibility while leaving a chemical trace (pheromone) behind them. The pheromone is detected by other individuals and set them in action to follow this route. If each possibility is followed by an equal number of ants, the shortest path will automatically carry the highest concentration of pheromone. Subsequently more ants will take this route till at the end most of the colony is using it. How fast the colony finds the shortest path depends mainly on the number of ants and the diffusion rate of the pheromone.1
Ants find the optimal route in a more complex environment with more than one food source too. Finding this becomes a task which is known by mathematicians as the travelling salesman problem. The essence of this problem is that each possible route has to be calculated and evaluated. The number of possibilities and therefore the necessary calculation power increases exponentially with increasing number of cities – or in our case – food sources. Ants, obviously, find their way without the use of a computer. In a situation with 4 food sources for example – with each food source placed on a corner of a rectangle – ants are able to find a remarkable shortest path. See here a paper (unfortunately in Dutch) about ants finding the so-called Steinerpoints. Steinerpoints are points at which 3 line pieces (paths) meet with angles of 120°. Noticeably this angles can also been found by clustering of 2 soap bubbles. Surface tension tends to minimize the individual surfaces which leads to system of 3 physical forces searching for an equilibrium. If both soap bubbles have an identical surface tension this equilibrium is found at angles of 120°.2
Obviously in many different systems appear comparable solutions, in a biological setting with ants as well.
Note that not all ants are following the shortest path. In a normal situation there is always a small number of ants exploring new routes. This is no waste of energy, but an investment in future. These ants show others alternative routes when former food sources are depleted. Therefore it is not necessary that the complete colony needs to go exploring again. This strategy saves time and energy globally. The ant colony specialized itself with 2 types of ants: explorers and foragers. This form of self-organisation leads to a more efficient strategy and is more resistant against – or adaptive to – changing available food sources.
It is easy to imagine when other programs (like defending against predators or reproduction) come into play the behaviour of the system becomes quite complicated with numerous solutions, strategies, specialisations and different ways of adaptivity.
Conclusion. The self-organising properties of the ant colony are dealing with finding optimal paths and specialisation. In bioarchitecture we are looking for some geometrical structure as a result of self-organisation. The former touches this in a tiniest way, the latest completely not.
Pack of cyclists. Another case of a CAS is a pack of cyclists. Cyclists tend to cycle as quick as possible over a course to a finish line. (To keep things simple I’ll leave team play strategies unattended). In fact the cyclists are kept within the borders of 1 dimension. Birds flying in a swarm have 2 and fishes gathered in a school have even 3 degrees of freedom. Principally these systems are equal but they differ in the degrees of freedom and as we shall see in a different reciprocal relation. Cyclists have to keep distance with others in order to prevent crashes. Contrarily they keep their distance small to benefit from reduced air drag when riding in a pack. Generally speaking the average speed of a pack is higher than of an individual cyclist. This implicates that in a race different cyclists lead the pack. For an individual cyclist, however, it is useful not leading the pack. These contradicting conditions shape the pack of cyclists with remarkable dynamics.
To some extent the dynamical structure of a pack might be compared with a fluid. Along the sidewalks the pack shows capillary-like behaviour. Narrowing parts in the course lead to congestion and a decrease in speed, while in the narrowing parts itself accelerations occur. Furthermore the pack seems to have a certain temperature. Sometimes a pack behaves like a rigid – almost solid-like – fluid when the pack is closed, width and driving at low speeds. It shows dynamical gas-like behaviour when cyclists tend to escape from the pack time after time. It’s interesting to investigate how long this comparison goes on, with phase changes in particular. In literature a lot of systems with moving people are described by models of fluid dynamics. For instance the traffic flow and the behaviour of people entering or leaving in stadiums.
Packdynamics differ from 2d and 3d models of swarms and schools. The difference is found in the different reciprocal relation between pack and the individual cyclist. The pack an the individual cyclist have conflicting intentions. An individual cyclist wants to finish first, alone, while birds and fishes want to stay together safely till the end. So packdynamics are quite interesting despite of its mono-linearity.
conclusion. The self-organisational structure of a pack of cyclists leads to a geometrical structure. The shape depends on the number and speed of cyclists, the air drag, the width of the course and the ultimate goal to finish first. The spatial structure has a changing reciprocal feedback at the individual cyclist. This changing influence leads to a dynamical but instable system.
- Scott Camazine, Jean-Louis Deneubourg, Nigel R. Franks, James Sneyd, Guy Theraulaz, & Eric Bonabeau (2003), Self-Organization in Biological Systems. Princeton University Press.
- D’Arcy Wentworth Thompson (1992). On Growth and Form. Cambridge University Press. Abridged edition by John Tyler Bonner.
—–
The examples show that the kind of agent will lead to a typical geometrical structure,