Design genetics

On the other hand, product adaptability is the capacity that a physical solution has for accepting modifications that improve its performance or that allow it to perform new functions. Again, depending on whether these modifications derive from expected or unexpected  changes in the requirements of users or other stakeholders, product adaptability can be specific (SPA) or general (GPA). Product adaptability is usually of relevance to users because it allows them to upgrade and/or customize a product in accordance to changes in their needs. Product adaptability usually implies the replacement of parts or the incorporation of attachments. For closed-architecture products, part replacement and the development and instalment of attachments are conduct only by the original equipment manufacturer (OEM). For open-architecture products, these activities are conduct also by other manufacturing companies, specialised contractors and even the users themselves. 

Adaptable product design

The creation of the first pencil, if it is meaningful to speak of such an event, would have to had involve first, the very concept of marking a piece of paper in a voluntarily and controlled way, then a seek for the material that could allow the execution of such task and then, a conscious or unconscious decision about the size and shape in which this material needs to be prepared. Ideally, such material would be light enough to be lifted and manipulated; strong enough to be pressed down on a piece of paper and pulled along without breaking; stiff enough so as not to change its shape under this action; soft enough so as not to tear the writing surface; and of such a nature as to leave a visible line on the paper. [This key point illustrates how the mere conceptualisation of an artefact conditions it to be used in conjunction with others. Thus artefacts are intrinsically complementary].

While lumps of lead or charcoal could certainly be serviceable as primitive pencils, they were also easily to criticise. Writing or drawing with a lump for an extended period of time can cramp the fingers and cause a diminishment in the quality of what is being written or drawn. Additionally, the relative bulkiness of the lump would hide from the view of users the very things that they are tracing thus making detailed work a difficult task. Finally, while the line made with a lump of lead might not be as dark as one would like, the line made with charcoal might be too dark and smudgy both on the paper and the hands.

There were of course, alternative device that allowed people to write and draw. Reed and quill pens have been available since the 4th century B.C. However, both devices required preparation of their points and repeated dipping in ink (which was at risk of being spilled or smeared). Another alternative was the “stylus” which consisted of a metallic device with a sharp end that was used to incise surfaces of wax. Nonetheless, just as the reed and quill pens, the metallic style exhibited major inconveniences. To begin with, it can only be used to mark surfaces previously prepared so it wouldn’t work on paper. Secondly, it could be used as a weapon and because of this in many places its use was forbidden or restricted.

Beyond the immediate complications the lead and charcoal lumps, the reed and quill pens and the metallic stylus supposed for their users, jointly, these limitations constituted a wealth of knowledge about what a device for writing in paper should look like. If a lump of lead was uncomfortable, the new writing device should be shaped to fit comfortably in the hand. If the bulkiness of the lump of charcoal blocked the users’ view, the new writing device should be made smaller. As far for the darkness of the line, designers of the epoch might try to heat the lead or mix it with other materials to achieve a darker mark. [This key point illustrates how existing artefacts provide a sort of toolkit for conceptualising new solutions].

However, as no other writing material was known, users had to settle with what was available. What finally accelerated the emergence of the pencil as we know it today, was the discovery around the middle 16th century of an easily mined abundance of graphite (also known as black lead) in northwestern England. In 1599 the natural historian Ferrante Imperanti wrote on black lead “it is much more convenient for drawing than pen and ink, because the marks made with it appear not only on a white ground, but, in consequence of their brightness, show themselves also on black; because they can be preserved or rubbed out at pleasure; and because one can retrace them with pen, which drawings made with lead or charcoal will not admit”. [This key point illustrates how the function and symbolic load of a product develops by contrasting its performance with other products].

By 1610, lumps of black lead were sold regularly in the streets of London to be wrapped on paper or fit into wooden, copper, silver or gold tube cases to prevent hand for being stained. As great as were the advantages, offered by cased-black-lead devices, they no doubt had some disadvantages as well. Depending on exactly how the case held the piece of black lead, it might easily work itself out, get loose, slide back into the case or even fall out of it.

In view of this situation and making use of advancements in techniques and tools for wood cabinet making, a Keswick maker developed a manufacturing system that consisted in having rods of black lead glued and enclosed in pieces of pine or cedar. In this way, not only the piece of black lead was firmly hold, it could also be efficiently exposed as the wood was whittled away with a knife. Since users were used to sharpening their reed and quill pens, this represented no inconvenience for them. Additionally, wood casing provided structural strength that allowed the black lead to be slender enough to be formed into a fine point that facilitated fine writing or drawing. [This key point illustrates how users can actively contribute to the enhancement of an artefact. In this case, the expertise of the users with sharpening instruments, made the pencil a more precise instrument. Additionally this key point shows how an artefact can inherit features not only from similar objects (reed and quill pens) but also from objects that belong to different categories of solutions].

This manufacturing process remained the standard until the English black lead became scarce. To address the lack of material, makers tried to make pencils using black lead extracted from other parts of Europe. However, this alternative material had such a level of impurities, that pencils made with it scratched and teared the paper. In an attempt to stretch the available supplies of English black lead, makers mixed dust and powder of it with binding agents such as gum, shellac and wax but the leads produced in this way were too difficult to sharpen and use without breaking and the did not produce a mark of very good quality. [This key point describes a situation in which changes made on an artefact (the pencil) creates areas of opportunity for improving complementary objects. In view of the situation described above, knife and paper makers could have implemented improvements on their products to couple with the features of existing pencils].

Such situation remained until 1794 when Nicolas-Jacques Conte developed a process for manufacturing pencils that consisted of mixing powdered black lead from which impurities were previously removed, with potter’s clay and water and rubbing the wet paste into long rectangular molds. When the leads dry, they were taken from the molds, packed in charcoal, sealed in a ceramic box and fired at high temperature. Since these leads could not easily be planed flat, they were inserted in wooden cases of a modified design. These cases (developed by early German pencil makers) had a groove about twice as deep as the thickness of the rod of black lead. A slat of wood was then glued in over the lead to completely filled the grove and the pencil was ready to be finished as desired.

Even though the technical success of Conte’s process, it was believed that new pencils could not write as nearly as well as these made with English black lead. However, when the choice is between a very good but prohibitively expensive pencil and an affordable good enough one, concessions could be made. [This key point illustrates how materials and artefacts just like users, exhibit limitations. A material can not be exploited perpetually (at least in the same way). Similarly, the behaviour or performance of an artefact is constrained by the materials of which it is made of. Thus, just as artefacts are design to couple with our physical capabilities, there are times in which users have to couple with those of the artefacts and materials].

Reference:

Due to such common capacity for self-organisation, Jane Bennett considers that both nonorganic and organic bodies exhibit “conatus” which is a force that Spinoza describes as a “trending tendency to persist”. From and ontological perspective, having conatus implies for any body, the dual possibility of affecting and being affected by other bodies whether they are nonorganic or organic.

According to Deleuze, these two possibilities (affect and being affected) vary inversely one to the other but their sum is constant. In this regard, Lucretius highlights that there is a proportional relation between the degree of internal complexity of a body and its possibilities of affect and being affected. Concretely, this author states that “And whatever possesses within it more forces and powers, it thus shows that there are in it most kinds of primordia and diverse shapes”.

Similarly, Spinoza highlights that there is no existing body that is not actually composed of a very great number of extensive parts, parts that come to it from elsewhere. Following this line of thought, Jane Bennett claims that both nonorganic and organic bodies link together through a process of “addsorbiton”. Such process is defined by her as a gathering of bodies in which they form a coalition and yet preserve something of their conative nature. [The concept of “addsorbtion” might be useful to describe how artefacts acquire new features and/or components. Ex: Smartphones that “addsorb” a Leica camera, a Bose speaker, etc.].

Through addsorbiton Bennett explains, cells and atoms integrate into materials, materials into components and organs and these, into artefacts and living beings. Once this point is achieved, artefacts and living beings can integrate into what she refers to as an “assemblage”, which is nothing more than an ad hoc grouping of nonorganic and organic complex bodies that are able to function in despite of the frictions and tensions that exist between them and between the constituent parts of each. [The concept of “assemblage” might be useful to describe and assess interacting systems of the type: artefact-user-artefact].

Using Spinoza to explain the above, Bennett claims that within an assemblage, both simple bodies (components and organs) and complex bodies (artefacts and living beings), exhibit conatus. In the case of the former, conatus is expressed as an inertial tendency to persist while in the latter, conatus refers to the effort required to maintain the specific relation between their constituent parts. [The concept of “conatus” might be useful to describe on the one hand, the wear of a product and its components due to its action on another device and viceversa; and on the other, the financial and technological limitations that condition the function, behaviour and structure of such artefacts].

Further exploring the nature of assemblages, Bennet highlights its dynamic nature. Because within an assemblage, each body regardless of its scale, suffers the actions on it by others, if it is to persist, it must constantly seek to creatively compensate for the alterations it suffers while being subject to the element of chance or contingency intrinsic to any event. As a result of this, the arrangement of an assemblage can change, that is, existing “alliances” between its parts can change and dissolve and new ones can emerge. [This key point might provide a theoretical foundations for improving the functionality of a product and/or delaying its obsolescence].

Although assemblages have uneven topographies because some points at which their components cross paths are more heavily trafficked than others, Bennett characterise their self-organisation as decentralised. To support this claiming she introduces the concept of “fractal causality” which occurs when affectations become infused into the organisation of an assemblage and thus cannot be said to be fully different form the effects engendered.

As an example of the above, Bennett describes the blackout that took place in the United States on August 2003 and which affected 50 million people over approximately 24000 square kilometers. From her perspective, the blackout was the endpoint of a cascade of voltage collapses, self-protective withdrawals from the electrical grid, and human decisions and omissions.

What seems to have happened on that day was that several initially unrelated generator withdrawals in Ohio and Michigan cause the electron flow pattern to change over the transmission lights, which led, after a series of events including one brush fire that burnt a transmission line and several wire-tree encounters, to a successive overloading of other lines and a vortex of disconnects. One generating plant after another separated from the grid, placing more and more stress on the remaining participants.

The U.S.- Canada Power Outage Task Force reported that the blackout was caused by a variety of agents and situations which included first electricity, with its internal differentiation into “active” and “reactive” power; the power plants, understaffed by humans but overprotective in their mechanisms; transmission wires, which tolerate only so much heat before refuse to transmit the electron flow; a brush fire in Ohio; energy-trading corporations, who by legal and illegal means, had been milking the grid without maintaining its infrastructure; consumers, whose demand for electricity grows and is encouraged by the government without concern of the consequences; the federal Energy Regulatory Commission whose Energy Policy Act of 1992 deregulated the grid, separated the generation of electricity from its transmission and distribution, and advanced the privatisation of electricity, etc.

Reference:

The advent of the office chair

There is no single inventor of the office chair; the elements that define it (movement mechanism, adjustable features and casters), all appeared on different chairs at different times. [This key point illustrates how artefacts from a single category of solutions can exchange or inherit features].

The first movement mechanisms for office chairs were developed in the United States in the 1840s and 50s with steel coils, cast iron components and steel leaf springs. For example, Thomas E. Warren’s Centripetal Spring Armchair of 1849 featured arched steel leaf springs that allowed the chair to flex in any direction. [Although not mentioned here, the author suggests that the interest of endowing chairs with movement capacity was inspired by rock chairs. Again, artefacts from a single category of solutions can exchange or inherit features ].

The earliest known example of a chair on wheels was a William-IV-Style armchair modified by Charles Darwin for his study in Kent, England in the 1840s. He replaced the legs of his chair with cast-iron bed legs mounted on casters so that he could move from specimen to specimen with greater ease. [This key point illustrates how artefacts can exchange or inherit features with or form those that belong to a different category of solutions].  

The exact origin of adjustable features is unknown, but by the 1880s CWS Keighley produced models on which the resistance against the reclining of the chair could be adjusted.

Factors that constrain the design of office chairs

Hierarchical organisation described by Frederick Winslow Taylor in 1911 on his book “The Principles of Scientific Management”, had a long-lasting effect on chair design. From the early twentieth century, up until the 1990s, executives, managers, and secretaries typically sat on chairs that reflected their status. As well as being more robustly constructed, executive chairs tended to be made of costlier materials and sometimes had more sophisticated movement mechanisms. By the late 1980s, it was common for an office chair series to offer at least three hierarchical distinctions and sometimes even four or five. [This key point illustrates how artefacts are made to resemble a system of beliefs. In view of this situation, when attempting to develop a solution that complements the function of an existing product, designers should also make sure that the incoming solution matches or complements the symbolic value of the product with which it will be related].

However, as the personal computer (PC) became common in offices during that same decade (1980), the hierarchical workplace start to dissolve into the collaborative and communal organisations of multifaceted workers that we know today. Back in the 80s, this situation merged with an interest in designing more ergonomic office chairs that in turn emerged during the past two decades after information gathered about the human body during WWII entered the public domain through books such as Henry Dreyfuss’s “Measure of Man” and Diffrient’s “Humanscale”.

Before PCs, files and machines were located throughout the office so employees used to move around far more than they do with their own PC. Thus, while ergonomic design during the 70s and early 80s focused on supporting the body, ergonomic design from the mid 80s focused on supporting the body for eight consecutive hours of seated PC use. [This key point illustrates how the introduction of a new product which apparently has no direct relation with the office chair can en up affecting its design severely].

To couple with the new ergonomic requirements, designers counted with a growing set of materials and manufacturing techniques. While the earliest office chairs were made with wood, cast-iron, and steel bar or sheet that were upholstered with batting and fabric, in the early twentieth century materials such as steel tube, sand-cast aluminum, aluminum and bakelite were already being employed in chair production.

Later on, technologies of WWII allowed the use of die-cast aluminum, moulded fiberglass and plastic resin, rubber mounts, industrial strength glues and compound-moulded plywood for chair fabrication. By the 1960s, transparent thermoplastic and injection moulded plastic were employed in the fabrication of chairs such as the D-49 and the Pollock chair. Since then, plastics have undergone a rapid evolution, with frequent introduction of higher-performance plastics. There isn’t an office chair on the market today that doesn’t employ some form of this material.

The next factor that severely impacted office chair design is the emergence of sustainability as a societal goal. Because of this, manufacturers are now striving to eliminate aspects of the manufacturing process that are harmful for the environment. As a result of this, in 2003 HermanMiller presented the Mirra chair which was 69 percent recyclable by weight, made with 42 percent recycled content, and was design to disassemble easily for recycling or to have its parts replaced.

According to the author of the book, the most recent factor that is conditioning chair design is the mobile device. Today’s smartphone, tables, phablets and laptops, allow people to work just about anywhere and in any position. Because of this, in 2009 three chairs were released that accommodate a broader range of movement: the ON chair, the Generation chair and the 360 chair. [Again this key point illustrates how the introduction of a new product which apparently has no direct relation with the office chair can en up affecting its design severely. Will these affectations propagate to the desk, tableware, cubicles, etc.?].

Reference:

Donna Haraway does this, motivated by the idea that ontologically heterogeneous partners become who and what they are in relational material-semiotic worlding. Natures, cultures, subjects and objects do not preexist their intertwined worldings. Pigeons for example, have very old histories of becoming-with human beings. On chapter 1, she presents a couple of stories in which people and pigeons tangle together in innovative ways that render each other capable of a finite flourishing. [This key point illustrates how human practices do not precede artefacts, instead they nurture and co create each other].

California Racing Pigeons and Their People: Collaborating Arts for Worldly Flourishing

In Project Sea Hunt in the 1970 and 80s, the U.S. Coast Guard worked with pigeons who were better at spotting men and equipment un open water that human beings. Indeed, pigeons were accurate 93 percent of the time, compared to human accuracy in similar problems of 38 percent. The pigeons perched in an observation bubble on the underside of a helicopter, where they pecked keys to indicate their finds. When they worked with people instead of isolation, pigeons were nearly 100 percent accurate. Clearly, pigeons and coast Guard personnel had to learn how to communicate with each other. In no mimetic ways, people and birds had to invent pedagogical and technological ways to render each other capable in problems novel to all of them. [This key point illustrates how to enable cooperation between two multispecies players an interface is required. In this case, observation bubbles and keys were required so Coast Guards and pigeons can dialogue. Based on the above, research efforts could be focused on developing design methodologies for “bridging” artefacts and technologies. This key point also highlights how people can adapt their capacities to those of other agents (in this case pigeons). This situation might not be exclusive of multispecies players, but might also apply for artefacts].

PigeonBlog

This project which took place in August 2006, gathered together homing pigeons, artists, engineers, and pigeon fanciers in order to gather and distribute information about the quality of air in Los Angeles. This project who was coordinated by Beatriz da Costa consisted in equipping racing pigeons so real-time air pollution data could be gathered and contrasted with the information provided by local government. In order to make this possible, artists-researchers and pigeons fanciers had to render each other capable of mutual trust so they can ask the birds for their confidence and skill. Additionally, artists-researchers and pigeons had to learn to interact and train together with the mentoring of the pigeon fanciers. In short, all the players render each other capable: they became-with each other.

Reference:

From this point of view, what some audiophiles call “the absolute sound” is not absolute in the sense of being universal, but is rather a value formed through the process of acquiring cultural capital; a process that, for audiophiles develops through exposure to music and to high-end audio systems for music reproduction at home. On this matter, J. Gordon Holt commented on the first issue on the magazine “Stereophile Reports on Equipment” that:

“…the hard fact is that perfection, like the end on the rainbow where the pot of golds is stashed, is always a hilltop away. And as you advance, it recedes, because every subtle improvement in the fidelity of sound is reciprocated by an enhancement of the perfectionist’s ability to hear ever more subtle imperfections”. [This key point shows how artefacts’ features and users’ skills co-create each other].

High-end audio systems are small-scale constructions in which the user is as much part of a feedback loop as the technical components, and in which aesthetics concerns about sound quality and desires for interactivity motivate user action and interaction with these components. High-end audio systems were often comprised of four main parts, each of which could be considered a small system itself:

1.     Source components, such as record players;

2.     A preamplifier, which provided source selection, volume and tone controls;

3.     A power amplifier, which amplified the signal to a level that could drive the final component;

4.     Loudspeakers;

The characteristics or devices that audibly disrupt the aesthetics qualities of music, and thus music’s emotional power, prompt audiophiles to manipulate system components or their arrangement in the room to reduce or eliminate that audibility. In consequence, the feeling that CDs were negatively affecting the sound of high-end audio systems led to more direct interaction between audiophiles and CD playback, and to the emergence of technologies that enabled system-building practices that were more compatible with the audiophiles’ cultural capital. [This key point illustrates how users can actively improve the performance of existing objects by making them work together in a complementary way].

Prior to the CD and CD playback release, some audiophiles feared that the sampling standard employed to develop CDs and CD playbacks would be too low to provide sufficient fidelity. Once the advent of these technologies such fears were corroborated. An early adopter wrote on The Absolute Sound Magazine:

“After listening to a CD and LP copy of the same performance of Strauss’s “Also sprach Zarathustra” I realised that while the vinyl was sweet and airy, the CD was harsh and constricted. There was no musicality, just sound”.

In addition to the affectations exerted over music, other audiophiles pointed out the negative effects the CD technologies had over sound systems performance as a whole. Specifically, audiophiles claimed that CD technologies obscured differences between system components that would otherwise be audible with analog sources. [This key point illustrates how not considering the devices with which an incoming product will interact, can generate a negative impression on users].

Through direct manipulation and exploration of CD playbacks internal components, audiophiles discovered that while some feature of them were standard throughout several brands, there were non-standardized components that appeared to be contributing to diminish the fidelity of music reproduction. In other words, the sound engraved in CDs may be a lot better than CD players have been so far designed to extract. Such realisation, lead to critics and analysis that helped manufacturers to upgrade CD playbacks to a point that progressively became acceptable for the audiophile community. Curiously, the technical improvements that allowed this, were not implemented on the digital or “high-tech” part of the CD players, but rather in their analog electronics and mechanic components. [This key point illustrates that when developing a product that combines several technologies (in this case analog and digital), no prioritisation should be given to any of these technologies, but rather they should be coupled together in a balanced way. Additionally this key point illustrates how the meaning of product develops in relation to the artefacts with which it interacts. In this case, CDs were initially perceived as “poor quality” devices because when played on CD players they jointly provided a lame experience (according to audiophiles). Only later, it was realised that CDs were no the problem].

Because of the above, Oudshoorn and Pinch state that users and technology are in fact “two sides of the same problem”, rather than “separate objects”. Users can have a profound influence on the development of a variety of technologies and technological systems and bring on the table priorities that vary according to their community and cultural backgrounds.

Reference:

Redesign however is not exclusive to existing products. New products are generally derived from similar products (Li, Kou, Cheng, & Wang, 2006). In fact, more than 75% of all engineering design activities involves reusing existing prior design knowledge to solve new problems (Iyer, Kalyanaraman, Lou, Janyanti, & Ramani, 2003; Lou et al., 2003).

The authors present a redesign approach aimed at combining two or more products from a product cluster to generate a new artefact. The approach contemplates the following steps:

Choose a target product

Choose two or more references from a product cluster

Ask users to describe and rank their needs for each reference separately.

Identify components with similar function in each reference.

Ask designers to weight the impact of each common component on each user need.

Analyse interactions between key components using Taguchi analysis method.

Apply structured design principles to resolve design conflicts between key components.

What follows is a draft of how three GD techniques, shape grammars (SG), cellular automata (CA) and genetic algorithms may be integrated into compound GD model for assessing product evolution. This initial proposal is based on the design approaches that Singh (2011) identifies for each of these techniques:

Shape grammars: emergent shapes (function follows form). This technique could be used to generate a hierarchical set of components (components, assemblages of components, assemblages of assemblages of components, etc.)

Genetic algorithms: this technique could be used to identify optimal combinations of the components and assemblages developed in the previous phase.

Cellular automata: this technique could be used to model the growth of the artefact as it goes through several stages of its life cycle

In design, genetic algorithms (GAs) have been mainly used for optimisation (Mitchell, 1997; Salge et al., 2008) not only because the quality of solutions tend to increase with each generation, but also because they allow an undirected exploration of the design space. Commonly used terms in GAs with respect to design are:

Gene: the smallest unit of a genotype

Alleles: alternative forms of genes

Genetic structure: a set of genes with a certain order or relationship

Genotype or chromosome: the genetic constitution of a design, rather than its physical appearance

Phenotype: the observable properties of a design, its form.

A basic technical description of a GAs have been proposed by Singh (2011):

Components and requirements: alleles (building blocks), chromosomes/genotypes (combinations of building blocks), phenotypes (solution), population (set of solutions), genetic operators and fitness function.

Rule application: Usually one operator is applied at a time on a sub-set of the population.

Main advantages: regular design evaluation and improvement, multiple solutions, optimisation, disruptive innovation.

Main limitations: progress slows down after achieving near optima solutions.

Level of accuracy: contingent on problem formulation, i.e., choice of genotypes and fitness function

Core development aspects of GAs in the context of design are the following:

User intervention: low designer intervention once fitness functions, genotypes and termination conditions are defined.

Based on the ideas discussed above, the author distinguishes five models of what Mitchell (2005) refers to as digitally mediated design. The first of these models is the computer aided design (CAD) model. In this model, designers manipulate geometrical shapes to develop 2D and 3D representations of physical objects. These representations are then analysed using predictive methods and design modifications are implemented accordingly. The second model is the digital formation model. In this model, designers can either manipulate the topology of digital objects using modifiers such as non-uniform rational b-splines (NURBS), and/or establish  structural inter dependencies between various geometrical objects.The third model is the generative model. In this model, designers specify rules, relations and principles that lead to the emergence of complex geometrical shapes. The fourth model is the performance-based model. In this model, digital geometry results from the simulation of the environmental conditions in which the product is supposed to be used. The fifth model known as compound model, is that which integrates performance simulation with generative and formative processes.

In what follows, the description of the generative model is expanded in order to specify the role that genetic algorithms (GAs) play within it. In short, the generative model is the design of rules, relations and principles that enable the emergence of complex shapes. Currently, two sub approaches to the generative design model exist: shape grammars and evolutionary design. As described by the author, shape grammars consists in the specification of a set of mathematical relations that will guide the sequential assembly of predefined geometrical components. In evolutionary design in contrast, form emerges by means of a genetic algorithm. Because of this, the term "form" in this sub approach, is used to describe the structure of strings of data not geometrical bodies. Perhaps the main limitation in the use of GAs is that the evolutionary process is completely automated which leaves no room for designers to actively interact with it. Other evolutionary models that are inspired in the concept of morphogenesis are beginning to play an important role in digital design and require further revision. 

Elements of a GA

1. Populations of candidate solutions

2. Fitness function: determines the extent at which a candidate solution satisfies pre-selected criteria. The fitness of each candidate solution within a population is represented in the fitness landscape. 

2. Selection operator: selects candidate solutions in the population for reproduction. The fitter the solution, the more times it is likely to be selected to reproduce. 

3. Crossover operator: randomly chooses a locus and exchanges the subsequences before or after that locus, of two candidate solutions. 

4. Mutation operator: randomly flips some of the bits in a candidate solution. Mutation can occur at each locus in a string. 

How a GA works?

1. Create a population of N candidate solutions

2. Calculate the fitness of each design solution in current population 

3. Repeat the following until a new population is created:

3.1. Select a pair of parents from current population; being probability of selection an increasing function of fitness. 

3.2. Use probability Pc to determine if crossover will take place. If no crossover occurs, create two offsprings that are exact copies of the parents. If crossover occurs randomly select a locus at which parents will exchange their string sequences, and create two offspring.  

3.3. In any of the cases above, mutate the two resulting offspring at each locus of their strings with probability Pm. 

4. Replace population 1 with population 2.

The Workflow

The workflow initiated with the construction of a simplified 3D model of a building.

Analysis grids were applied to the roof, south facade and the several floors at the work plane to conduct daylight factor (DF) analyses.

The design of the shading devices for the building parts described above, was explored independently by altering four parameters (angle, depth, number of devices and distances between them).

At the next stage, Galapagos GA solver was used to optimize daylight factor (DF) and solar irradiation (SI). On the basis of this, the number of shading devices, their depths and their proximity were defined.

Once these three parameters were fixed, only the tilt angle of the shading devices was left to be manipulated in the design exploration process. The possibility of changing orientation of the shading devices along the height was explored.

At the last stage, all floors were simulated together and optimize to achieve the target DF and to minimize variance between the analysis grids.

The workflow described above highlights the following aspects of GA-Oriented Design:

GA-Oriented Design commences with preliminary research and data gathering that enables designers to identify the parameters the GA will use to explore the design space.

GAs can be linked to parametric modeling tools to produce 3D shapes for each of the constituent components of a design solution.

Solution generation

These technologies attempt to assist and automate designer’s actions. Within this category the authors identify:

Parametric modeling systems which represent geometric relationships organized in a hierarchical binary tree structure that is automatically updated and visualized on the screen when changes in the parameters values occur. Although it facilitates design reusability, parametric model has limitations in terms of the variation beyond its preconceived scope, prematurely reducing the range of options to explore.

Expert systems, which apply existing knowledge of design domain in similar situations emulating human decision making based on rules. The main limitation is that these systems are limited to already known solutions.

Case-based reasoning systems, which reuse and adapt previous solutions for similar new problems. The problem is defined as a set of constraints that are satisfied in order of importance. The main disadvantage is that these systems do not learn and have issues synthesizing a large number of constraints for complex problems.

Generative design systems in which the resulting model evolves from an initial state through iteration by operations that imply geometrical and topological transformations. Generative design can lead to apparently creative outcomes since every new combination of parameters brings the opportunity to look for the emergence of new properties or affordances from the resulting composition. Although it is an approximation to the co-evolving dialogue between problem and solution, these systems lack reformulation mechanisms.

Agent-based design systems which rely on the interaction of active autonomous entities to create shapes that satisfy material, fabrication and geometric constraints. These systems have the limitation that they can both produce unexpected remarkable results or fall into constraint circularity and non-sense loops.

Solution evaluation

These systems analyze solution candidates and provide feedback to support decision-making. Within this category, authors identify:

Performance evaluation systems which estimate how a product executes a given function under stress. This design framework focuses on what the product should achieve instead of how it should be built. The disadvantage of these systems is that it is difficult to execute them in early design stages because of ambiguity and incompleteness in the design definition.

Design rule checking systems, which evaluate the fulfillment of rules and provide feedback. However, the designer must interpret the reports and provide solutions. Determining how to make or suggest corrections in real time during the design process remains a challenge.

Constraint-based systems, which capture design knowledge in the form of constraints and requirements that must be satisfied by the design. If constraints are reached or broken immediate feedback is provided, supporting the co-evolution of the problem and the solution. However, constraint-based methods only detect when constraints are violated and do not provide clues about how to address the problem.

Solution selection

These systems reflect trade-offs, values and preferences through quantitative indicators complemented with relative weights to reflect issues of certainty of information and preferences of decision makers. Within this category, authors identify:

Utility methods, which attempt to formally describe the preferences of the decision maker. These methods are hard to implement during the early stages however, because they rely on explicit information that is not always available at these stages.

Multi-criteria decision methods, which establish metrics to evaluate how a design fulfills given objectives. These approaches aggregate data into results to measure performance, which may also be normalized to establish a level of comparison of different design attributes. However, because the comparison is based on quantitative values, it is difficult to apply such methods in early design stages.

Optimization methods, which enable designers to define and search through large spaces of designs.Standard optimization approaches however, require a high level design specifications before design decisions can be made, thus it is not suitable for early design phases.  

Integration of the generation, evaluation and selection processes

Within this category, authors identify the following approaches:

Custom, which links generation, evaluation and selection via one-to-one selective data exchanges. E.g. software that links analysis packages with conceptual tridimensional models of buildings to eliminate file exchange and improve the interactivity between design, evaluation and selection. The linkages are labor intensive in terms of implementation and with limited reusability because of the specificity of the integration.

Interoperability, which involves exchanges among systems that are based on industry standard neutral formats. Examples can be found in Building Information Modeling (BIM) paradigm in the Architecture, Engineering and Construction (AEC) industry. These exchanges poorly preserve parametric features and have not yet attempted to represent the simultaneous interaction of different sources of design knowledge while designing.

The general structure of a GA can be considered to be analogous to the process of Darwinian evolutionary theory in which the characteristics are transmitted from one generation to the next by genes and organisms evolve under the pressure of fitness proportionate reproduction.

How a Genetic Algorithm works?

The GA commences its search of the design space not from a point based on an initial guess but from a number of randomly selected points. To commence a global search of the design space, an initial population of design solutions is generated. Each solution is represented as a string of variables (constituent elements) that can acquire different values (alleles). The fitness of each design solution is then determined by evaluating the extent at which they satisfy  set of predefined criteria. Designs are then assigned with a “selection probability” that is proportional to their level of fitness. In this way, while any design may be selected as a candidate for the next generation, chances are that the most fit will be selected several times and the least fit not at all.

Whilst some of the selected designs are passed without modification from one generation to the next, some of them are selected to crossover. This means that a portion of the selected designs will be arranged in pairs so they can exchange part of the information contained in their strings. A limited amount of new information is introduced into the process by means of a mutation operator which causes randomly selected digits within the combined strings  to change value in accordance with some predetermined level of probability. Mutation can also be applied to uncombined strings (designs that were not subjected to crossover).      

Chunks of constraints: implicit relations that describe important aspects of the design problem such as the dependency among the geometry, limitations, cost and timing of an artifact’s components (On the role of computational support for designers in actions).

Conceptual structures: arrangements of components that convey specific meanings. E.g. the hallway (composed by walls, windows, etc.) is an architectural concept that conveys “accessibility”  (On the role of computational support for designers in actions).

Design crossover: a mechanism of genetic algorithms that causes the exchange of information between some of the selected designs within a population. Crossover enables the better adapted traits to spread among generations.  (Developments on the use of the genetic algorithm in engineering design).   

Designer-centric computational tools: systems that support the actions designers execute while they manipulate the patterns that drive the arrangement of the physical parts of an artifact (On the role of computational support for designers in actions).

Design mutation: a mechanism of genetic algorithms that causes randomly selected digits within design genotypes to change value in accordance with some predetermined level of probability. Mutation enables the search for good solutions to be extended into regions of the design space for which the genetic code either was not exhibited by the original population or has been eliminated due to its association with some unfit characteristics.  (Developments on the use of the genetic algorithm in engineering design). 

Design schema: encrypted logic of organization that structures and gives identity to a design  (On the role of computational support for designers in actions).

Efficiency criteria for design: the proportion of desirable effects and the efforts expended to obtain them at each stage of an artifact’s life cycle (The Semantic Turn, page 189). 

Explicit knowledge:

Premature fixation: the resistance to make design changes resulting from the premature complexity of the structure of the models (On the role of computational support for designers in actions).

Product cluster: group of heterogeneous products that use different physical components to perform similar functions (Redesign for product innovation)