Generative Design in Construction Essay

In the last two centuries, events such as industrial revolution, the growth of large-scale manufacturing industries, and mass production of goods have resulted in drastic changes in the automotive, agriculture, architecture and construction fields. These changes, especially in the car industries, led to the development of car factories that used sophisticated methods of design and production. The development of these industries greatly motivated early architectures including Walter Gropius, Van Der Rohe, and Le Corbusier who incorporated automotive design methods into architecture thereby ushering in an era of use of the computer as a design tool (Woodbury, 2010). In other words, the actions of early architectures led to the development of parametric design method. According to Weisberg (2008), the parametric design method is a technique enables an architect to intelligently design objects through the application of relationships and rules by use of a computer. The relationships and rules are defined in the parametric program used and have become increasingly easy to manipulate recently hence increase in speed of generating several iterations of the desired design in 3-dimension.

Is your time best spent reading someone else’s essay? Get a 100% original essay FROM A CERTIFIED WRITER!

Save 25%

The use of parametric design enables architects to control form hence allowing them to react to the context, the environment, and the respective regulations to create an exclusively digital workflow. In addition, the technique enables architects to explore and include fundamental aspects of construction such as materials, structural properties, and manufacturing procedures in the design phase. As a result, it has transformed the process of architectural design from that one of evolution to one that is iterative, reactive, and generative (Thompson, 2016).

While the parametric design has significantly benefited architecture through the transformation and demystification of complex issues into reasonable and comprehensible decisions, it has also introduced new design problems. The successful use of the technique demands a deep understanding of mathematics, geometry, and computer programs. Importantly, the architect must always remember that he must be the control and the master of the tool and not the tool as the control. Smith (2007) criticised the use of parametric design by highlighting its shortcomings. They include front-loading, difficulties in anticipation of flexibility, and drastic changes in breaking models. Even for experienced and competent designers, the networks of explicit functions weaved by the model sometimes become so vulnerable that starting from scratch is easier than making changes. Also, flexible models tend to be rigid in practice by restricting changes rather than embracing them (Weisberg, 2008).

Generative Designs

Generative design is a process/technique that incorporates evolutionary approach of nature into the design. The process starts with the designer defining the design goals in the generative design program together with parameters like manufacturing methods, cost constraints, and construction materials. Through the application of cloud computing, the technique explores limitless design options, tests the configurations, and learn from every iteration the feasibility of the options. In the recent past, the process has been hailed by scholars and architectures for enabling designers to come up with new options beyond the level at which a human can develop at the most effective design level.

Many generative design models are based on parametric modeling and algorithmic modeling. Parametric models are based on algorithms thus enable the designer to have more computational control during the design process. As noted by Aish and Woodbury (2005), parametric models are highly adaptable and flexible to the design criteria and requirements which make them popular during design exploration in dynamic and complex settings. Also, their parametric control of form is significantly valuable in performance-based designs because they enable the incorporation of performance analysis into design analysis.

The use of parametric modeling as a design synthesis technique allows for the expansion of the design space to explore several variances of a similar parametric model. As a result, the design principle defined in parametric relationships enables the architecture to explore a combination of design choices through time, assess design alternatives, and add value to the design template during the process of design (Aish and Woodbury, 2005).

Performance-Based Design

Performance-based design is a technique used to design a new structure or evaluate the integrity of an existing building that yields better outcomes compared to the conventional design methods (Wen, 2001). Through the use of performance-based design approach, a designer is able to work collaboratively with the client to determine the desired performance goals that might include structural strength and energy efficiency. Unlike the conventional methods of design that utilize codes to attain design objectives, the performance-based design does not follow specific method to realize the performance goals. Due to lack of adherence to specific goals, a designer has the freedom to develop unique methods and tools to assess the life cycle of the construction process starting from procurement to evaluation of outcomes. Design optimization and design rationalization are some of the methods that designers commonly use to manipulate parameters and achieve performance goals.

According to Moller et al. (2015), design optimization refers to the process of determining the best design tools that lead to the realization of the desired outcomes in an economical and timely way. Design optimization's primary objective is to help an architect to rationally search among many alternatives the best-suited design to meet the objectives. For the selected design to be acceptable, it must satisfy the design constraints. Design optimization automatically alters the design variables to enable the designer to identify the maximum or minimum objective function and satisfy all the necessary design constraints.

Typically, most optimization problems are expressed mathematically as shown below.

The optimization problem comprises of three main parts. They include:

• A vector of input data that defines all the likely designs
• A set/s of objective functions that define the objectives of the design
• An optional set of constraint functions that evaluates the practicability/feasibility of all designs.

After defining the optimization problem, the focus shifts to finding sets of input data that best suit the objective functions while satisfying the constraints. To solve such problems, designers use various strategies known as optimization algorithms (Talatahari, 2014). According to Talatahari (2014), there are two categories of optimization algorithms that include stochastic methods and deterministic methods. Deterministic methods realize the solution by directly applying a sequence of defined stages while stochastic methods realize the solution by introducing a certain degree of randomness.

On the other hand, design rationalization entails the elimination of unwanted variation through the elimination, reduction of complexity, and use of opportunities made available by the prefabrication and manufacturing approaches (Yoshimoto, 2009). Typically, use of appropriate analytical methods enables designers to benefit from significant cost savings. For instance, a designer can use finite element analysis for analysis of flat slabs instead of using the conventional methods described in the design codes.

Although design rationalization and design optimization use different strategies, they are somewhat used by designers independently or together to achieve performance goals. In particular, optimization is used when an unwanted design is inevitable. Unwanted designs are those that prioritize objective functions. However, less important functions are considered as constraints hence the need to address them. On the other hand, design rationalization is an approach used to achieve modular designs (Baldwin and Clark, 2000). In other words, design rationalization handles complex systems that do not have well-defined procedures or guidelines. Design rationalization aims at modularization to explain the metrics as well as to synthesize and communicate the design. It focuses on systematically supporting the design process and building confidence across the design stakeholders.

Designing in Performance

In the recent past, the idea that the availability of simulated performance feedback together with the automation and incorporation of performance analysis into the initial stages of design would lead to better performing designs has increasingly become widespread. The widespread of that idea has prompted scholars to explore the concept of "designing-in performance". Lin and Gerber (2014) defined "designing-in performance" as the technique for the provision of performance feedback to improve design exploration and the decision-making process that is unavailable in a conventional design process specifically during the initial stages. The development of multidisciplinary design optimization (MDO) to offer the required performance feedback has so far indicated effective ways of addressing the challenges of the existing performance-based processes. A study by Lin and Gerber (2014) developed an MDO framework called Evolutionary Energy Performance Feedback for Design and investigated its effects on the early stages of design.

From the results, Lin and Gerber (2014) established that the development of EEPFD led to the achievement of the criteria for the provision of a "designing-in performance" environment for designers. In particular, the results showed that the framework led to the generation of design alternatives and the simultaneous evaluation of the alternatives. In addition, the framework intelligently identified alternatives with better fit performance and provided a trade-off study for all the generated outcomes for use by decision makers. EEPFD as an advanced and unique design approach for use during early design stage can be verified by determining its ability to offer a better solution space within a given time limit and a demonstration of adaptability to a broad range of design conditions. The results obtained by Lin and Gerber (2014)'s study showed that EEPFD satisfied both conditions. Following the satisfaction of these conditions, it is true to say that EEPFD is viable for future studies to determine whether designers can practically use the generated data and design options to improve their decision making in the early stages of design (Lin and Gerber, 2014). It is also important for the future studies to focus on gaining a better understanding of user-oriented Genetic Algorithm settings that are available through EEPFD. Apart from geometric variability, it is important to measure the effects of genetic algorithm settings on the process and problem empirically. Currently, the real impact of the genetic algorithm settings on the entire solution pool is unknown and not quantified. Importantly, the optimum settings needed to offer the most effective solution are yet to be identified.

In conclusion, EEPFD can be applied in future studies based on the different observations of the unique needs of early stages of architectural design and early stages of other industries. Even with similar design specifications and energy requirements but different conceptual designs, many distinct resulting performance boundaries are noted. This suggests that the performance levels of generated design options are heavily influenced by the early design stage problem. Thus, the capacity of EEPFD to quickly determine the potential performance of several varying conceptu...