AIGenerative DesignOptimizationSimulated AnnealingTechnology

Exploiting Simulated Annealing in Generative Design of Commercial Spaces

Generative design frees us from the limitations traditionally imposed on designers It is happening right now and provids designers with unimagined and innovative solutions to increasingly demanding customers in a complex and competitive world.

Architectural planning is iterative by nature. Planners navigate regularly through trade-offs that stem from conflicts between the physical space and obstacles within, regulatory limitations, assumptions they make regarding the quality of the user flow in space, proportions we like, stakeholders’ goals, etc. It seems, sometimes, that there are endless things to consider, which yields an enormous number of options. If we could only come up with more iterations, we would get the optimal plan, one that fulfils the wants and needs of everyone involved. 

In retail, the floor plan has direct impact on the revenue; therefore, it is crucial to obtain the best possible plan, one that reflects the brand’s identity and its commercial objectives.   

Generative Design, is an approach that uses computation to generate numerous design options based on pre-defined goals and constraints and enables the fine tuning of the design by changing the minimal and maximal values of parameters in which the code meets the set of pre-defined preferences. This approach has many advantages, since it allows planners to experiment and explore new planning concepts, get a large number of design options based on data, get an optimized design according to a specific criterion, and create a basis for discussion about the quality of the planning among the stakeholders.

For example a 300 Sqm store admits an immense number of options but, Which one is the best?

To answer this question, Xenia uses  Simulated Annealing, a randomized algorithm that approximates the global optimum of a given function. In metallurgy, Annealing is a technique that uses heating and controlled cooling of a material to reduce the defects of its crystals by increasing their size. In the same way, the simulation of annealing can be used to find an approximation of a global minimum for a function with a large number of variables to the statistical mechanics of equilibration (annealing) of the mathematically equivalent artificial multi-atomic system. This notion of slow cooling, is interpreted in the Simulated Annealing algorithm, as a slow decrease in the probability of accepting worse solutions as the explore solution. Accepting worse solutions is a fundamental property of metaheuristics because it enables a more extensive search for the global optimal solution. In general,  simulated annealing algorithms work as follows: At each time step, the algorithm randomly selects a solution close to the current one, measures its quality, and then decides to move to it or to stay with the current solution. This is based on either one of two probabilities between which it chooses on the basis of the fact that the new solution is better or worse than the current one. During the search, the “temperature” is progressively decreased from an initial positive value to zero and affects the two probabilities: at each step, the probability of moving to a better new solution is either kept to 1 or is changed towards a positive value; instead, the probability of moving to a worse new solution is progressively changed towards zero.

Making a decision is governed by a quality function, which is used to evaluate the quality of the current planning state and states that are near the current state. The code grades each planning state using  a combined evaluation function, which combines several different criteria.

 Following are diagrams that demonstrate various evaluation functions that Xenia uses as part of the optimization process:

View of the store by passersby:

Measures the visibility of the store as viewed by passersby.

View of the store from the entrance:           

Measures the visibility of the store viewed from the store entrance.

View of the store from the counter:             

Measures the visibility of the store view from the counter area.

Store Capacity:

Measures the number of fixtures in this planning state.

Entrance Size and Location    

Determines the size and the location of the entrance to the store and measures its impact on other parameters such as visibility and control.


Divider Size:                            

Determines the size and the location of a dividing element in the store and measures its impact on other parameters such as visibility and control.

Generative design frees us from the limitations traditionally imposed on designers. It makes the technology becomes a an active part of the process working within the goals, constraints and preferences defined by the user. The computer is not used to reflect the design but to co-generate it.

Leave a Reply

Your email address will not be published.