We have been working on our second generative design case study, focusing further on light industrial sites.
Our scripts have been developed to add an increased volume of initial inputs/variables which will influence the outcome.
In this video we let the model run to see the initial outputs. The model created 2000 options initially.
Our initial output gave us a range of different extremes.
The graph on the right highlights the key output evaluation measures we set to be able to compare each option.
The graph shows 8 different options out of 2000.
From the outputs we decide that we would look to maximise the buildings on the site. This resulted in the model creating the options below.
We selected 4 options which were optimised to the site and analysed them against the original scheme we were given. The green results in the graphs represents the original scheme.
From the analysis we selected Site C. At this point we developed the site further in Revit.
One of the biggest challenges using computational design is knowing when to stop. We believe the finer design detailing and fine-tuning needs to be completed by the architect.
The architect will manually adjust certain areas such as landscape area or parking area to suit the needs of the client. This manual intervention is crucial to the design process.
Below is our final analysis showing the original site we were given, the optimised computer output and then the manually refined version of the final computer output, made by the architect.
You can see the architect decided to reduce the road area and increase the parking area and landscaped area. This manual intervention was made so the result would work to the original brief.
Below is our optimised result rendered in Revit and Lumion.