Applying Generative AEC Dynamics to a Parking Garage
By Jesper Wallaert and Sylvester Knudsen for Autodesk University
This article explains how MT Højgaard has automated and optimized the process of designing a parking garage, from early concept to final estimation and simulation. We will take you through the workflow and show the software possibilities that include Autodesk Revit, FormIt, Project Fractal, and Dynamo.
The Parking Garage Configurator
MT Højgaard has undertaken the self-built construction of a parking garage for our new headquarters in Søborg, Denmark. The Virtual Design and Construction (VDC) department was asked to join the project and come up with some cost alternatives by optimizing some of the parameters behind the rule-based design of the concept. The standard parking garage concept is also developed to address the Danish market regardless of the site. We ended up having a data-driven design solution from early conceptual design all the way to construction documents and object-based estimation with four products connected through Dynamo.
Phase 1: Parametric Concept Model
We decided to use Dynamo for the creation of a rule-based parametric concept definition. We identified the parameters of the standard concept structure such as levels, lengths, number of parking spots, as well as other factors like total cost and average price per parking spot, all of which play an important role in the success of the project. We quickly created the foundation for a Dynamo approach.
The first phase would focus on the preparation of a proof of concept definition providing the base for the early-stage conceptual design model. In Phase 2 we would use the base definition from Dynamo Studio, and via Dynamo for Revit build a Revit generator. To provide a better end-user platform we ended up using Dynamo Studio and the Dynamo Customizer to bring the design into FormIt 360. The combination of the two products made it possible for us to meet our client inside FormIt. In order to evaluate the site we used FormIt and created the site context to try some further options with the dynamic model.
We built an output dashboard in the definition to dynamically show outputs for number of parking spots and price estimation in real-time while adapting to the client’s wishes. After finding the best fit for the client the outputs from the conceptual design model (Phase 1) are then used as inputs in the Revit generator definition which creates the detailed design model in Revit.
Phase 1.1: Generative Design
During the development we got access to the Project Fractal alpha program and with only a little adjustment prepared the definition to work inside Fractal. Project Fractal runs your Dynamo definition through the cloud to analyze a design space with possible inputs and helps the designer determine the configurations that best meet the criteria of fitness. Now we had a solution to automate the optimization process of finding cost alternatives. After adding a cost mechanism and color visualization on top of the original definition we fed it through Project Fractal to see which combination arrived closest to the cost target. From the options generated, we could then refine until we found five to ten optimal cost alternatives to present to the client.
While this does not necessarily design the entire parking garage automatically, it is an excellent tool to create thousands of garage scenarios optimized for parameters we choose and to validate a particular decision in the early stage of a design.
Phase 2: Parking Garage Generator
After locking the conceptual design of the parking garage structure, the next step is to turn the dead geometry created in Dynamo, FormIt, and Fractal into hardcore BIM objects. The process of converting the parameters identified in the concept model into a Revit model is based on a Dynamo graph which has the same foundation as the script for generating the concept model. Although the big difference is that instead of making Dynamo geometry, it now creates fully parametric Revit families, which can be used for estimation, fabrication, and so on.
Basically the script uses the data that was found in the concept model, and from that data generates the entire structural model of the parking garage. Form the data collected from the concept model, the script has built-in rules based on the data defining things like number of columns and beams, placement of structural reinforcement, the appropriate number of levels, and much more. From this data we are able to automate almost all of the modeling process.
To make sure that objects naming, parameters, and classification are always done in a standardized way, a special Revit template is created for modeling. This template contains all the parts used in the structure with about two or three alternatives for each object. By making sure the naming convention is always done in the same way, we create a link to our VDC software that enables us to do automated quantity take-offs and 4D simulations.
Phase 3: 4D/5D Model
In the last phase of the project, we go from 3D to 4D and 5D, through our VDC software iTWO. In iTWO we can perform quantity take-offs, estimations, and simulation of the building process. By using the generated Revit model, we are able to almost fully automate the processes of quantity take-off, estimation, and simulation. By using standard naming and classification, an iTWO template is made so that every time a new model is loaded into iTWO everything is set up and ready to give accurate price estimation and time scheduling.
By automating these processes we can present the customer with alternative solutions for the structure, which means the customer can now see the consequence of every choice, a lot faster. As a contractor we also get a lot of benefits through the simulations of the building process, as we can build the parking garage as many times as we want on the computer before we start the actual building process.
Sylvester Knudsen is completing a master’s degree in the field of Building Informatics at Aalborg University, while working at one of Denmark’s biggest general contractors, MT Højgaard, as part of the Computational Design and Construction group. The group’s goal is to implement and develop computation in the company’s daily workflows. Passionate about BIM, utilizing data, and computational workflows, Sylvester strives to make better projects by using data for better decision making.
Jesper Wallaert is working as VDC specialist at MT Højgaard. Jesper uses his experience providing Virtual Design and Construction research to share knowledge of digital building solutions. Jesper also manages new implementation and further development of computational design and construction at the MTH organisation.
Want more? Read on by downloading the full class handout at AU online: Applying Generative AEC Dynamics to a Parking Garage.
