Imagine you are designing a new building. Part of your job is to understand and communicate answers to questions like the following:
Questions 1 & 2 treat the building as a structure. They are part of a larger question: What will the building look like? Tools such as Revit will help you communicate the answer.
Questions 3 & 4, on the other hand, regard the building as a system. They relate to the larger question: How will the building function? The recently released Insight 360 tool will automatically convert your Revit model into an energy model, and help you estimate the building’s environmental performance. The tools of the future will help you do even more.
In the future, architects will truly design buildings as systems. Their models will incorporate not only the physical characteristics of a building, but also the control logic driving its HVAC system and the social interactions that occur among its occupants. Human behavior affects the building, as people take action to influence indoor temperature, lighting, and ventilation. The building in turn affects human behavior. Future systems design tools will help architects simulate and visualize these complex relationships.
Here's what a systems design tool might display for the benefit of an architect:
The above is a simulation of hotel at around 12:00 noon. The warm outdoor air is heating the building from the outside in, as shown by the red-to-blue color gradient. The yellow cylinders represent hotel guests, and the yellow speedlines reveal their movements. Some guests are entering the building (bottom middle). Some are having lunch in the restaurant (bottom right). Some are traveling to and from their hotel rooms on the upper floor (top).
If the architect fast forwards to 7:00 pm, she might see the following:
It is now warmer inside the building than outside. The green glow effects indicate windows being opened by occupants who want to accelerate the cooling of the indoor environment.
Autodesk’s Complex Systems Research group is investigating how to help people design systems. Our publications on customizable simulated behavior, new building simulation algorithms, and multiscale visualization techniques describe the architectural design research that led to the images above. We also study systems approaches that could be applied not only to sustainable buildings but also Internet of Things devices, biomedical advancements, and other pressing challenges.
The systems design tools of the future must be easy to use, yet they must also scale. This suggests a combination of visual programming to simplify the task of modeling behavior, and systems theory to ensure that even the most challenging design problems become tractable.
One promising form of visual programming is called dataflow programming. It’s the same type of programming found in Autodesk’s open source parametric design tool, Dynamo. Dataflow programming is a perfect fit for at least two aspects of systems modeling: initializing a simulation based on a set of parameters, and aggregating simulation results into a set of performance metrics. This realization led us to the following interface layout:
The layout above places the simulation elements (blue) in the middle, with dataflow elements (yellow) on either side. The dataflow graph on the left is processed before the simulation, and the dataflow graph on the right is processed at the end. The black dots on the far left are ports representing parameters. The dots on the far right are ports representing performance metrics.
The simulation itself is based on an area of systems theory known as the Discrete Event Systems Specification, or DEVS. Our paper on the DesignDEVS simulation environment describes this approach in more detail. The basic idea is that simulation models communicate over time via messages that travel along links. The image below shows how these links might appear in a visual programming tool:
Combining the dataflow programming elements with the DEVS simulation elements, this is what a systems model might look like:
These mockups are described in our paper, Designing DEVS Visual Interfaces for End-User Programmers.
Update (April 2018): Node networks similar to those illustrated here can now be developed in C++ and simulated using the SyDEVS open source library.
The most challenging aspect of a complex systems model is that it must inevitably incorporate the expertise of multiple collaborators. People must work together, and future design tools must help them.
The images above depict a systems model with 4 simulation models. To promote collaboration to the fullest extent, each simulation model should correspond with a different area of expertise. For example, one model might represent the flow of heat through a building, another might represent natural and artificial lighting, another the behavior of building occupants, and the last might estimate resource consumption. In other words, each component of the overall system is itself system. And within these systems there may be more systems.
The success of future systems design tools depends on the extent to which they help experts collaborate. Their promise is to give designers a better understanding of how things will function, and how people will experience the built environment.
Here is a video showing how some of the systems described above may interact and affect each other. Enjoy!