Data Driven Architecture

The Idea

More Details about this project at: http://zachsoflin.com/design-thesis/

While design based on performance is not a new concept, it can be applied to architecture in a way never thought possible before. Two conditions make it possible:

1.The age of computation in architecture and the part that it plays in the design process is continually growing.Currently, it is becoming much more an integral part in the development of design and the direct form of architecture.

2.Today’s humans are the most measured species in history. Never before has society known so many details and intricacies of how something works then it knows about humans past and present.

The coupling of these conditions lays out a direction for a new architectural process. The application of this process is what this project explores: In a world moving away from shallow aesthetic and toward performative space, how can architects produce a more accurate and performative form? This project sets up an automated framework to which data-sets are applied in order to generate and influence architectural form directly. The goal being a more performative & enjoyable space for its users.

WHAT IS A FRAMEWORK?
It is a basic structure underlying a system or concept. In this case it is the structure of a system based on given parameters that generate architectural form.
THE IDEA...
The generated framework could be applied to any site to produce a architectural product based on its users and context.

An analogy that works well with this concept is that of a automobile. All automobiles start with a basic unchanging framework that includes steering, wheels, and axles. It also has a specific size limitation (depending on the regulation for that type of vehicle). All of these factors could be considered a framework, or something to work within. It allows for changes while maintaining basic elements. Vehicles are then produced based on user needs and market demand. So, while the basic framework remains unchanged, the form and shape of the vehicle can vary quite drastically. This is the same idea as the concept of frameworks this project is trying to get across. Architects would lay out the basic framework and parameters for the building to fit within. Then the form would be driven by its context and users. The stark difference between vehicles and architecture also makes the argument for this type of process which is that buildings are much more complex than vehicles. While a cars parts may be more complex, there are many more parameters when it comes to a typical building. Human intuition can handle much of this and, no one can argue, does a fine job. The objective is not to get a computer to design an entire building. Rather, the objective is to use computation to aid in the processes it is good at. Finding solutions to quantitative problems is one of its strengths, where as lifestyle and feel is not. This creates
a team between architect and computer and allows the architect to layout a framework using intuition and then fill it in with form using computation.

The Base Framework

Using GIS demographics information (shown above) one can start to estimate the population of each surrounding building based on square footage, program, and number of units. Then, using a automated connection to Google Maps, directions from each surrounding building to the given site are drawn. After this, based on the results of a survey to determine tolerable walking distances, the buildings are parsed. For example, if the average walking distance to grab a load of groceries is 2 minutes, then only buildings populations within a 2 minute walk of the site are included in the population count. Buildings that are in range are then listed. Using the GIS data for these buildings gives an accurate snapshot of the user groups demographics. As stated previously, this process is automated, so as survey results are changing and updating so is the determined potential user group.

Using the results from the previous section, the demographics of each user group can be determined. These demographics are matched with different brands for each desired program to find the best result. An automated Google Places search is done as well to find other programs in the area, their size, and characteristics. Using numbers taken from each programs industry leader, one can calculate the needed square footage of each program in the area. For example, if there are 28,000 potential users for a grocery store, and the ideal grocery sqft per capita is 7.2 sqft (National Grocers Association), then, the needed sqft is 202,000. The difference between what exists in the area and what is needed shows how large that specific program should be. Then, this data and the demographic information is combined to determine which brand would perform best in each program.

The Tower Framework

The framework for the tower units is based on the typical floor size for each floor which is derived from FAR (Floor to
Area Ratio) and other factors. The idea is that through user input, users can start to lay out and determine what their unit looks like. In a way, this could be considered justified form in that users wanting to cantilever outward can do so, but must
pay for the increase cost of structure. This can be seen in the typical floor plan below showing that as one moves their unit outward, the $/sqft increases. The idea provides a framework (seen above) that users can work within to design their personal units. This framework is made up of 12’x12’x12’ modules which can be placed within a given grid and assigned a program (details about the program types to come). Associated with the program and all of the modules are costs. This ensures everyone is truly getting what they paid for. This also extends into how the units interact with their neighbors like view or patios. The following pages describe the process of producing a unit and what complexities arise when
doing so.

Once begun, the optimization process takes all units from its database and sets them to a default general location. It then begins to measure its values and determines whether the priorities were met or not. Below shows a visual image of how this process would work: 1. Placing the unit 2. Testing its values 3. Repeating.

The optimization process works on quantitative formulas, basing its decision on trying to maximize or minimize a given value. In this process priorities are scaled values, given a rating from 1-10. If the priority is met perfectly it receives a 10.
If not, it receives a lower value. All priority values are added to determine a final rating for the optimization. Since the goal of the optimizer is to get the highest number, it will continue to shuffle units, learning from its past, until the best result is
received.

Once all values have been assembled and the result is optimized, the tower’s geometry is solidified. This finalizes the
building allowing construction to begin.

Final Design

Aerial View

View From Wacker Drive

Roof Top Area

Typical Unit

Lower Level - Holds the parking and resident storage programs.

Base Level - The main entrance from the street level and pedway to the retail programs.

Wacker Level 1 - Contains the private entrance for residents from Upper Wacker Drive. It also contains the second level of retail functions including a fullsize grocery store.