When is a design process truly data-driven design? The process needs to be driven by more than obvious data. In other words, the data must be significant, and not obvious. Results from parametric energy modeling in software is a good example. The program BEopt is at the top of my list.
Energy modeling in architecture is the process of creating a building model that can be used to understand energy flows. Energy is constantly flowing in a building. Examples include solar energy and exterior temperatures. Also.internal heating and cooling loads; air infiltration; heating and cooling sources such as boilers and air conditioner. These are all examples of energy being produced and moved in and out of a building. We often feel the effects of energy flows. A building is cold or hot. A heating bill is large or small. The sun feels good, or is overbearing.
Architects know design decisions influence energy flows. How can a design respond? Perhaps a preliminary design has a lot of glass on the south side. Is it too much, not enough, or just the right amount? A fairly sophisticated spreadsheet can answer specific questions about energy efficiency. Solving for a specific outdoor and indoor temperature, a spreadsheet-level calculation can tell how much energy flows through an envelope: the building’s heat loss or gain. In the case of window design, a spreadsheet can determine a satisfactory glazing level for that environmental condition.
But, the environment is constantly changing. A building cycles through cooling and heating periods throughout the day. How can an architect account for the constantly changing environmental and internal conditions? Using software whose purpose is to crunch large amounts of building and environmental data points provides the answer.
We use a program called BEopt to create energy models of buildings. The name is an acronym meaning Building Energy Optimization. It is available here from the National Renewable Energy Lab.
BEopt models the features of a building that define energy flows. The modeling is akin to filling in cells of a massive, multi-dimensional spreadsheet. BEopt takes user input data and performs computations on that data to see how energy flows in the building. Examples of user input data include the type of envelope, windows and doors. Add to this thermostat set points, types of energy-using equipment, lights and plugs. User input also includes schedules of energy use throughout the day or week; location and orientation of the building.
BEopt takes the user data and performs a series of calculations to determine energy flow, starting on January 1 at midnight. The calculations occur every hour, every 30 minutes, or even every 15 minutes. Is it warm inside and cold outside? Envelope losses are part of the calculation. Are there people inside? Body heat is part of the calculation. Does the indoor temperature cause a thermostat to turn on the heat? For every 15-minute period throughout the year, BEopt performs the calculations. Simulating a year, BEopt performs about 35,000 15-minute calculations.
This is a lot of calculating. In fact, it is far more than can be done by hand, or even on a spreadsheet. The data set is huge. In fact, the data set is too large for an architect sitting at a desk with a calculator, or even in front of a computer with Revit opened. Happily, BEopt takes all this data and produces a single data point. Namely, this particular building, with this set of parameters, used this much energy.
But wait! There is more.
Modeling and Optimizing
Knowing how much energy a specific building consumes is great. Our model may tell us the heating bill is $3,000 during a typical year. Can we reduce that heating bill? How? Smaller windows? More roof insulation? Better appliances? Rotate the building 30 degrees clockwise?
There are dozens of variables bearing on energy use. We may increase wall insulation, not knowing how much that will help. Will the wall insulation increase overshadow too little roof insulation? What we really need is a program that looks at dozens of variables in relation to all the other variables. We need a program that can look at many data points for each variable, and see how they respond relative to each other.
BEopt does this. It can calculate flows for a specific building configuration 35,000 times against a specific set of year-long environmental data. BEopt can also look at multiple settings for dozens of variables. For instance, we can compare four window u-values, three wall r-values, five window to wall ratios, two types of light fixtures…all against each other. The software simulates hundreds of permutations in 15-minute time increments to see how each unique set of permutations compares with others.
This type of parametric data-driven design lets an architect compare hundreds of similar buildings in a few hours’ time. Each building instance- the one with triple pane glass, LED lights, and r-30 walls, for instance- produces a single data point representing its annual energy use. we can compare this point to all the points representing all the other building instances, with different values for energy consumption parameters.
Optimize for Cost
Energy efficiency measures usually cost money. Are the energy savings worth the extra insulation cost, or the window upgrade? A graph of energy savings against the mortgage and all energy bills over a period of time will show this. BEopt can do this as well. The graphic at this article’s heading is exactly this. It shows a graph of energy savings on the X axis and total cost of ownership on the Y axis. Optimizing energy savings and minimizing expense puts the building in the sweet spot.
Is this the building we want to build? Or, would we rather have a less expensive building, even though it uses slightly more energy? Conversely, are we interested in a more energy efficient building, such as net-zero, even though the cost is more? Finally, which parameter values satisfy those three different goals? Building design that includes BEopt modeling can answer those questions. That is data-driven design.