Predictive Energy Modeling at the Design Stage

An earlier version of this article appeared on www.sefaira.com.

Building energy models are often poor predictors of the actual performance of a building. There are many reasons for this -- among them, necessary assumptions about weather, construction, and usage that can never be perfectly accurate.

Of course, most energy models do not need to be predictive in order to be useful. Decisions about the shape of a building and efficiency of its envelope can be made by comparing the relative impact of different options.

Still, there are scenarios which require greater precision -- for instance, designing a Net Zero Energy building -- or a design team may want to better understand the uncertainty that is built into their model. In these situations, a recent Technical Memorandum from CIBSE, TM54, provides relevant guidance.

The remainder of this article will provide an overview of its recommendations and how they might be put into practice.

TM54 Overview

The primary recommendations from TM54 are:

• Use a Dynamic Simulation Model to estimate building energy use. (Most modern energy modeling programs fit this description.)

• Source complete and accurate data for operational parameters such as occupancy, schedules, lighting power, and plug loads, rather than relying on typical defaults.

• Present simulation results as a range rather than single deterministic numbers, to reect the fact that performance depends upon variable factors such as weather, occupancy, and operation. High and low estimates can be determined by exploring multiple scenarios.

• Compare results to existing energy use benchmarks to ensure that the results are reasonable.

Exploring Scenarios and Sensitivity

One of TM54’s primary recommendations is to assess the impact on energy use of varying key operational and design parameters. These variations can be used to present high and low estimates for performance, clearly indicating the uncertainty inherent in early-stage performance calculations.

A few of these factors are:

• Building occupancy & operating hours. Designers can create “high occupancy” and “extended hours” scenarios to understand the impact of these factors on performance.

• Equipment & lighting loads. Designers can study multiple scenarios, reflecting everything from “good practice” values to “worst case” scenarios that include poor management or the addition of unintended lighting and equipment.

• Efficiency of mechanical systems. HVAC systems do not always operate at their rated efficiency. Designers can look at realistic decreases to mechanical system efficiency.

• Weather data. This one can be more difficult to study, as sourcing good future climate data can be challenging. However, if future weather files are available designers test the impact of varying weather — including "higher emissions" and "lower emissions" scenarios as outlined by the International Panel of Climate Change.

While TM54 is focused on making models more predictive, the same methodology can be expanded to provide insight on the sensitive of design-related variables as well. Performing sensitivity analysis on factors like compactness, glazing ratios, shading amounts, building orientation, and envelope properties can help the architect focus on the factors that will have the biggest impacts on performance.

Sensitivity Analysis in Action

Below are example scenarios and sensitivity analyses for a 100,000 sq. ft. oce building in New York, NY. I explored five operational scenarios (left side of the chart), as well as the sensitivity of five design-related variables (right side of the chart).

The top five most sensitive parameters for this particular building were: (1) plug loads, (2) lighting loads, (3) glazing ratios, (4) operating hours, and (5) glazing properties.

Notably, three of these five are clearly impacted by the building design: glazing ratios, glazing properties, and lighting loads. And previous explorations have shown a 10 to 15% difference in energy use from varying building form alone. The architect’s decisions clearly have a major influence on the final performance of the building.

The remaining highly sensitive factors (plug loads and operating hours) present an opportunity for further engagement with the client, end users, and/or facilities manager to better understand, plan for, and monitor the intended and actual usage of the building.

In line with TM54, we would present the estimated energy use as a bar graph with error bars:

The error bars represent the high and low results from the sensitivity analysis. This communicates the range of likely outcomes, and can open a discussion about the factors upon which the final (measured) energy use will depend.