Eliminating Data Uncertainties with Calibrated Building Energy Models

Nov. 20, 2014
Calibrated building energy models pinpoint exactly how systems operate. Onset’s Evan Lubofsky explains when and why it’s important to collect such refined data.

By Evan Lubofsky, Onset

Building energy models (BEMs), which simulate the view of energy use in a building, have been playing a more important role in building performance evaluation over the past few decades. While commonly used in building design, they can be particularly important in retrofit projects, where energy savings need to be identified and energy conservation measures proposed.

Despite their potential usefulness, the value of simulated models has been called into question in recent years as a number of studies have shown wide variances between predicted energy usage from models and measured building performance.

“There are skeptics out there who talk about energy models not being accurate enough and say that they can’t be used to guarantee results,” said John Nott, principal energy engineer with Atlanta-based Griffith Engineering. “However, I think it’s a misnomer. The reality is that accurate energy modeling is a very in-depth, difficult process. Skepticism is being fueled by the fact that a lot of people just aren’t doing it right.”

The industry’s answer to “doing it right” has been calibrating models to some form of measured or “real” data. At its simplest, calibrating an energy model is taking various inputs to a model – HVAC and lighting controls, building envelope, weather information, etc. – and adjusting them based on actual measured use. In doing so, the quality of the model inputs can be improved, and theoretically, the resulting model will be more reliable, credible, and useful in predicting future energy use and savings.

The need to calibrate

Beyond providing more legitimacy to energy models, there are a number of specific factors driving the need for calibration.

In performance contracting, for example, investment-grade energy audits typically have performance guarantees in place that specify a certain amount of energy savings once energy conservation measures (ECMs) have been taken. These financial guarantees provide a way for building owners to obtain and finance the necessary capital energy projects for their facilities – but the estimates need to be accurate.

“When involved in a performance contracting project, you have to deal with the fact that you’re going to guarantee results to your client,” Nott said. “If you say you’re going to save the client $100,000, they have to be confident enough to take that figure to a bank or other financier to get financing for it. In these cases, a calibration of your energy model is required.”

Measurement and verification, or the process of verifying energy savings after ECMs have been implemented, also requires models to be calibrated. Option D of the International Performance Measurement and Verification Protocol (IPMVP) covers “Calibrated Simulation” and calls for direct measurements of specific building equipment, and calibration of both the base and post-retrofit energy models.

Approaches to calibration

There are three primary ways in which energy models are calibrated: calibrating to utility data; calibrating to Building Management System (BMS) trend data; and/or calibrating to data collected with portable data loggers and meters.

Calibrating models to utility data is the simplest and most basic approach, and the rule of thumb is to have at least 12 months of utility data on hand. This data is typically coupled with site survey input from building occupants and operators. However, since utility data reports on whole building energy use, it’s not equipment-specific. So while you may to gain a sense for how a building is performing in general, you won’t get much insight as to which particular system(s) are causing increased energy use.

“If you’re a building owner and just want to know generally how your building is performing vs. knowing how much energy specific building systems are using, then calibrating to utility data is fine,” said Eveline Killian, a senior engineer with Burlington, VT-based Cx Associates.  “You may even be able to figure out that the source of your increased energy use is HVAC-related, but you won’t be able to know if it’s the chiller specifically, or the heat pumps, or something else.”

Marcus Sheffer, an energy consultant and partner with Wellsville, PA-based 7group, feels that a model can only be considered calibrated when it simulates the actual energy use of each of the major energy end uses – heating, cooling, and lights.

“A lot of people don’t understand the need to gather data for each significant energy end use in a building,” he said. “You can’t just calibrate to annual gas and electric use. It’s inadequate data gathering, which is something we see frequently when reviewing LEED projects.”

Another approach to calibrating models is using trend data from the BMS if one exists. This goes beyond utility and survey data by providing equipment-specific energy use data on a daily or even hourly basis. While BMS data can provide a decent amount of resolution, there can be some limitations.

“If a building management system hasn’t been calibrated, it can be way out of spec,” said Killian. “We often see this with buildings that haven’t been commissioned. And if the system is out of calibration and you’re counting on its trend data to inform your model, you can run into problems.”

According to Nott, the potential issues don’t stop there.

“BMS settings often get overridden by building operators, which can result in things like space temperatures not being set back to where they should be,” he said. “And, the BMS doesn’t always have sensors where you need them, so often times there are areas of a building where no trend data is available.”

Finally, it’s becoming more common to calibrate models to trend data collected with portable data loggers and meters. This approach typically provides the finest level of data resolution and information about how a building is performing, and is typically used as part of the investment-grade auditing process where performance guarantees are in place.

It’s becoming more common to calibrate models to trend data collected with portable data loggers and meters. This approach typically provides the finest level of data resolution.

The process of calibration is aided by data loggers significantly,” said Nott. “It’s one of those things where if we’re doing an investment-grade audit, we’re going to walk your facility, talk to occupants and staff, check control systems, and then put data loggers out to see how the building systems are actually operating. It’s a layer of due diligence we add to the process.”

Nott and his team routinely rely on a number of different data logging tools in modeling projects. This includes using temperature loggers to log space temperatures and determine if setbacks are in place, light on/off loggers to determine light use patterns during occupied and unoccupied times, and motor on/off data loggers to determine when fans, pumps and motors are operating.

“When we data log, we’re generally trying to capture a particular number that can be extrapolated out to represent standard building operations. It also gives us a way to validate how well a BMS is doing what it is trying to do. Data logging eliminates a number of critical uncertainties involved in energy modeling.”

Killian and her colleagues at Cx Associates also use portable data loggers in building evaluations, though usually only on larger building systems since calibrated modeling is typically a time-intensive and expensive process.

“We look at the influential pieces of equipment and their schedules to see how they are performing, so we know, for example, what the energy usage of the lighting systems is vs. a chiller plant,” she said. “Those are the kinds of modeling inputs that are very important. Using trend data from loggers enables us to calibrate models to the specific building equipment, its performance, and the manner in which it’s being used.  This makes the model very accurate to the present usage profile and as well as enable its use to predict future consumption and efficiency opportunities.”

A word about the weather

Commercial building energy use is typically influenced heavily by outside weather: HVAC heating in most climates is proportional to outdoor air temperature; peak cooling is affected by solar heat gain. Therefore, energy models must be normalized for the weather in order to be accurate and reliable.

The most common approach is to incorporate Typical Metrological Year (TMY) weather data files from local or regional weather stations into the model. While these files do an adequate job reporting on average weather conditions for every hour of a typical year, they aren’t always able to capture weather extremes. And, depending on the location of the weather station being polled, the data may not necessarily be representative of climate conditions at the building site.

“We use TMY3 weather data to normalize energy usage, in order to analyze the building performance over a typical year,” said Killian. “But calibrating a model to logger or utility data is usually done using outside air data loggers placed onsite.  Depending on the location of the nearest weather station, there can be significant differences between the local weather station and the individual building. For example, we have experienced that the weather station in Central Park can report temperatures that are five degrees below the temperatures at a facility within Manhattan due to the radiated heat from surrounding buildings.”

According to a 2011 report published by the American Society of Heating, Refrigeration and Air-Conditioning Engineers (ASHRAE) entitled, “Modeling and Calibration of Energy Models for a DoD Building,” a team of building engineers found that statistical weather data can vary up to nearly 23° F from weather data measured with site-specific weather stations.

Due to potential variations between offsite and onsite weather data, a growing number of energy auditors and consultants install weather stations and/or temperature data loggers right outside the building and use the site-specific data to calibrate their models.


Beyond making energy models more accurate, reliable and credible, calibration can add value in a variety of ways. It can help ensure energy performance guarantees, better quantify energy savings in measurement and verification projects, and provide building owners with important feedback on how buildings are actually performing. The process of calibrating models can be as basic as analyzing utility bill data, and as advanced as deploying portable data loggers throughout a facility to collect hourly (or less) interval data. And depending on site location, availability of local weather data, and the impact outside weather has on a building’s energy performance, there may be good reason to incorporate site-specific weather data into an energy model instead of relying on average weather condition data from offsite stations.

“It’s critical to gather as much information as you can about your building,” said Nott. “The more measured, real information you have, the fewer reasonable estimates you need to make when calibrating models. While there are always the components of skill and experience involved in energy modeling, the fewer estimates the modeler has to make, the more accurate the energy model will be.”

 Evan Lubofsky is a writer and communications director for Onset, a Cape Cod-based company that manufactures data loggers and weather stations for building performance monitoring.

About the Author

Kevin Normandeau | Publisher

Kevin is a veteran of the publishing industry having worked for brands like PC World, AOL, Network World, Data Center Knowledge and other business to business sites. He focuses on industry trends in the energy efficiency industry.