# Energy Modeling Versus Reality

January 06, 2010
A version of this article appears in the January/February 2010 issue of Home Energy Magazine.
SHARE
Modeling software can be very useful for helping you understand how a building works. As I’ve described in previous articles, you can calculate heating and cooling loads, HVAC air delivery requirements, and duct layouts. You can tweak construction assemblies to your heart’s content to see, at least in theory, how a building’s energy consumption changes. You can change orientation and glazing to estimate the impact of various passive-solar strategies. The combinations of variables are endless.

Underlying this process are usually complex algorithms based on research done by dozens of well-respected researchers from some of the most prestigious institutions in the world. The key question is, How well do these modeling programs reflect reality? After all, if we want to predict the effect of a change to a building, we have to know exactly how that change affects the building’s energy consumption. This question raises a variety of issues.

First, you have to decide how to measure reality. That’s usually done in some standardized
unit, such as kWh for electricity or thermsmfor natural gas. However, there may be other useful metrics—daylight penetration, occupant comfort, humidity levels, temperature changes, room-by-room load calculations, and so on.
Next you have to decide when to measure energy consumption. The choice is simple for new buildings—you start once the building is finished and occupied. Unfortunately, particularly in residential construction, you rarely go back after at least one complete heating and cooling season, collect the utility bills, and compare the actual energy consumption to the modeled consumption. It’s just too much trouble.

For existing buildings, if you have the utility billing history, it’s possible to analyze that history and separate out the heating, cooling, and baseline loads. Armed with that information, you can, at least in theory, calibrate a model of the building to the real-world data. With a well-calibrated model, you should then be able to predict how building improvements will affect utility costs. That’s the premise behind most residential modeling software.

Utility Bill Disaggregation

Analyzing a utility bill history, often called disaggregation, is a straightforward process. You can find a detailed description, for instance, in the Saturn Energy Auditor Field Guide. Here’s a simplified summary, assuming electricity for cooling and natural gas for
heating and the availability of at least 12 months of utility bills:

▪ Add the three lowest monthly electricity usage amounts together, divide by 3, and multiply by 12. That’s the annual electric baseload.
▪ Do the same for natural gas consumption. That’s the annual natural gas baseload.
▪ Subtract the annual electric baseload from the annual electric load. The remainder is the cooling load.
▪ Subtract the annual natural gas baseload from the annual gas load. The remainder is the heating load.

This approach is a good starting point for analyzing a home’s loads, but it has three serious limitations. software Energy Modeling Versus Reality The most important limitation has to do with accuracy. Without a detailed understanding of both the house’s mechanical equipment and the behavior of the occupants, this approach can produce only rough estimates. For instance, if the family tends to smother the house with Christmas lights during all of December and January, and the primary heating source is electric, the heating load will be overestimated. If several family members visit for several days over the holidays, the additional hot water usage will be incorrectlytreated as heating load. If there is a pool pump that runs several hours a day during the summer months, that electricity will be lumped in with the cooling load. Another limitation has to do with the billing information provided with net metered PV systems. Some utilities provide only net consumption data for each billing period. Instead of stating exactly how much electricity was consumed and how much was produced, they provide only the kWh that the consumer had to pay for. It’s impossible to calculate baseload information with those aggregate data. The third limitation has to do with midstream changes to the house. Perhaps you have a detailed 12-month utility history for a home in a cold climate zone. Unfortunately, the family swapped its 15-year-old 58% efficient water heater for a 98% condensing tankless unit in the eighth month of the history, which happens to be September. You might see some obvious changes in gas consumption that help you to quantify the difference between the old and new water heater. However, if the family tends to take more and longer showers in the fall because it’s colder and the kids are going to school, the changeover of water heaters might not be so obvious.

Modeling—The Next Level of Detail

A better way to predict the effect of building changes on energy consumption is to create a computerized model of the building. There are lots of programs to do that. DOE maintains an extensive list of these programs at http://apps1.eere.energy.gov/buildings/ tools_directory/alpha_list.cfm. On this list, there are a handful of programs used by HERS raters, home energy auditors, and Home Performance with Energy Star contractors. I’ve discussed REM:Rate and EnergyGauge in previous articles. Both of these programs are RESNET certified to do HERS ratings. Treat, which I will discuss in detail in a future issue, is another popular choice. (Treat should be RESNET certified by the time you read this article.) In California, EnergyPro is the most widely used application for analyzing existing homes. Regardless of which program you decide to use, the process is pretty much the same. You do a detailed takeoff of the home, including information about orientation, glazing, floor area, volume, insulation, and so on. You may also conduct duct testing or blower door testing and record the results. Finally, you make a detailed inventory of the mechanical equipment and appliances, and collect information about occupant behavior. (The level of detail you need depends on the program you are using.) You then enter all this information into the software and ask it to calculate the home’s energy usage. All of these programs calculate energy usage in both units and dollars, if you enter utility rates.
At this stage, in a perfect world, the projected energy usage should match the actual utility bills fairly closely. In fact, at least in my experience, it seldom does. Although I haven’t done this kind of modeling many, many times,
partly because I work mainly on new construction, I always find the results disappointing. I’ve discovered a few reasons for this (no doubt there are quite a a few reasons that I haven’t yet discovered):

▪ Most software applications just aren’t very good at accommodating unusual occupant behavior, such as the Christmas light extravaganza. They typically assume that there are always the same number of people in the house; that they all use the same amount of hot water; and that utility usage other than heating and cooling loads is consistent from month to month.
▪ None of the programs I’ve used is able to handle tiered rate structures properly. Some don’t allow tiers at all—you have to do your own analysis to come up with an average rate that is then used to (inaccurately) calculate utility costs. Some allow tiers, but not enough levels (five levels is not unusual in my part of the country).
▪ There are often limitations on the number or type of appliances or mechanical systems that you can specify. Restricting the data to one refrigerator or one A/C condenser is too limiting and results in inaccuracies.
▪ Some applications don’t allow certain types of equipment, such as solar panels or solar hot water collectors, to offset electrical consumption or hot water usage.
▪ There may be limitations on the type of building structure that you can specify. One program has a library of shell structures that cannot be edited and that doesn’t include stucco as an exterior finish. It’s not very useful for the California climate.
▪ It’s very easy to enter incorrect or inappropriate data. I’ve yet to use a modeling program where I’ve understood all the data entry parameters and how they affect the final results. Some items that seem critical have little or no influence on the computed numbers. Conversely, some seemingly unimportant item may have a huge effect on the outcome. That’s clearly my problem, but I suspect I’m not the only person with that problem.

I don’t raise these issues to criticize any particular software application—it’s very hard to create a program flexible enough and accurate enough to predict building energy usage. I mention them just to call attention to the fact that just because it’s software, that doesn’t mean it’s right. It’s important to understand the limitations of your tools.

Tweaking to Nirvana

Let’s assume you have software that can handle all or most of the parameters of the house you are analyzing, and you have historical utility data, and you have what you think is an accurate picture of the house’s construction and mechanicals. You plug all that information into the program, you press the Compute button, and you look at the results. Most likely some or all of the modeled utility costs won’t even be close to the actual utility costs. What can you do (besides utter a few choice words)?

The first thing you should do is reexamine all your data. Maybe you entered the utility rates incorrectly. Maybe you added or dropped a digit on the blower door or duct test CFM results. There are dozens of possibilities.

If nothing turns up, the next step is to start playing with some of your assumptions. If predicted electrical consumption is lower than the real thing, you might want to look at your appliance inventory. Maybe you didn’t know about the really old refrigerator in the garage that holds only a six-pack of beer. Maybe you’re making assumptions about the thermostat setback parameters that aren’t accurate. Again, there are dozens of possibilities. The problem is that it can be very timeconsuming to quiz the occupants to figure out what’s missing in your analysis, or tweak the modeling parameters, all in an attempt to get the utility cost predictions to match the utility bills. It’s like sealing the envelope on an existing building—you should only do what’s cost-effective. The model-versus-real world mismatch becomes even more important when you start suggesting improvements, and the software calculates utility savings and payback periods. How can those numbers possibly be reliable?

There is no easy solution to this dilemma.
You can pick your software tools carefully and understand thoroughly how to use them, minimizing user errors. Unfortunately, that doesn’t help you when the program predicts thatreducing duct leakage from 23% to 10% will reduce your client’s utility costs \$350 a year and pay for itself within 3.25 years. You have to be suspicious and, more importantly, make your clients understand that those numbers are only estimates—and that they should be
suspicious as well. If a bank is willing to accept those hypothetical numbers to finance an energy-efficient mortgage, so be it. The bank generally doesn’t come back a year later to check the reductions. It should, however, also be aware of the risks and caveats.

Over the next few years, we can expect to see an increased emphasis on energy efficiency, deep energy reductions, net zero energy homes, the 2030 Challenge, and myriad other initiatives taking place all over the world. Let’s hope that software manufacturers will incorporate features into their programs that will make it easier to predict real-world energy consumption accurately—especially for existing buildings with historical data. I plan to look at this issue in more detail in future articles. If you know of any residential software applications that you think are good at modeling energy consumption and costs, please contact me. I’d like to take a closer look.

Steve Mann can be reached through e-mail at steve@green-mann.com. Krigger, John and Chris Dorsi. Saturn Energy Auditor Field Guide. Helena, Montana: Saturn Resource Management. To purchase the book, go to http://srmi.biz.
• 1
• FIRST PAGE
• PREVIOUS PAGE
• NEXT
• LAST