Base year selection can be a challenge in a real building, but a pragmatic approach can make things easier. As discussed in my previous article, 10 Steps To a Useful Energy Model, a base year is a representative year of building utility data used to help calibrate/validate an energy model. It is also used as a starting point to judge savings when applying energy retrofits to a model.
There are various ways to perform analysis of energy data while selecting a base year. A few of them will be covered here, but on real projects, with real buildings, there will often be a very easy way to select a base year – the only year of data you have. When asking for energy data it’s common to request three or more consecutive years worth. This is not going to happen every time. If less than twelve months of data is provided and the client desires a full, whole-building energy model, it may be very difficult to achieve a satisfactory accuracy.
Questions to Ask Regarding Utility Data
If a single twelve month period is provided, the modeller should examine the year carefully for unusual patterns and interview the building owners/operators to determine how representative the given data really is. Were there changes in tenants during the period? Were there additions, renovations or fit-ups? Was any major equipment malfunctioning, replaced or repaired? Were any energy retrofits completed during the period? These types of questions will allow the modeller to judge if it would be best to choose a different base year, or at least allow them to account for any differences during modelling. Ideally, the base year will contain no significant changes to tenancy, scheduling, building systems or equipment.
Other important things to understand about the provided utility data are:
- Are the readings monthly totals? Quarterly? Annual?
- Are the values consumption values such as kWh from a meter or delivery amounts such as for an oil tank being filled.
- Sometimes dollar amounts will be the only available data and energy will need to be calculated. If this is the case, be sure to understand what the energy rates were at the time (not just current rates) and whether or not the values included taxes, fixed monthly costs, etc.
- Is there a single meter or are there several sub meters? Does the data include them all?
- Do the observed energy patterns make sense? If the building is electrically heated, there should be a seasonal increase in electricity. If the building is closed for the summer there should be significantly reduced energy consumption during that season.
More Advanced Techniques for Base Year Selection
Ideally, several years of data will be available. In this case, the previously mentioned questions are still important, but additional techniques become useful. The simplest is referred to as a regression analysis and involves comparing the provided utility data with weather data for the year in question. Degree days are often used for this purpose. For those not familiar with degree days, several excellent articles describing their advantages and disadvantages can be found here and here. The technique basically correlates energy consumption with degree days, which then lets you judge the “goodness of fit” (R2 value) of the utility data. If the energy data correlates strongly with the degree days, then the data can be said to represent the building well. It also indicates that the data is strongly weather dependant. If there is little correlation between the energy data and degree days, the data is not weather dependent and so doesn’t provide as much information about how well the data may fit. A more detailed comparison would correlate energy consumption per day with degree days per day. More information about regression analysis can be found here.
Another common technique is called cumulative sum of differences (CUSUM) analysis. This technique is useful not only for base year analysis but also for watching for changes in a building’s energy consumption patterns during the course of its operation. The technique makes use of degree days again and allows a simple model of the building to be created. This model predicts the “expected” energy use based on the utility data and the degree days. This expected value is then compared to the actual value in the utility data and the differences are summed over time. A building operating as it should would have a relatively straight line on the generated graph (meaning the sum of the differences is low). If there is a sudden change in the graph, this could indicate a change in the building that would make it more difficult to use as a base year. More information about CUSUM analysis can be found here.
Normalization for Weather
A typical energy simulation will use some type of average weather data such as a 30 average for a nearby weather station. If the base year is used directly during calibration, this may introduce error if the chosen year was abnormally hot or cold. To achieve the better results, the base year weather energy data should therefore be normalized using weather data recorded during the actual base year. This is not always done, however, as there may not be enough weather data available or the desired accuracy may not require it.
Available Tools and Resources
Several tools exist to help in performing base year analysis.
- RETScreen Plus: A free excel-based tool created by Natural Resources Canada offers CUSUM, regression analysis, control charts, and various other tools.
- Metrix: This paid program is geared towards verification of energy savings, but offers comprehensive analysis of utility bills.
- ASHRAE Guideline 14: Contains detailed instructions on model calibration and base years.
Selecting a good base year has a strong effect on the results of an energy model so its important to choose a good one. There will be limits, however, based on the available data so ask lots of questions and state clearly what choices were made and why to ensure the client understands the chosen base year.