by Richard Johnson, Director of Sustainability, Rice University

When are reductions in energy consumption verifiable savings?

With the emergence of the ACUPCC and increasing focus on energy costs and supplies, universities across America are pursuing measures to reduce their energy consumption and their greenhouse gas emissions. As these schools attempt to measure their results and document savings, I ask how do they really know when they are saving energy?

Let’s assume that a campus building is metered for all utilities, and that these utilities can be tracked on a weekly basis. And further, let’s assume a two-week experiment, and that at the beginning of the second week space temperatures in the building are changed as part of a new campus building temperature policy to reflect what is considered to be a more efficient range. If the meter readings were lower in week two than week one, can a utility manager conclude that the energy conservation measure was a success? Given our experience at Rice University, we would argue that the answer is no.

The energy consumption of a building from one time period to the next is influenced by a number of variables, including outdoor temperature, humidity, time of day, day of the week, and day of the year. In the example above, week two could have been significantly cooler than week one, potentially leading to a false conclusion about the effectiveness of the new policy, and even masking unintended consequences of changing space temperatures. However, by creating a weather-normalized baseline model for energy consumption as our energy managers have done at Rice and then comparing this baseline against actual meter data, we submit that utility managers can be much more confident in interpreting their results.

Figure 1

How might one visualize this? Figure 1 presents one week of data for chilled water consumption at our student center, the Rice Memorial Center. The y-axis expresses chilled water consumption, and the x-axis represents time. The red line shows the modeled baseline for chilled water consumption for that building. The variation in the red baseline between daytime and nighttime is obvious, reflecting that we use more chilled water to condition the building during the day than we do at night. And yet, while the model for each day looks generally similar in shape, it is not exactly the same, because in reality these days were of course not the same. The blue line represents actual consumption, drawn straight from the chilled water meter at the building in near real-time. What we see is that due to a variety of conservation measures enacted in that building during the summer of 2009, actual chilled water consumption is now consistently well below the baseline model. Prior to these initiatives, the baseline and the actual meter readings would have been quite similar. These results are weather-normalized: we’re not having to guess whether the savings might be related to a cold front or a series of cloudy days.

Figure 2

We can use this system to express cumulative building-level savings (or losses) from electricity, chilled water, and steam in dollars. Figure 2 shows daily utility expenditures for the Rice Memorial Center over a 30-day period. The green bars represent actual daily costs, while the black lines are the predicted costs according to the baseline model. Notice how each day has a different predicted consumption? The blue space between the green bars and black lines indicates savings. On the right side of Figure 2, we see that over a 30-day period, we saved $4,931.49 in steam, $1,618.11 in chilled water, and $780.13 in electricity, for a total utility savings of $7,329.74.

The ability to plot meter data against a predictive baseline is a game-changer for campus energy conservation. Every two weeks, we hold an interdepartmental meeting to review the performance of a number of our campus buildings using this tool. Sometimes we see unexpected results that trigger maintenance work orders. Sometimes we find buildings whose nighttime setback temperatures have been placed in an override mode and need to be restored (and we can see the amount of money that we lost as a result of that decision). In the case of our own facilities building, when an unexpected electrical load caused us to consume more electricity than predicted by the model, we were able to estimate the size of the additional load, and our maintenance manager tracked it down to a baking booth in the paint shop that had been switched on and left on for several days. As one of my colleagues frequently observes, this tool allows us to shine the bright light of truth on how we’re consuming energy on our campus.

Rice’s approach to energy modeling is now the basis of a campus energy management product in development by Incuity Software, a subsidiary of Rockwell Automation. We are working to embed within this system the ability to track greenhouse gas emissions, which would enable us to display and report campus-level and building-level predicted and actual carbon footprints, divisible by type of utility. The position of our energy management team is that unless energy consumption is tracked against a weather-normalized baseline, we are suspicious of claims of actual savings. The implications for greenhouse gas reporting are clear: as we develop our inventories and compare them with previous years, did we enact measures that genuinely reduced our emissions, or did cooperative weather make us lucky? Without a proper baseline, we just don’t know.