| ABSTRACT | | | | facility. |
| Utility Bill Tracking systems are at the center of | | | | Like Benchmarking, you can determine your own |
| an effective energy management program. | | | | rules of thumb for your buildings, however, your |
| However, some organizations spend time and | | | | range of acceptable Load Factors will vary based |
| money putting together a utility bill tracking | | | | upon building type and climate. Rules of Thumb |
| system and never reap any value. This paper | | | | may not be that helpful though. Like |
| presents three utility bill analysis techniques which | | | | Benchmarking, just identifying the buildings with |
| energy managers can use to arrive at sound | | | | unusually high and low Load Factors, relative to |
| energy management decisions and achieve cost | | | | the other buildings in the portfolio, should be |
| savings. | | | | sufficient. |
| INTRODUCTION | | | | Load Factor Summation |
| Utility bill tracking and analysis is at the center of | | | | Load Factor can be used to identify billing and |
| rigorous energy management practice. Reliable | | | | metering errors, buildings that are not turning off |
| energy management decisions can be made | | | | equipment, and buildings with suspiciously high |
| based upon analysis from an effective utility bill | | | | demands. While Benchmarking can identify buildings |
| tracking system. From your utility bills you can | | | | most likely to yield large energy efficiency |
| determine: | | | | payoffs, Load Factor Analysis can point to easily |
| - whether you are saving energy or increasing | | | | resolved scheduling and metering issues. |
| your consumption, | | | | WEATHER NORMALIZATION |
| - which buildings are using too much energy, | | | | Another important utility bill analysis method is to |
| - whether your energy management efforts are | | | | normalize utility bills to weather. Weather |
| succeeding, | | | | Normalization allows the energy manager to |
| - whether there are utility billing or metering | | | | determine whether the facility is saving energy or |
| errors, and | | | | increasing energy usage, without worrying about |
| - when usage or metering anomalies occur (ie. | | | | weather variation. |
| when usage patterns change) | | | | Suppose an energy manager replaced the existing |
| Any energy management program is incomplete | | | | chilled water system in a building with a more |
| if it does not track utility bills. Equally, any energy | | | | efficient system. He likely would expect to see |
| management program is rendered less effective | | | | energy and cost savings from this retrofit. |
| when its utility tracking system is difficult to use | | | | A quarter-million dollar retrofit is difficult to justify |
| or does not yield valuable information. In either | | | | with results like this. And yet, the energy |
| case, fruitful energy savings opportunities are lost. | | | | manager knows that everything in the retrofit |
| Many practical energy managers make the smart | | | | went as planned. What caused these results? |
| choice and invest in utility bill tracking software, | | | | Clearly the energy manager cannot present these |
| but then fail to recover their initial investment in | | | | results without some reason or justification. |
| energy savings opportunities. How could this be? | | | | Management may simply look at the figures and, |
| This paper introduces three simple and useful | | | | since figures don't lie, conclude they have hired |
| procedures that can be performed with utility bill | | | | the wrong energy manager! |
| tracking software. Just performing and acting | | | | There are many reasons the retrofit may not |
| upon the first two types of analysis will likely save | | | | have delivered the expected savings. One |
| you enough money to pay for your utility bill | | | | possibility is that the project is delivering savings, |
| tracking system in the first year. The three topics | | | | but the summer after the retrofit was much |
| are Benchmarking, Load Factor Analysis, and | | | | hotter than the summer before the retrofit. |
| Weather Normalization as shown in Table 1. | | | | Hotter summers translate into higher air |
| BENCHMARKING | | | | conditioning loads, which typically result in higher |
| Let's suppose you were the new energy manager | | | | utility bills. |
| in charge of a portfolio of school buildings for a | | | | Hotter Summer -> Higher Air Conditioning Load |
| district. Due to a lack of resources, you cannot | | | | -> Higher Summer Utility Bills |
| devote your attention to all the schools at the | | | | In other words, the new equipment really did |
| same time. You must select a handful of schools | | | | save energy, because it was working more |
| to overhaul. To identify those schools most in | | | | efficiently than the old equipment. The figures |
| need of your attention, one of the first things | | | | don't show this because this summer was so |
| you might do is find out which schools were using | | | | much hotter than last summer. |
| too much energy. A simple comparison of Total | | | | If the weather really was the cause of the higher |
| Annual Utility Costs spent would identify those | | | | usage, then how could you ever use utility bills to |
| buildings that spend the most on energy, but not | | | | measure savings from energy efficiency projects |
| why. | | | | (especially when you can make excuses for poor |
| Benchmarking Different Categories of Buildings | | | | performance, like we just did)? Your savings |
| When benchmarking, it is also useful to only | | | | numbers would be at the mercy of the weather. |
| compare similar facilities. For example, if you | | | | Savings numbers would be of no value at all |
| looked at a school district and compared all | | | | (unless the weather was the same year after |
| buildings by $/SQFT, you might find that the | | | | year). |
| technology centers administration buildings were at | | | | Our example may appear a bit exaggerated, but |
| the top of the list, since administration buildings | | | | it begs the question: Could weather really have |
| and technology centers often have more | | | | such an impact on savings numbers? |
| computers and are more energy intensive than | | | | It can, but usually not to this extreme. The |
| elementary schools and preschools. These results | | | | summer of 2005 was the hottest summer in a |
| are expected and not necessarily useful. For this | | | | century of record-keeping in Detroit, Michigan. |
| reason, it might be wise to break your buildings | | | | There were 18 days at 90degF or above |
| into categories, and then benchmark just one | | | | compared to the usual 12 days. In addition, the |
| category at a time. | | | | average temperature in Detroit was 74.8degF |
| Different Datasets | | | | compared to the normal 71.4 degF. At first |
| You can benchmark your buildings against each | | | | thought, 3 degrees doesn't seem like all that |
| other (as we did in our example) or against | | | | much; however, if you convert the temperatures |
| publicly available databases of similar buildings in | | | | to cooling degree days, the results look dramatic. |
| your area. Energy Star's Portfolio Manager allows | | | | Just comparing the June through August period, |
| you to compare your buildings against others in | | | | there were 909 cooling degree days in 2005 as |
| your region. Perhaps those buildings in your | | | | compared to 442 cooling degree days in 2004. |
| portfolios that looked the most wasteful are still in | | | | That is more than double! Cooling degree days are |
| the top 50th percentile of all similar buildings in | | | | roughly proportional to relative building cooling |
| your area. This would be useful to know. | | | | requirements. For Detroit then, one can infer that |
| Occasionally, management decides that their | | | | an average building required (and possibly |
| organization needs to save some arbitrary | | | | consumed) more than twice the amount of |
| percentage (5%, 10%, etc.) on utility costs each | | | | energy for cooling in the summer of 2005 than |
| year. Depending upon the goal, this can be quite | | | | the summer of 2004. It is likely that in the Upper |
| challenging, if not impossible. Energy managers can | | | | Midwestern United States there were several |
| use benchmarking to guide management in setting | | | | energy managers who faced exactly this problem! |
| realistic energy management goals. For example, | | | | How is an energy manager going to show savings |
| our school district energy manager might decide | | | | from a chilled water system retrofit under these |
| to create a goal that the three most energy | | | | circumstances? A simple comparison of utility bills |
| consuming schools use only $0.80/SQFT. Since | | | | will not work, as the expected savings will get |
| this is about as much as the lowest energy | | | | buried beneath the increased cooling load. The |
| consuming schools are currently using, this could | | | | solution would be to apply the same weather data |
| be an attainable goal. | | | | to the pre- and post-retrofit bills, and then there |
| If you can find a dataset, you may also be able | | | | would be no penalty for extreme weather. This is |
| to benchmark your buildings against a set of | | | | exactly what weather normalization does. To |
| similar buildings in your area and see the range of | | | | show savings from a retrofit (or other energy |
| possibilities for your buildings. In any case, | | | | management practice), and to avoid our |
| benchmarking will focus your energy management | | | | disastrous example, an energy manager should |
| efforts and provide realistic goals for the future. | | | | normalize the utility bills for weather so that |
| Rules of Thumb | | | | changes in weather conditions will not compromise |
| New energy managers often search for a "rule of | | | | the savings numbers. |
| thumb" to use for benchmarking. An example | | | | More and more energy managers are now |
| could be: "If your building uses more than $2 | | | | normalizing their utility bills for weather because |
| SQFT/Year then you have a problem." | | | | they want to be able to prove that they are |
| Unfortunately, this won't work. Different types of | | | | actually saving energy from their energy |
| buildings have different energy intensities. | | | | management efforts. |
| Moreover, different building locations will require | | | | In many software packages, you can establish |
| differing amounts of energy for heating and | | | | the relationship between weather and usage in |
| cooling. In San Francisco, where temperatures are | | | | just one click. Because the one-click "tunings" that |
| consistently in the 60s, there is almost no cooling | | | | the software gives you are not always |
| requirement for many building types; whereas in | | | | acceptable, it does help to understand the |
| Miami, buildings will almost always require cooling. | | | | underlying theory and methodology so that you |
| Different building types, with their characteristic | | | | can identify the problem tunings and make the |
| energy intensities, different weather sites, and | | | | necessary adjustments. The more you know |
| different utility rates all combine to make it hard | | | | about the topic the better. The section that |
| to have rules of thumb for benchmarking. | | | | follows explains in a little more detail the basic |
| However, energy managers whose portfolios are | | | | elements of weather normalization. |
| all close by, can develop their own rules of thumb. | | | | How Weather Normalization Works |
| These rules will most likely not be transferable to | | | | Rather than compare last year's usage to this |
| other energy managers in different locations, with | | | | year's usage, when we use weather normalization, |
| different building types, or using different utility | | | | we compare how much energy we would have |
| configurations. | | | | used this year to how much energy we did use |
| Benchmarking Buildings in Different Locations | | | | this year. Many in our industry do not call the |
| There are some complications associated with | | | | result of this comparison, "Savings", but rather |
| benchmarking. Suppose you were the energy | | | | "Usage Avoidance" or "Cost Avoidance" (if |
| manager of a chain store, and you had buildings in | | | | comparing costs). Since we are trying to keep |
| different national locations. Then benchmarking | | | | this treatment at an introductory level, we will |
| might not be useful in the same sense. Would it | | | | simply use the word Savings. |
| be fair to compare a San Diego store to a | | | | When we tried to compare last year's usage to |
| Chicago store, when it is always the right | | | | this year's usage, we saw disastrous results. We |
| temperature outside in San Diego, and always too | | | | used the equation: |
| hot or too cold in Chicago? The Chicago store will | | | | Savings = Last year's usage - This year's usage |
| constantly be heating or cooling, while the San | | | | When we normalize for weather, we use the |
| Diego store might not have many heating or | | | | equation: |
| cooling needs. Comparing at $/SQFT might help | | | | Savings = How much energy we would have |
| decide which store locations are most expensive | | | | used this year - This year's usage |
| to operate due to high utility rates and different | | | | The next question is how to figure out how much |
| heating and cooling needs. | | | | energy we would have used this year? This is |
| Some energy analysts benchmark using kBtu | | | | where weather normalization comes in. |
| SQFT to remove the effect of utility rates | | | | First, we select a year of utility bills to which we |
| (replacing $ with kBtu). Some will take it a step | | | | want to compare future usage. This would |
| further using kBtu/SQFT/HDD to remove the | | | | typically be the year before you started your |
| effect of weather (adding HDD), but adding HDD | | | | energy efficiency program, the year before you |
| (or CDD) is not a fair measurement, as it | | | | installed a retrofit, or some year in the past that |
| assumes that all usage is associated with heating. | | | | you want to compare current usage to. In this |
| This measurement also does not take into | | | | example, we would select the year of utility data |
| account cooling (or heating) needs. Many | | | | before the installation of the chilled water system. |
| thoughtful energy managers shy away from | | | | We will call this year the Base Year . |
| benchmarking that involves CDD or HDD. | | | | Next, we calculate degree days for the Base |
| Different Benchmarking Units | | | | Year billing periods. Because this example is only |
| Another popular benchmarking method is to use | | | | concerned with cooling, we need only gather |
| kBtu/SQFT (per year), rather than $/SQFT (per | | | | Cooling Degree Days. |
| year). By using energy units rather than costs, | | | | Base Year bills and Cooling Degree Days are then |
| "rules of thumb" can be created that are not | | | | normalized by number of days. Normalizing by |
| invalidated with each rate increase. In addition, the | | | | number of days (in this case, merely, dividing by |
| varying costs of different utility rates does not | | | | number of days) removes any noise associated |
| interfere with the comparison. | | | | with different bill period lengths. This is done |
| Benchmarking Summation | | | | automatically by canned software and would need |
| Benchmarking is a simple and convenient practice | | | | to be performed by hand if other means were |
| that allows energy managers to quickly assess | | | | employed. |
| the energy performance of their buildings by | | | | To establish the relationship between usage and |
| simply comparing them against each other using a | | | | weather, we find the line that comes closest to all |
| relative (and relevant) yardstick. Buildings most in | | | | the bills. This line, the Best Fit Line, is found using |
| need of energy management practice are easily | | | | statistical regression techniques available in canned |
| singled out. Reasonable energy usage targets are | | | | utility bill tracking software and in spreadsheets. |
| easily determined for problem buildings. | | | | The next step is to ensure that the Best Fit Line |
| LOAD FACTOR ANALYSIS | | | | is good enough to use. The quality of the best fit |
| Once you have identified which buildings you want | | | | line is represented by statistical indicators, the |
| to make more efficient, you can use Load Factor | | | | most common of which, is the R2 value. The R2 |
| Analysis to concentrate your energy | | | | value represents the goodness of fit, and in |
| management focus towards reducing energy or | | | | energy engineering circles, an R2 > 0.75 is |
| reducing demand. | | | | considered an acceptable fit. Some meters have |
| What Load Factor is | | | | little or no sensitivity to weather or may have |
| Load Factor is commonly calculated by billing | | | | other unknown variables that have a greater |
| period, and is the ratio between average demand | | | | influence on usage than weather. These meters |
| and peak (or metered) demand. Average demand | | | | may have a low R2 value. You can generate R2 |
| is the average hourly draw during the billing period. | | | | values for the fit line in Excel or other canned |
| What Load Factor Means | | | | utility bill tracking software. |
| High Load Factors (greater than 0.75) represent | | | | This Best Fit Line has an equation, which we call |
| meters that have nearly constant loads. | | | | the Fit Line Equation, or in this case the Baseline |
| Equipment is likely not turned off at night and | | | | Equation. The Fit Line Equation might be: |
| peak usage (relative to off peak usage) is low. | | | | Baseline kWh = |
| Low Load Factors (less than 0.25) belong to | | | | (5 kWh/Day * #Days ) + ( 417 kWh/CDD * |
| meters that have very high peak power draws | | | | #CDD ) |
| relative to the remainder of the sample. These | | | | Once we have this equation, we are done with |
| meters could be associated with chillers or electric | | | | the regression process. |
| heating equipment that is turned off for much of | | | | Base Year bills ~= Best Fit Line = Fit Line Equation |
| the day. Low Load Factors can also be associated | | | | The Fit Line Equation represents how your facility |
| with buildings that shut off nearly all equipment | | | | used energy during the Base Year, and would |
| during non-running hours, such as elementary | | | | continue to use energy in the future (in response |
| schools. | | | | to changing weather conditions) assuming no |
| Load Factors greater than 1 are theoretically | | | | significant changes occurred in building consumption |
| impossible , but appear occasionally on utility bills. | | | | patterns. |
| Isolated instances of very high or low Load | | | | Once you have the Baseline Equation, you can |
| Factors are usually an indicator of metering errors. | | | | determine if you saved any energy. How? You |
| One school, Tyler MS, consistently has a much | | | | take a bill from some billing period after the Base |
| lower Load Factor than the others (hovering | | | | Year. You then plug in the number of days from |
| consistently around 20%). Low Load Factors can | | | | your bill and the number of Cooling Degree Days |
| be ascribed to either very high peak loads or | | | | from the billing period into your Baseline Equation. |
| very low loads during other hours. In this case, | | | | Suppose for a current month's bill, there were 30 |
| we cannot blame the Load Factor problem on | | | | days and 100 CDD associated with the billing |
| "peaky" cooling loads, as the problem exists all | | | | period. |
| year. A likely cause can be that Tyler MS is doing | | | | Baseline kWh = |
| a better job at shutting off all lighting and other | | | | ( 5 kWh/Day * #Days ) + ( 417 kWh/CDD * |
| equipment at night than the other schools. One | | | | #CDD ) |
| school (Jackson MS) typically has higher Load | | | | Baseline kWh = |
| Factors than the other schools. One reason may | | | | ( 5 kWh/Day * 30 ) + ( 417 kWh/CDD * 100 ) |
| be that lighting, HVAC and other equipment is | | | | Baseline kWh = 41,850 kWh |
| running longer hours than at Tyler MS. | | | | Remember, the Baseline Equation represents how |
| A good energy manager would investigate what | | | | your building used energy in the Base Year. So, |
| building operational behavior is contributing to the | | | | with the new inputs of number of days and |
| low Load Factor values (and consequently | | | | number of degree days, the Baseline Equation will |
| relatively high demand) for Tyler MS, and would | | | | tell you how much energy the building would have |
| investigate whether the demand could be | | | | used this year based upon Base Year usage |
| decreased. Inquiring about whether Jackson MS is | | | | patterns and this year's conditions (weather and |
| turning off equipment at night is also advisable. | | | | number of days). We call this usage that is |
| Load Factor Rules of Thumb | | | | determined by the Baseline Equation, Baseline |
| Load Factor analysis is an art, not a science. | | | | Usage. |
| Different building types (i.e. schools, offices, | | | | Now, to get a fair estimate of energy savings, |
| hospitals, etc.) will have different Load Factor | | | | we compare: |
| ranges. Since hospitals run many areas 24 hours a | | | | Savings = How much energy we would have |
| day, one might expect higher Load Factors than | | | | used this year - How much energy we did use |
| for schools, which can turn off virtually everything | | | | this year |
| at night. Also many things contribute to a | | | | Or if we change the terminology a bit: |
| particular building's Load Factor. A building left on | | | | Savings = Baseline Energy Usage - Actual Energy |
| 24 hours a day can still have a low Load Factor if | | | | Usagewhere Baseline Energy Usage is calculated |
| there are large peaks each month - for example, | | | | by the Baseline Equation, using current month's |
| a 20 bed hospital that has a scheduled MRI truck | | | | weather and number of days, and Actual Energy |
| visit once each month. The MRI demand is large, | | | | Usage is the current month's bill. |
| and can greatly impact the Load Factor of a small | | | | |