Managers and executives can benefit immensely from clarity about energy consumption and savings, as well as risk costly mistakes from its absence. Despite loud claims, most M&V systems create more confusion and delusion, than clarity. Here are recommendations on what to watch for in an M&V reports and software to avoid making costly investment mistakes.

Vilnis Vesma, an energy efficiency specialist from England, has discussed some faults of M&V software in a brilliant article called Weaknesses in energy monitoring and targeting software. I’d like to list several critical flaws in M&V software and M&V reports in general, ignoring which imposes significant business risk on any decision-maker.

Watch for quality of baseline models for energy efficiency

Not every formula makes sense, even if it looks scientific.

To assess quality of regression analysis, statistics has developed several tests, such as – R2, t, p, cv. These tests should be taken as seriously as the formula itself. If R2 is low, the model in nothing but a random number generator. If t and p are off the charts, the model has a high risk of generating random numbers, despite high R2.

If your statistics is a bit rusty – consult somebody who does not stand to benefit from M&V conclusions.

Do this BEFORE you make financial and operational decision based on M&V results.

Results of any calculation or model are not exact numbers

Treating numbers resulting from model as exact ones is wrong. Calculating precision is as important as calculating the approximate value. Exact numbers with 100% certainty do not exist in modeling. Outcome of any engineering calculation, including regression analysis, is always “X ±precision with Y% confidence”.

Any software that reports exact numbers of savings or saving forecasts covertly imposes a risk on decision-maker.

On the same note, presenting calculated numbers with numerous digits is plain misleading. In financial terms, savings forecast in exact kWh is as sensible as sales forecast in dollars and cents.

If last month’s sales can be calculated to a dollar, last month’s energy saving can’t be calculated to a kWh or to a m3.

Model of energy consumption must reflect physics and business

Calculation of savings in accordance with best practices should be done on the basis of regression analysis, which results in a formula that models energy consumption as it relates to consumption drivers, such as production volume, temperature, etc.

Regression model must reflect actual physics of the process it models, or else it is useless.

For example:

  • model of AC system consumption should have a negative coefficient to temperature, because higher temperature causes higher AC consumption
  • model of plant consumption should have positive coefficients to products, because when more products is produced – more energy is consumed
  • most energy consumption models must be linear, unless you know that physics is not linear

Use of fancy polynomial models to increase R2 puts decision-maker at risk of wasting money.

If you found this post useful, you may find our posts on adjacent topics useful too:

Which has to come first: Energy metering or Energy Audit?

More energy data may not benefit business – and what does