Every so often if you work in energy management you hear that defeatist whine - you can't manage what you can't measure.
Sure I get it, measuring things is good ! However, sometimes a manager has to deal with missing information. In this post I am going to look at a very common problem with energy management - the missing meter reading. Look it happens, so let's get used to it already!
The classic report of a string of meter readings is shown showing cumulative true readings
reported in light blue and then in purple reported values. During our example, twice the reading technology fails. At the end of the missing period a sudden spike of apparent consumption occurs as our reported reading suddenly catches up with our true reading.
The question I am asking is what are we to do with the data?
- Bury head in sand - This is the easy and most common path. The automatic reading company (for that is all they are "smart" is a misnomer) says, aha, but you must have data that corresponds to fiscal metering (the money), otherwise it will cause confusion - so we report what we know !
- Linear interpolation - You can't manage what you cant measure (there it is) so lets just join the dots that represent what we do know with a straight line. Essentially this assumes a period of constant average consumption.
- The "Smart fix" - Looking at the lines above, we can see a pattern of consumption, when we do get readings, so we can fill in the patterns on the basis of the information that we do have
The correct action, naturally is to remember the source data (the reconciliation is necessary), and delete all the unknown data ! - If we do not know a pattern, to assume it is to suggest that metering at this resolution is unnecessary, because we know what is happening anyway.
This is the difference between financial and management accounting, and an energy manager accounts for energy - but must also manage it, If he needs reconciliation, he needs it at billing intervals (not at half-hourly intervals) so yes the big picture is to take the diffenece between two known readings as sacred. However, if we care about knowledge regarding distribution of load, we are working with statistics. A missing data item is not a problem (and it is far better than an observation that re-inforces a pattern that may be wrong - to do so adds no value (and can confuse) - The temptaton to make data pretty must be avoided. Report and work with what you know - do not give yourself the artifical comfort of looking at fraudulently complete data, and thereby innure yourself to the truth. Best solution, use the evidence to make your meter data provider work harder and then get on with the job - Managing with the measurements you have, not excusing yourself on the basis of those you don't have !
