For this reason, back in September, we introduced the concept of a ‘base range’ to our daily Rapid Efficiency Feedback (REF). Unlike a ‘baseline’, which shows an unrealistically precise energy profile from analysing historical usage patterns during a ‘training’ period prior to the introduction of daily feedback, a base range shows the range of expected profiles. We’re not the first to introduce this concept (‘confidence intervals’ are a mainstay of statistics) but our base range is considered a milestone achievement because it highlights the consistency of past performance within and between days and can therefore draw attention to what might be considered a ‘significant’ slip or improvement – precisely and to the 15-minute interval.
In October we launched another significant advance – the ‘potential’ profile, also known as ‘best in class’. We calculate a building’s potential much like we calculate base ranges and ‘best ever’ (launched in May last year and featured in a behaviour change study we presented at the ACEEE conference in August), by using machine-learning technology to analyse past patterns of performance. Our judgement about a building’s potential efficiency for any given conditions takes into account patterns we’re able to observe from comparable buildings under comparable conditions in other places, sometimes on the other side of the world.
A ‘best ever’ profile, on the other hand, only looks at the building itself and focuses on when it was operating at its most optimised under the conditions.
Best ever and potential profiles are calculated by looking at all performances up until the present, not just an historical training period. This is because we want to reflect the most relevant information and for a building where performance is continually improving, that means the most recent information is most important.
Given all the complications and complexity involved with running a building and recognising that no two days are ever exactly the same, we have just recently introduced a ‘target range’ that conveys our level of uncertainty about what would constitute ‘best ever’ performance during any given 15-minute period of a day. This is illustrated in the plot which we’ve extracted from one of the daily REF messages.
There are a few things worth noting:
- The target range is much tighter than the base range. This is because, in this case at least, the building has become more consistently operated – much more consistently than in the past (during the pre-commencement training period) when its performance varied more widely and not necessarily in proportion to the work needing to be done.
- The target range only appears above the ‘best ever’ performance (i.e. only one side is shaded). Best ever represents the improvement limit we’ve been able to observe whereas the red-shaded target range highlights the exceptional performance zone.
- The target range is below the base range, but not always. In this example we can tell the building has been on an improvement trajectory for a while because REF knows that performance significantly below the base range is achievable. The green and red shaded areas will start out closer together and not spread apart until savings have been demonstrated under the conditions.
- The vertical grey-shaded area ‘beat best ever’ highlights the times during the day when performance was significantly better than anything ever observed before.
Just as an athlete can only hope to achieve his / her ‘best ever’ every once in a while, the ‘best ever’ should be thought of as something to aspire to, not something to consistently beat – after all it’s a moving target! The target range is different and hopefully quite helpful. It shows when a building is in ‘the zone’, and that level of performance should be considered a realistic goal for every building under all conditions.
Do you think a target range simplifies or complicates our messaging around building optimisation, and why? We’d love to hear from you!