Last week I attended a thought-provoking presentation on digital twinning (DT) by the energy manager at Glasgow University, which has built digital twins for five of its buildings. It’s not a topic I know much about but I was interested because, going by what it says on the tin, it sounded like potentially a good tool for what I would call ‘discrepancy detection’ as a way of saving energy. In other words, spotting when a real building’s behaviour deviates from what it should be doing under prevailing circumstances, which will nearly always incur a penalty in excess energy consumption. The other potential benefit of DT to my mind would be the ability to try alternative control strategies on the virtual building to see if they yielded savings, and what adverse impacts there might be on service levels. This would be less intrusive than the default tactic of experimenting on live occupants.

Unfortunately I came away with the impression that we are still a way off achieving these aims. The big obstacle seems to be that DT is not dynamic – it only provides a static model. That surprised me a lot, and if any readers have evidence to the contrary, please get in touch. Another misgiving (and to be fair, the presenter was very candid about these issues) was the cost and difficulty of building and calibrating a detailed virtual model of a building and its systems. Then there is the question of all the potential influencing factors that you cannot afford to measure.

My conclusions are in two parts. One is that simulating the effect of alternative control strategies would have to be done with software short of a full DT implementation, in other words, using much-simplified dynamic block models. The other is that discrepancy detection is probably still best done with conventional monitoring-and-targeting approaches using data at the consumption-meter level, with expected consumption patterns derived empirically from historical observations rather than from theoretical models.

Good morning

FEATURED EVENT: BUILDING INSULATION

This two-hour technical briefing on 19 May covers the materials and techniques currently specified for improving the thermal performance of buildings. It includes revision of basic principles, discussion of the limitations of each type of product, and a summary of relevant UK regulations and standards. Details are at https://vesma.com/z112 and as ever your readers’ discount code is EMR2012.

Other forthcoming events are listed at http://vesma.com/training.

–o–

U IS FOR UNCERTAINTY

When we plot energy consumption against driving-factor values on a scatter diagram, the points don’t fall exactly on the regression line. The degree of dispersion is described by a parameter called the ‘coefficient of determination’, commonly known as R-squared, which tells us how much of the variation in energy consumption is explained by the regression model. When all the points fall exactly on the line, the model explains all the variation in energy consumption and R-squared has a value of one. If there is no relationship between consumption and the chosen driving factor, R-squared would be zero. If R-squared is 0.9 it means that the model explains 90% of the observed variation in energy consumption with the remaining 10% being attributable to errors or factors that were not taken into account.

There are two common misconceptions about R-squared. One is that on a heating system, a low value of R-squared signifies poor control. This is not necessarily the case, as the following thought experiment will show. Consider a well-controlled heating system whose consumption is assessed against a reliable local source of degree-day data. Whatever value of R-squared is observed, if you were to substitute degree-day statistics from a more distant weather station in the regression analysis, R-squared would go down, even though the heating system continues to be well-controlled. So beware: low R-squared might be telling you more about the quality of the model and your data than about the behaviour of the thing you are monitoring.

The other common misconception is that there is a threshold for R-squared (0.75, or 0.9, or whatever) below which your regression model cannot be trusted. There is no such cut-off. If you have chosen the most relevant driving factor and a straight-line model is plausible, you have got the right model and a low R-squared value just means it is not as reliable as it could be. In practice that simply means that a deviation has to be bigger before it can be treated as something that didn’t happen by chance. By refining the model you will improve your ability to discriminate between real faults and random variation. So it’s not a question of do you trust the model or not; the question is: “given a plausible model, how much uncertainty is there in its predictions?”. Hence the idea, introduced in an earlier bulletin, of tuneable +/- control limits on charts showing the history of deviation from expected consumption.

In the next issue: V is for verification of savings. Meanwhile if you missed any earlier issues you can catch up at http://EnManReg.org/azmt

–o–

MAVCON 21: CALL FOR PAPERS

We’ll be running our measurement and verification conference again as a series on Wednesday afternoons starting on 20 October. By popular request the first two sessions will be a refresher on basic principles and good practice, and the final sessions will be aimed at advanced practitioners but open to all. If you have an idea for a presentation (even if you don’t want to do it yourself) please let me know now.

Kind regards
Vilnis

Good morning

FEATURED EVENT

We’re hearing a lot in the news, including ambitious government announcements, about ‘decarbonising heat’. Most of the media coverage is about the domestic sector but the chances are that, as a reader, your focus is more on commercial, public-sector or industrial buildings or larger-scale residential facilities. You may have heard that experience with biomass and heat-pump installations has not always been positive, and you will want to understand the problems and pitfalls. You may also be getting questions about how hydrogen in the public supply might play out, and need ready answers.

We’ve therefore arranged a half-day intensive workshop on the practical realities of decarbonising heat in non-domestic buildings, with a team of experienced experts to talk about the lessons that have been learned and the technical and other issues that organisations face in coming decades. Details are at https://vesma.com/z200 and as ever your readers’ discount code is EMR2012.

–o–

T IS FOR TARGETS

In energy management the word ‘target’ has two distinct meanings. The first is the ‘aspirational’ target, usually set from on high without regard to practicability, to reduce consumption by x% within a certain time. It’s not a particularly smart approach. In fact in a large organisation it is almost guaranteed to fail, and here’s why. Top management sets a reduction target of x%. There being no easy, transparent and equitable way to do anything else, all departments adopt the same x% target and pass it down the chain to the lowest-tier managers. For some of them, x% is impossible to achieve so they fail. For others it will be achievable or even easy. They will save x% and then probably stop trying. Why would they over-achieve? They have other work to worry about and anyway we all know if we beat our target our managers will just give us a harder target next time. So take the roughly x% saved by the successful ones, blend this with the lower savings achieved by the first group, and you have an aggregate failure.

For me, achievability is key, and when I talk about a performance ‘target’ I mean just maintaining the best performance you can demonstrably achieve. In other words, avoid accidental excess consumption (see next article). This may not be ambitious but it is worth doing; using regression or other modelling methods supported by cusum analysis it is possible to ensure that everything has its own achievable performance characteristic. The word ‘achievable’ is crucial: it’s much more likely to get buy-in than the megaphone-management targeting that I described earlier.

I said achievable ‘characteristic’ because my concept of a target differs from common understanding in another important respect. My ‘target’ is not expressed as an annual kWh figure, nor indeed as a performance indicator, but in terms of an expected consumption quantity dynamically linked to relevant driving factors, meaning that you can track performance at whatever interval you want.

–o–

£50 FOR THE CHARITY OF YOUR CHOICE?

If you’ve ever stumbled over something that was wasting energy in a manner that was easily avoidable, I’d like to hear about it. Every month we’ll donate £50 (or the equivalent) to the charity nominated by the person whose entry is deemed the best by our guest judge. Details at http://EnManReg.org/starspot

Kind regards
Vilnis