A biennial international conference on building simulation would have been a brilliant excuse to visit San Francisco if I hadn't already been planning to spend August and September there anyway. As it was, it was a great opportunity to reflect on my research and talk to some interesting people.
Although Energy Performance Contracting wasn't a conference theme, modelling energy consumption is fundamental to understanding the savings that are likely to arise from an EnPC. However, like any tool, understanding its limitations is also vital. Much has been written about the potential for modelling to lend a spurious air of accuracy to predictions and the need for caution in interpreting the results. Sensitivity analysis has a big part to play in testing and understanding models and their results and it was really encouraging to see how much the field has moved on in recent years. I saw only a handful of papers using old fashioned (and inappropriate) one-at-a-time sensitivity analysis. Most employed global sensitivity analysis techniques and not just as "proof of concept" but as a mainstream tool. The need to make sure that you cover all possible combinations of inputs when you do your sensitivity analysis is a real hobby horse of mine (as is the fact that it's not much more difficult than the one-at-a-time approach - have a look at Andrea Saltelli's website if you'd like to know more) so that was really encouraging.
My paper was catchily titled "Exploring the commercial implications of measurement and verification choices in energy performance contracting using stochastic building simulation". The aim was to use simulation to try to encompass the uncertainty that exists when you undertake an energy efficiency retrofit in an existing building when time and resources for detailed surveys are sadly lacking. Underlying this was the question of the approach to measuring and verifying savings - how much difference does it make which option you choose? The answer, as always, is complicated, deeming savings reduces risks for the ESCO, important when margins are tight but leaves some clients with expensively procured guarantees which offer no real protection.
Although there were only a handful of papers directly applying simulation to EnPC, there was some really interesting work, particularly from Simon Ligier with a methodology for EnPC using simulation and global sensitivity analysis to derive estimates for energy savings which take account of the uncertainties involved.
The great thing about such a large conference though was the opportunity to get a broader perspective and a flavour for what else is going on. The discussions around calibration were illuminating, while techniques are advancing all the time, the challenges of accommodating the full range of uncertainties remain.
Another fascinating discussion was Luisa Scambia's presentation on the differences between the way that students and more experienced professionals approach building simulation. Some real food for thought in how the two have different priorities in their modelling and what the implications of that might be.
