March 2018 - M&V Focus Issue #1

With the fast development of new and relatively cheap data collection and analysis tools, we hear a lot about what is called M&V 2.0 or advanced M&V. This topic motivated our search for articles for this first issue of M&V Focus. We invited Colm Gallagher to tell us more about his research on the difficulties of M&V in industrial applications and how machine learning techniques could help extract valuable knowledge contained within complex data sets collected in industrial facilities. Paul Calberg-Ellen and Eric Vorger volunteered an article which deals with recent breakthroughs in the world of energy simulation and the corresponding questions about the application of the IPMVP to the new capabilities offered by energy building simulation programs. David Jump accepted to re-post an article previously shared on his company’s blog on the concept of M&V 2.0 and normalized metered energy consumption. Many months ago, Greg Kats suggested it would be nice to tell the story of the early days of the IPMVP. The opportunity became obvious with the launch of this first issue of M&V Focus. Greg searched his notes and recollected his memory to tell us more about the early days of the IPMVP. To complement this set of core articles, we feature a practical exercise on non-routine adjustments for a real Option C case. This idea comes from Colin Grenville who originally proposed this exercise during an interactive workshop session at a conference in the U.K.


By Paul Calberg-Ellen and Eric Vorger*

This article deals with recent breakthroughs in the world of energy simulation and the corresponding questions about the application of the IPMVP to the new capabilities offered by energy building simulation programs.

Recent breakthroughs of some building energy simulation programs

In a recent article, Simon Ligier et al., 2017 explain how building energy simulation programs allow ESCO to propose a transparent and easy-understandable adjustment equation while using, at the same time, the power of energy simulation to realize risk analysis related to the uncertainties of some parameters such as building characteristics, occupants behaviour or weather variability.

By using Monte-Carlo analysis, the building energy simulation programs generate hundreds of simulated energy consumption corresponding to different combinations of the explanatory parameters, by taking into account their uncertainty. Energy simulation specialists are quite knowledgeable of these techniques. For example, commercially available solutions such as the plug-in AMAPOLA of the Building Energy Software (BES) PLEIADES developed by IZUBA and KOCLIKO, or the BES OpenStudio are implementing them.

Figure 1


But we can go even further, as described by Ligier et al. We can develop a regression model from the hundreds simulated energy consumptions to determine an adjustment equation. By using a quantile regression instead of a simple linear regression, we can assess the adjustment equation at a given level of risk. This allows the ESCO to propose a guaranteed energy consumption, here called “Guaranteed Consumption Limit at a risk α,” GCLα:

• Which takes into account the uncertainties of the influent parameters, and turn it into a quantified risk on the energy consumption;
• Which, at the same time, is accompanied by an adjustment model, such as an easy-understandable linear model, to neutralized the effects of variables whom the ESCO is not responsible, dependently of the contract’s terms, as weather or occupancy for example.


Figure 2

Consequences in M&V approach

Regarding M&V, this new approach of the determination of an adjustment model coupled with the use of energy simulation program gives rise to a question: which IPMVP option should be applied to this kind of technique? Some of the considerations summed up in the Table 2 of the IPMVP Core Concepts 2016 will guide the choice of option. According to the definition given for each option in this table, the use of an energy simulation program suggests to choose Option D. However, in the manner savings are calculated, the use of measured data associated with a routine adjustment would rather suggest selecting Option C. The table below summarizes our thoughts.


Our analysis suggests there is a need of a real mix between methodological tools developed for Option C and Option D for the following reasons. Among other things, you must ensure the proper calibration of the energy simulation program. But also, you have to ensure that the adjustment model chosen is correct, based on some indicators such as CV(RMSE) and R².

To be more specific, it seems to us that we cannot deem that it is sufficient to consider both Option C and Option D IPMVP recommendations to assess the consistency of such a new approach which couples the determination of an adjustment model and the use of an energy simulation program.

Regarding the current M&V recommendations contained in the IPMVP, we believe that the appropriate process for reviewing the consistency of both simulation and adjustment model would have to be studied globally. We also believe that a new set of recommendations should eventually be developed in the IPMVP to address the numerous questions that are already arising – and will continue to rise – with the use of the tremendous new capabilities of building energy simulation programs.

evo globe 25x25

(*) Paul-Calberg-Ellen is Deputy Director Energy at Association Régionale Biomasse Normandie in Caen area, France. He is also a member of EVO’s Extended Training Committee.

Eric Vorger is an Associate Founder with Kocliko, a French firm based in Bordeaux, France. The company offers energy optimization services for commercial, institutional and large apartment buildings.


1. Simon Ligier, Maxime Robillart, Patrick Schalbart, Bruno Peuportier. Energy Performance Contracting Methodology Based upon Simulation and Measurement. Building Simulation 2017, Aug 2017, San Francisco, United States. <>. <hal-01556848>