Moderator: Dominique Osso, EDF R&D
“In the beginning was the data”. This session will focus on data and how they can be exploited at their full potential. The three papers presented in this session will provide a deep insight into the use of data.
First, Daniel Cabrera et al. present lessons learned in Switzerland on efficient lighting programs. They show how comparison of ex-ante and ex-post data helps to better understand the ex‑ante estimates. They find that studying the gap between ex-ante and ex-post energy savings estimations for specific programs should not be enough to validate ex-ante estimations. The fact that savings estimations match with the actual savings doesn’t mean that the assumptions taken for the calculations are correct.
These findings can be exploited to elaborate some criteria that can be used to decide under what conditions ex-ante estimates can be used (i.e. only if each one and all of the independent variables have an acceptable accuracy) and improve the cost efficiency of the impact evaluation without scarifying accuracy.
Next, Filip Milojkovic examine a large datasets of more than 80,000 German users that monitors their energy consumption of their household via a dedicated website. User’s account shows how much energy is consumed at a glance whether it is heating, electricity, water but the focus in this paper is on space heating. It is a first attempt to use these user-generated data for regression modelling of energy savings.
The analysis of monthly heat energy consumption provide assessment of the influence of refurbishment activity on energy consumption. Results allow the robust conclusion that the insulation of roof, wall and basement lead to statistically significant and relevant energy savings on average. At the opposite, for window replacement the results are mixed and inconclusive.
The final paper from Emily Cross proposes a constructive path forward toward a future energy efficiency impact evaluation paradigm in the era of “Big Data” (i.e. advanced real-time data analytics). The paper explains the requirements for leveraging high-frequency data for an automated analysis of energy savings, and how the IPMVP could be expanded to explicitly include high-frequency data. The author also supports measurement of first year impact savings focusing only on whole building savings for projects above a signal-to-noise ratio in the range of five to ten percent of annual pre-installation consumption.
The paper concludes that accurate, direct, census measurement and evaluation of entire populations of energy consumption is consistent with the broader industry goals relative to climate change, emerging ‘white certificate’ policies, and can be achieved at lower cost than current impact evaluation methods by using advanced accelerated analytics, with relatively modest changes to data tracked by programmes.
PAPERS / PRESENTATIONS
When Can We Trust Deemed Savings? [paper]
Daniel Cabrera, University of Geneva
Jean-Luc Bertholet, University of Geneva
Bernard Lachal, University of Geneva