An energy efficiency program analysis to understand the gaps between ex-ante and ex-post evaluations
Résumé
There is a gap between whole house energy consumptions calculated by engineering models and observed consumptions as reviewed in literature. This is especially true after retrofitting actions but the reasons are not well understood. Is it due to assumptions made in models, to uncertainties on input data, notably concerning behaviour, or to a no controllable random error? In order to tackle this problem, a statistical modeling approach has been developed to study deviations before and after refurbishment on a panel of refurbished dwellings in France of which annual consumptions " before " and " after " are known. In the first step, a statistical model of deviations between consumptions calculated by an engineering model and observed ones before retrofitting is established. Explanatory variables are building typology, energy systems data and household behaviour data. This model allows to identify the main sources of deviations before refurbishment and to quantify their effects. In the second step, a similar statistical model of the deviations between simulated and observed consumptions after retrofitting shows that the variables linked with the refurbishment (i.e. type of measure…) do not explain errors, excepting set point temperature change. But the observed effect (the error decreases with an increase of set point temperature after retrofitting) does not go to the logically expected direction. A modeling flaw seems be partly the cause for this result. Uncertainties of this work and prospects are also discussed.
Domaines
Sciences de l'ingénieur [physics]Origine | Fichiers produits par l'(les) auteur(s) |
---|
Loading...