From in-situ measurement to regression and time series models

Abstract

Data analysis methodologies are crucial to learn insights from data and to create more trust in the assumptions used for energy performance assessment. Indeed, continuous performance monitoring should become a more diffuse practice in order to improve our design and operation strategies for the future. This is an essential step to reduce incrementally the gap between simulated and measured performance. In fact, assumptions in simulation represent a significant source of uncertainty when estimating the energy performance of buildings. This uncertainty affects decision-making processes in multiple ways, from design of new and refurbished buildings to policy making. The research presented aims to highlight potential links between experimental approaches for test-facilities and methods and tools used for continuous performance monitoring, at the state of the art. In particular, we start by exploring the relation between in-situ measurement of thermal transmittance (U) and regression-based monitoring approaches, such as co-heating test and energy signature, for heat load coefficient (HLC) and solar aperture (gA) estimation. After that, we highlight some recent developments in simplified dynamic energy modelling using lumped parameter models. In particular, we want to underline the scalability of these techniques, considering relevant issues in current integrated engineer design perspective. These issues include, among others, the necessity of limiting the number of a sensors to be installed in buildings, the possibility of employing both experimental and real operation data (and compare them with design data as well) and, finally, the possibility to automate performance monitoring at multiple scales, from single components, to individual buildings, to building stock and cities.