A daily baseline model based on transfer functions for the verification of energy saving. A case study of the administration room at the Palacio de la Madraza, Granada

Abstract

Energy consumption in the building sector presents a high potential for reducing it by means of interventions to improve the energy efficiency of the building and/or its installations. After these interventions, it is necessary to ensure the energy impact expected by measurement and verification of savings protocols. The energy characterization of the starting situation, which takes place from the baseline, is one of the fundamental pillars of the measuring and verification of savings process. The baseline is necessary to determine the savings made after introducing an improvement measure. In this regard, researchers have provided various solutions for obtaining it. From the literature review, mathematical models stand out, which require very detailed knowledge of the building or a large amount of training data to get a good adjustment in the results. In addition, these models do not have the capacity to adapt to changes not previously considered in the knowledge base of the model. The baseline methodology proposed presents a combination of two models based on transfer functions. Both models are easily identifiable with experimental data and minimal monitoring is required to adjust them. The proposed formulation gives the models the ability to characterize the dynamic effects linked to the thermal inertia of the building and the operating conditions of the building by any variables that there may be. In addition, due to the linking of the coefficients of the models with the building's characteristic energy parameters it is possible to perform quality estimates with different input variables from those used in their calibration. The proposed methodology has been validated experimentally in a building with the use of HVAC systems, and a very variable level of occupation, with average relative daily errors of less than 10% and a monthly maximum relative error of less than 5% in the estimate of HVAC consumption. In addition, to illustrate the usefulness of the proposed methodology, its possible application has been described.