• Energy Research

Abstract The goal of optimal temperature control in buildings is usually to ensure thermal comfort with minimal energy consumption. In intermittently occupied buildings, this presumes the ability of the controller to recover in due time the building from setback. Model Predictive Control (MPC) is considered among the best candidates for this task due to its ability to use occupancy schedule and weather forecasts for optimal temperature control. However, the use of the classical cost function within MPC does not allow to achieve the objectives of minimal energy consumption and optimal restart of the heating system. Therefore, a new cost function is introduced, which minimizes the energy consumption while maintaining the thermal comfort in the building. The obtained linear optimization problem is formulated to fit into the canonical form of Linear Programming method.

Experimental identification of the dynamic models of heat transfer in walls is needed for optimal control and characterization of building energy performance. These models use the heat equation in time domain which can be put in matrix form and then, through state-space representation, transformed in a transfer function which is of infinite order. However, the model acts as a low-pass filter and needs to respond only to the frequency spectrum present in the measured inputs. Then, the order of the transfer function can be determined by using the frequency spectrum of the measured inputs and the accuracy of the sensors. The main idea is that from two models of different orders, the one with a lower order can be used in building parameter identification, when the difference between the outputs is negligible or lower than the output measurement error. A homogeneous light wall is used as an example for a detailed study and examples of homogeneous building elements with very high and very low time constants are given. The first order model is compared with a very high order model (hundreds of states) which can be considered almost continuous in space.

Autoregressive models with exogenous (ARX) are widely used for parameters identification of walls although their parameters do not have a direct physical interpretation. This paper uses measured data and an ARX model, obtained by deduction from a thermal network, for identifying the parameters with physical meaning of a wall tested under real weather conditions. The physical parameters are the U-value, the dynamic solar energy transmittance and the effective heat capacity. The paper presents the whole chain of transformations from thermal networks to ARX models for demonstrating that the procedure used for recovering the physical parameters guarantees their unicity. The estimations of the physical parameters are presented together with their uncertainties.

Abstract Non-intrusive loading monitoring (NILM) provides a smart solution to the problem of electrical energy monitoring of households at the appliance level. In blind disaggregation, the power level of each appliance is not known a priori. In this paper, we propose an event-based blind disaggregation algorithm that uses Gaussian mixture models (GMM) for clustering to automatically detect two-state appliances from the aggregate data. The benefit of using Gaussian mixture models over other clustering methods is that they can automatically learn the statistical distributions present in the data. This is beneficial, especially when the appliances have similar power consumptions. Since Gaussian mixture models do not determine the number of clusters automatically, we use Bayesian information criteria (BIC) to determine the number of clusters. The blind disaggregation method is tested with data from a real house collected by a smart meter which samples the aggregate consumption at 3.4 kHz and also from Reference Energy Disaggregation Dataset (REDD) public data, sampled at a frequency of 1 Hz. We compared the performance of our algorithm with other unsupervised methods and found comparable performance. We also compared the Gaussian mixture model with mean shift clustering in a blind disaggregation. We saw an improved performance by using Gaussian mixture models instead of mean-shift clustering in various accuracy measures.

Quick U-building (QUB) is a method for short time measurement of energy performance of buildings, typically one night. It uses the indoor air temperature response to power delivered to the indoor air by electric heaters. This paper introduces a method for estimating the expected measurement error as a function of the amplitude and the time duration of the input signal based on the decomposition of the time response of a state-space model into a sum of exponentials by using the eigenvalues of the state matrix. It is shown that the buildings have a group of dominant time constants, which gives an exponential response, and many very short and very large time constants, which have a small influence on the response. The analysis of the eigenvalues demonstrates that the QUB experiment may be done in a rather short time as compared with the largest time constant of the building.

The quick U-building (QUB) method is used to measure the overall heat loss coefficient of buildings during one to two nights by applying heating power and by measuring the indoor and the outdoor temperatures. In this paper, the numerical model of a real house, previously validated on experimental data, is used to conduct several numerical QUB experiments. The results show that, to some extent, the accuracy of QUB method depends on the boundary conditions (solar radiation), initial conditions (initial power and temperature distribution in the walls) and on the design of QUB experiment (heating power and duration). QUB method shows robustness to variation in the value of the overall heat loss coefficient for which the experiment was designed and in the variation of optimum power for the QUB experiments. The variations in the QUB method results are smaller on cloudy than on sunny days, the error being reduced from about 10% to about 7%. A correction is proposed for the solar radiation absorbed by the wall that contributes to the evolution of air temperature during the heating phase.

State-space representation is essential in the theory of dynamic systems. This paper introduces a methodology for obtaining state-space representation from the thermal models of elementary components of a building by the conjunction of two methods: 1) assembling of thermal circuits and 2) state-space extraction from thermal circuit. These methods are fully illustrated on a very simple model and tested on a real house of about 100 m 2 on which detailed measurements were achieved for 40 days at a time step of 10 min. The errors obtained between the measurements and the simulation results are in the order of $\pm$1 {\textdegree}C for a single zone and $\pm$2 {\textdegree}C for seven thermal zones. Besides simulation, parameter identification and control, the methods for assembling thermal circuits and extraction of state-space representation may be useful in Building Information Modelling (BIM).

Abstract Dynamic models are needed for automatic control and online fault detection and diagnosis of desiccant wheels used in solar cooling systems. The parameters of these models need to be experimentally identified during operation. A model based on the heat and mass transfer equations, given in state-space representation, is developed. Using this model, a gray-box approach is developed to identify experimentally the thermal transfer coefficients. The Nusselt number and the heat and mass transfer coefficients have been obtained by defining the unknown parameters of the model as a function of a single parameter, which itself depends on the heat transfer coefficient, and by expressing the other parameters as a function of the geometrical and physical properties of the wheel. The difference between the values of the thermal transfer coefficients given in literature and those obtained by this method is less than 10%. Therefore, the gray-box identification method may be adopted as a tool for experimental estimation of the thermal coefficients under various design and operation conditions for the desiccant wheels.