Thermal Load Prediction Research Articles

A Nonlinear Autoregressive With Exogenous Inputs Artificial Neural Network Model for Building Thermal Load Prediction

Abstract Thermal load prediction is a key part of energy system management and control in buildings, and its accuracy plays a critical role to improve building energy performance and efficiency. Regarding thermal load prediction, various types of prediction model have been considered and studied, such as physics-based, statistical, and machine learning models. Physical models can be accurate but require extended lead time for model development. Statistical models are relatively simple to develop and require less computation time, but they may not provide accurate results for complex energy systems with intricate nonlinear dynamic behaviors. This study proposes an artificial neural network (ANN) model, one of the prevalent machine learning methods to predict building thermal load, combining with the concept of nonlinear autoregressive with exogenous inputs (NARX). NARX-ANN prediction model is distinguished from typical ANN models because the NARX concept can address nonlinear system behaviors effectively based on its recurrent architectures and time indexing features. To examine the suitability and validity of NARX-ANN model for building thermal load prediction, a case study is carried out using the field data of an academic campus building at Mississippi State University (MSU). Results show that the proposed NARX-ANN model can provide an accurate and robust prediction performance and effectively address nonlinear system behaviors in the prediction.

Failure analysis of reinforced concrete walls of cylindrical silos under elevated temperatures

Lessons learned from failures are an important input for improvement of structural design and consequently, for advancement of standards. As an integral part of the industrial infrastructure, silos are exposed to high operation and environmental loads. As a result, silos suffered from serious damage in the recent years. One of the major factors affecting serviceability of silos is related to temperature-induced cracking. However, prediction of an adequate thermal loading may be a rather complex task and the significance of thermal cracking itself is not quite emphasized in the current guidelines for the design of silos. As a result, engineering practice may omit or underestimate this effect. The authors aim to present the significance of thermal loading in silos under elevated temperatures and its influence on the design of structurally necessary reinforcement. This is performed with summarizing background information for determining thermal loads based on the literature overview and further, with examining thermal load actions in a case study of a concrete silo for storing bulk cement by means of anumerical analysis.

Gradient boosting machine for predicting return temperature of district heating system: A case study for residential buildings in Tianjin

Accurate prediction of the return temperature is critical to energy efficiency of the district heating system (DHS). The support vector machines (SVMs) and artificial neural networks (ANNs) have been widely applied to forecast energy consumption of the DHS recently. However, the parameters of SVM and ANN are difficult to be optimized due to their specific request for inputs and time-consuming training process. To explore the performance of the decision tree-based ensemble algorithms in the return temperature prediction task of the DHS, four return temperature prediction models based on the operational data of a heating system in Tianjin are established, namely Support Vector Regression (SVR), Multilayer perceptron (MLP), Random Forest (RF) and light gradient boosting machine (LGBM). The RF and LGBM are two typical decision tree-based ensemble algorithms. The historical supply temperature, outdoor temperature, relative humidity, wind speed and air quality index (AQI) are used as original inputs of the models. For computational models (SVR, MLP), the input features should be further transformed. The experimental results demonstrate that the LGBM model outperforms others in all standard evaluation measures. It shows that the tree-based ensemble models without complicated feature transformation achieves considerable results. Moreover, a week-based time series data splitting strategy is developed and compared with the traditional method. The experimental results show that the novel method can improve the performance of models except MLP. Overall, the performance of tree-based ensemble algorithms (RF and LGBM), SVR, and MLP are compared based on a case study in this article, illustrating the potential of the tree-based ensemble algorithms in thermal load prediction.

Cabin and Battery Thermal Management of Connected and Automated HEVs for Improved Energy Efficiency Using Hierarchical Model Predictive Control

Incorporating traffic information in power management optimization process for electrified and connected vehicles offers opportunities for improving fuel economy. Integrating the management of thermal load (such as those used for heating, ventilation, & air conditioning (HVAC) of the passenger compartment, and for the battery cooling) with the power management process can provide even greater benefits for connected and automated vehicles (CAVs). However, given the relatively slow dynamics associated with the thermal subsystems, the lack of reliable power and thermal loads prediction over an extended prediction horizon is the main challenge for efficient thermal management using model predictive control (MPC). This paper presents a hierarchical two-layer MPC scheme which exploits vehicle speed and traffic preview predictions over short and long prediction horizons to schedule optimal thermal trajectories for the cabin and battery cooling in hybrid electric vehicles (HEVs) via a novel intelligent online constraint handling (IOCH) approach. These trajectories are next incorporated into the vehicle-level controller to determine the proper power split between electric motor and internal combustion engine (ICE). We present the development and experimental validation of control-oriented models used for prediction of the vehicle thermal dynamics and loads over a long planning horizon. Compared to a more traditional single-layer MPC approach, the proposed two-layer MPC shows that depending on the driving cycle and traffic conditions, 2.2% to 5.3% reductions in HEV fuel consumption can be achieved for urban driving and congested city driving cycles, respectively, in CAV operation scenario. This fuel economy improvement is a direct result of taking proactive actions through real-time prediction and optimization to avoid conservative and inefficient thermal responses, while enforcing cabin and battery operating constraints.

Model-Based Occupant Tracking Using Slab-Vibration Measurements

Sensor-based occupant tracking has the potential to enhance knowledge of the utilization of buildings. Occupancy-tracking strategies using footstep-induced floor vibrations may be beneficial for thermal-load prediction, security enhancement and care-giving without undermining privacy. Current floor-vibration-based occupant-tracking methodologies are based on data-driven techniques that do not include a physics-based model of the structural behavior of the floor slab. These techniques suffer from ambiguous interpretations when signals are affected by complex configurations of structural and non-structural elements such as beams and walls. Using a physics-based model for data-interpretation enables deployment of sparse number of sensors in contexts of non-uniform structural configurations. In this paper, an application of physics-based data interpretation using error-domain model falsification (EDMF) is presented to track an occupant within an office environment through footstep-induced floor vibrations. EDMF is a population-based approach that incorporates various sources of uncertainty, including bias, arising from measurements and modeling. EDMF involves the rejection of simulated model responses that contradict footstep-induced floor vibration measurements. Thus, EDMF provides a set of candidate locations from an initial population of possible occupant locations. A sequential analysis that accommodates information from previous footsteps is then used to enhance candidate locations and identify trajectories among candidates. In this way, incorporating structural behavior in interpreting vibration measurements induced by occupant footsteps has the potential to identify accurately the trajectory of an occupant in buildings with complex configurations, thereby providing tracking information without undermining privacy.

Thermal load prediction in district heating systems

Optimal operation of district heating (DH) systems usually relies on the forecast of thermal demand profiles of the connected buildings. Depending on the purpose of the analysis, thermal request can be required at various levels, from building level to thermal plant level. In the case of demand response for example, thermal request is necessary at a building level to evaluate its applicability and at a plant level to determine the effects. Thermal request profiles are quite different, depending on the observation point. Total requests are not just the summation of the downstream requests, mainly because of the thermal transients. The heat losses also contributes to modify the curves, although generally in a smaller way. In this work, a multi-level thermal request prediction is proposed. This approach has the aim of evaluating the thermal request in the various sections of DH network with reduced computational resources. This includes a compact model for the prediction of building demand and a network model in order to compose together the requests at the various levels. The application to a portion of the Turin district heating network is proposed. This shows that the network dynamics significantly affects the evolution, especially at peak load.

Real time optimal control of district cooling system with thermal energy storage using neural networks

Thermal energy storage can be utilized as an effective component in energy systems to maximize cost savings when time-of-use (TOU) pricing or real-time pricing (RTP) is in place. This study proposes a novel approach that can effectively predict performance and determine control strategy of thermal energy storage (i.e., ice storage) in a district cooling system. The proposed approach utilizes Neural Network (NN) based model predictive control (MPC) strategy coupled with a genetic algorithm (GA) optimizer and examines the effectiveness of using a NN model for a district cooling system with ice storage. The NN offers a relatively fast performance estimation of a district cooling system with given external inputs. To simulate the proposed MPC controller, a physics-based model of the district cooling system is first developed and validated to act as a virtual plant for the controller to communicate system states in real times. Next, the NN modeling the plant is developed and trained during a cooling period so that the control strategy is tested under the RTP and TOU pricing. This model is optimized using the GA due to the on/off controls for the district cooling network. Finally, a thermal load prediction algorithm is integrated to test under perfect weather inputs and weather forecasts by considering 1-hour discretization in the MPC scheme. Results indicate that for the month of August, the optimal control scheme can effectively adapt to varying loads and varying prices to effectively reduce operating costs of the district cooling network by approximately 16% and 13% under the TOU pricing and the RTP, respectively.

Towards 4th generation district heating: Prediction of building thermal load for optimal management

One of the requirements for the transition from conventional district heating (DH) systems to 4th generation DH (4GDH) systems is the knowledge of system dynamics. Forecast of thermal request profiles of buildings is crucial to optimize the operating conditions. In fact, when this is available, the thermal load evolution at the plants can be estimated and proper energy saving actions can be implemented. In this paper, a smart and fast approach for estimating the daily thermal request of buildings in large networks is presented. The methodology uses only data available from the smart meters installed in the building substations (mass flow and temperature data). For the users where smart meter data are not available an alternative approach is proposed. The methodology is shown to be suitable for applications involving a) a very large number of buildings b) necessity of forecast of an area c) measured data, which might be affected by gaps d) low computational time requirements. Experimental data show that, despite the simplicity, the method predicts the thermal request very accurately. Furthermore, the forecasted thermal request are here effectively used with the aim of reducing the peak load in one of the largest DH systems in Europe.

A tuning methodology of Model Predictive Control design for energy efficient building thermal control

The Model Predictive Control (MPC) approach is based on the prediction of indoor and outdoor thermal loads in order to counter the deviation of the indoor temperature from the occupants' preference in advance. The minimization of temperature alteration allows for efficient energy use and improvement of indoor thermal environment. Unlike the classical reactive control, MPC is able to act in advance and to explicitly handle constraints on building variables when selecting the best scenario. This peculiarity makes MPC particularly suitable for the optimal energy management of buildings but, despite the increasing research activity in the field, the commercial technologies are lagging behind. Among the reasons are the lack of a well established design approach easily applicable to the buildings domain and the lack of a design optimization. In this paper a tuning methodology of MPC design for energy efficient building thermal control is presented. The tuning is performed on the controller parameters, and aims at identifying the best parameter set in terms of energy saving and temperature deviation from the chosen setpoint. To demonstrate the effectiveness of the methodology, a simulation based analysis using a model estimated on a real case study is presented. The methodology shows that by improving the control parameters it is possible to reduce energy consumption and improve thermal comfort for the final user.

Fast Conjugate Heat Transfer Simulation of Long Transient Flexible Operations Using Adaptive Time Stepping

The emerging renewable energy market calls for more advanced prediction tools for turbine transient operations in fast startup/shutdown cycles. Reliable numerical analysis of such transient cycles is complicated by the disparity in time scales of the thermal responses in fluid and solid domains. Obtaining fully coupled time-accurate unsteady conjugate heat transfer (CHT) results under these conditions would require to march in both domains using the time-step dictated by the fluid domain: typically, several orders of magnitude smaller than the one required by the solid. This requirement has strong impact on the computational cost of the simulation as well as being potentially detrimental to the accuracy of the solution due to accumulation of round-off errors in the solid. A novel loosely coupled CHT methodology has been recently proposed, and successfully applied to both natural and forced convection cases that remove these requirements through a source-term based modeling (STM) approach of the physical time derivative terms in the relevant equations. The method has been shown to be numerically stable for very large time steps with adequate accuracy. The present effort is aimed at further exploiting the potential of the methodology through a new adaptive time stepping approach. The proposed method allows for automatic time-step adjustment based on estimating the magnitude of the truncation error of the time discretization. The developed automatic time stepping strategy is applied to natural convection cases under long (2000 s) transients: relevant to the prediction of turbine thermal loads during fast startups/shutdowns. The results of the method are compared with fully coupled unsteady simulations showing comparable accuracy with a significant reduction of the computational costs.

Development of Building Thermal Load and Discomfort Degree Hour Prediction Models Using Data Mining Approaches

Thermal load and indoor comfort level are two important building performance indicators, rapid predictions of which can help significantly reduce the computation time during design optimization. In this paper, a three-step approach is used to develop and evaluate prediction models. Firstly, the Latin Hypercube Sampling Method (LHSM) is used to generate a representative 19-dimensional design database and DesignBuilder is then used to obtain the thermal load and discomfort degree hours through simulation. Secondly, samples from the database are used to develop and validate seven prediction models, using data mining approaches including multilinear regression (MLR), chi-square automatic interaction detector (CHAID), exhaustive CHAID (ECHAID), back-propagation neural network (BPNN), radial basis function network (RBFN), classification and regression trees (CART), and support vector machines (SVM). It is found that the MLR and BPNN models outperform the others in the prediction of thermal load with average absolute error of less than 1.19%, and the BPNN model is the best at predicting discomfort degree hour with 0.62% average absolute error. Finally, two hybrid models—MLR (MLR + BPNN) and MLR-BPNN—are developed. The MLR-BPNN models are found to be the best prediction models, with average absolute error of 0.82% in thermal load and 0.59% in discomfort degree hour.

Comparison of Wall Treatments and Meshes in Large-Eddy Simulations of Mixing Tees

Large-eddy simulations (LESs) for two different T-junctions are performed for the prediction of thermal mixing loads on piping. In particular, the effects of wall treatment and mesh on temperature and wall heat flux fluctuations are studied. Wall-resolved LES shows good agreement with an experiment having adiabatic walls, but using wall functions shows deviations in root-mean-squared (RMS) temperatures and cross-stream mean velocities. The simulations show increases in peak RMS temperatures with local mesh refinement, and hence, too-low peak values are obtained with wall functions. The highest temperature fluctuations occur locally near the T-junction requiring a dense mesh. Wall functions are unable to capture high wall heat fluxes at a sharp corner, but otherwise, the maximum RMS value is close to a wall-resolved LES. For a T-junction having a round corner, higher RMS heat flux is obtained with wall functions compared to a wall-resolved case. Wall functions show lower instantaneous heat fluxes than wall-resolved LES, but the wall functions nonetheless result in higher pipe wall temperature fluctuations due to lower frequency content.

Understanding how arc attachment behaviour influences the prediction of composite specimen thermal loading during an artificial lightning strike test

High speed image analysis of experimental lightning strike testing on composite specimens has demonstrated complex time-dependent arc attachment behaviour. To date simulation studies of lightning strike direct effects on composite materials have overlooked or idealised the observed arc attachment behaviour. This is significant as these preceding studies have also demonstrated joule (resistive) heating is a major contributor to composite material damage and thus the application of the current load is critically important. The objective of this paper is to quantify how arc attachment behaviour influences the prediction of composite specimen thermal loading during a simulated lightning strike test. This is achieved through a simulation study modelling experimental arc attachment behaviours referenced in the literature. The simulation results are compared with measured test specimen damage. The results quantify for the first time the significance of arc attachment behaviour, demonstrating how realistic representation of arc expansion and arc movement can alter the predicted thermal loading and generate thermal damage patterns comparable to measured experimental damage.

The re-optimization strategy of multi-layer hybrid building’s cooling and heating load soft sensing technology research based on temperature interval and hierarchical modeling techniques

To avoid expanding existing fossil power plants, building the energy performance management system is an effect method to cope with a high growth demand of electricity in China's urbanization process. The accuracy thermal load prediction is the first step to build an energy performance management system which has important significance for energy saving, decreasing air pollution and high-efficiency operation of HVAC system.This paper discusses the use of hybrid intelligent approaches and re-optimization strategy, a multi-layer hybrid model (APNN) has been proposed by hybridizing an auto-regressive model with exogenous inputs (ARX) and a particle swarm optimization neural network (PSO-NN) to make a good use of the comprehensive information of the meteorological data and historical data. For the purposes of improving prediction precision generalization ability of the multi-layer hybrid model, temperature interval and hierarchical modeling techniques were used. According to the re-optimization strategy, there are two improvements of the previous proposed APNN model, which are based on temperature interval and hierarchical modeling by solar radiation intensity. Compared with the basic prediction models, validation results show that accuracy of the optimized models are greatly improved. What’s more, the optimization of multi-layer hybrid building’s cooling and heating load soft sensing technology enhance learning and generalization capability of the basic APNN model.

Linear regression models for prediction of annual heating and cooling demand in representative Australian residential dwellings

Abstract: This paper presents the development methodology of linear regression models that were developed for the prediction of annual thermal loads in representative residential buildings across three major climates in New South Wales, Australia, and the assessment of the impact of building envelope upgrades. A differential sensitivity analysis was undertaken for sixteen building envelope parameters, with six parameters being identified as significant. These six parameters were then explored using EnergyPlus simulation, and a number of linear regression models developed from the simulation outputs. Random values for design parameters were generated, and the results of EnergyPlus simulations using these parameters were used to verify the outputs of the regression models. The differences between regression-predicted and EnergyPlus-simulated annual thermal energy requirements were of order 10%-15%. The coefficient of determination (R2) was over 0.90, indicating a good agreement between simulation and the regression models, and suggesting that the annual heating and cooling energy requirements can be forecasted with an acceptable accuracy using the regression models. It is envisaged that the regression models developed can be used as a quick alternative to building simulation for residential buildings of the area and the climate covered by our study, and can serve to rapidly estimate the likely energy savings/penalty during the retrofitting design stage when different building schemes and design concepts are being considered.

Comprehensive approach to modeling and simulation of dynamic soft-sensing design for real-time building energy consumption

This research develops a multi-layer hybrid soft-sensor model to improve the accuracy of building thermal load prediction using integrated data. The multi-layer hybrid model (autoregressive and par...

Field validation study of a time and temperature indexed autoregressive with exogenous (ARX) model for building thermal load prediction

Building load prediction algorithms are becoming an essential component of building energy technologies as intelligent building technologies are rapidly evolving and require accurate load predictions to make real-time operational decisions. This paper presents a field validation study of an autoregressive with exogenous (ARX) model, indexed with respect to time and temperature, and used for hourly building thermal load prediction with an aim for integration with real time predictive control strategies. Indexing of the ARX model implies that different sets of coefficients are used in the predictive equation depending on different time intervals and temperature ranges. Although many regressive prediction models have been proposed, no field validation has been reported in the literature, which is an essential step before implementation in actual practice. The validation study was carried out using field data from three buildings located in the main campus of Mississippi State University. The proposed model was able to predict hourly thermal load accurately and within the uncertainty bounds of the measured thermal load most of the time. Results also demonstrated that proper indexing of the model allowed it to capture different cooling and heating load profiles and abrupt changes in the load pattern.

Thermal Load Prediction Considering Solar Radiation and Weather

Abstract This study is aimed at presenting solar radiation and weather dependent thermal load modeling in microgrids. This article predicts the hourly thermal load for the next 24 hours by using the Elman neural network and neural wavelet neural network (WNN). The principles of the proposed hybrid algorithm are firstly analysed. Then the impact factors of solar radiation and weather, which have important effects on thermal loads, are studied. The performance of the algorithm is analyzed and validated via real thermal load data obtained from Haidian Writers Association, Beijing. Comparisons are undertaken among four kinds of forecasting results to verify the importance of the solar radiation and weather data. The cloud theory is introduced as an effective tool to analyse the precision of the predicted results, and the errors are represented by using three numerical characteristics.

The Prediction of Thermal Loads in Building by Means of the EN ISO 13790 Dynamic Model: A Comparison with TRNSYS

The heating and cooling requirements in buildings are primarily dependent on weather conditions and on the physical and geometrical characteristics of the envelope, and secondly on occupant behaviour. Likewise, regarding the electrical loads, in the presence of multi-generation systems that employ also renewable sources, the prediction of heating and cooling load profiles allows a better planning of the production units, with consequent economic benefits. Furthermore, in a future scenario of Smart Grids, a coordinated production of heating and cooling plants with adjacent systems could be achieved. In order to assess the best operational conditions, an estimate of heating and cooling needs is required in home automation systems to minimize or to rationalize the consumption of primary fossil sources, especially in the presence of a district thermal network. Heating and cooling demands in buildings can be determined by simplified calculation models, which must provide consistent results with those obtainable by more sophisticated commercial software, which are difficult to interface in these automation systems. Therefore, the dynamic model 5R1C described in the standard EN ISO 13790 that employs 6 parameters (5 resistances and 1 capacitor) was tested. This model is simple, requires limited computational times and can be implemented with common programming language. The 5R1C model, in function of the envelope characteristics, determines heating and cooling loads of buildings taking into account the effects linked to thermal inertia. Furthermore, if appropriately predicted climate data are provided, the expected load profiles can be obtained. In order to determine advantages, limitations and accuracy of the 5R1C model, a comparison with the TRNSYS software for three different types of buildings was conducted.

Analytical solution to thermal loading and unloading of a cylinder subjected to periodic surface heating

ABSTRACTAn attempt is made to develop an analytical model for the prediction of thermal loading into a partially plastic state and unloading into an elastic state of a cylinder subjected to periodic boundary condition. The uncoupled theory of thermoelasticity is used as the cylinder is heated or cooled slowly. Transient temperature distribution in the cylinder is obtained by the use of Duhamel's theorem. It is assumed that the ends of the cylinder are free and hence a state of generalized plane strain is operative in the axial direction. An elastic and two plastic regions with different mathematical forms of Tresca's yield condition are formulated and solved analytically. Linearly hardening material behavior is assumed. Sudden unloading approximation is used to model unloading into a plastically predeformed elastic region. The model is verified in comparison to a purely numerical solution and by observing satisfaction of equilibrium, interface, and boundary conditions in every stage of deformation.