Multi-zone Building Research Articles

Deep reinforcement learning optimal control strategy for temperature setpoint real-time reset in multi-zone building HVAC system

Determining a proper trade-off between energy consumption and indoor thermal comfort is important for HVAC system control. Deep Q-learning (DQN) based multi-objective optimal control strategy is designed for temperature setpoint real-time reset to balance the energy consumption and indoor air temperature. In addition, this study develops an EnergyPlus-Python co-simulation testbed to evaluate DQN control strategy in a simulation environment. A case study experiment is conducted to evaluate the performance of DQN control strategy for real-time reset of supply air temperature and chilled supply water temperature setpoint in a multi-zone building VAV system. The developed EnergyPlus-Python co-simulation testbed is used to train and test the DQN control strategy for performance analysis. The applied DQN strategy leans to update control actions (i.e. temperature setpoint) through interaction with the simulation environment. Simulation results show that the DQN control strategy is effective in finding a proper trade-off between the energy consumption of HVAC system and indoor air temperature. Meanwhile, the DQN control strategy can find a proper temperature setpoint reset sequence in smaller training episodes, and the control actions can be stable after ten DQN training episodes. This study provides a preliminary direction of deep reinforcement learning control strategy for temperature setpoint real-time reset in multi-zone building HVAC systems.

End-to-End Deep Reinforcement Learning Control for HVAC Systems in Office Buildings

The heating, ventilation, and air conditioning (HVAC) system is a major energy consumer in office buildings, and its operation is critical for indoor thermal comfort. While previous studies have indicated that reinforcement learning control can improve HVAC energy efficiency, they did not provide enough information about end-to-end control (i.e., from raw observations to ready-to-implement control signals) for centralized HVAC systems in multizone buildings due to the limitations of reinforcement learning methods or the test buildings being single zones with independent HVAC systems. This study developed a model-free end-to-end dynamic HVAC control method based on a recently proposed deep reinforcement learning framework to control the centralized HVAC system of a multizone office building. By using the deep neural network, the proposed control method could directly take measurable parameters, including weather and indoor environment conditions, as inputs and control indoor temperature setpoints at a supervisory level. In some test cases, the proposed control method could successfully learn a dynamic control policy to reduce HVAC energy consumption by 12.8% compared with the baseline case using conventional control methods, without compromising thermal comfort. However, an over-fitting problem was noted, indicating that future work should first focus on the generalization of deep reinforcement learning.

Evaluating SARS‐CoV‐2 airborne quanta transmission and exposure risk in a mechanically ventilated multizone office building

The world has faced tremendous challenges during the COVID-19 pandemic since 2020, and effective clean air strategies that mitigate infectious risks indoors have become more essential. In this study, a novel approach based on the Wells-Riley model applied to a multizone building was proposed to simulate exposure to infectious doses in terms of “quanta”. This modeling approach quantifies the relative benefits of different risk mitigation strategies so that their effectiveness could be compared. A case study for the US Department of Energy large office prototype building was conducted to illustrate the approach. The infectious risk propagation from the infection source throughout the building was evaluated. Different mitigation strategies were implemented, including increasing outdoor air ventilation rates and adding air-cleaning devices such as Minimum Efficiency Reporting Value (MERV) filters and portable air cleaners (PACs) with HEPA filters in-room/in-duct germicidal ultraviolet (GUV) lights, layering with wearing masks. Results showed that to keep the risk of the infection propagating low the best strategy without universal masking was the operation of in-room GUV or a large industrial-sized PAC; whereas with masking all strategies were acceptable. This study contributes to a better understanding of the airborne transmission risks in multizone, mechanically ventilated buildings and how to reduce infection risk from a public health perspective of different mitigation strategies.

A Review of Reinforcement Learning Applications to Control of Heating, Ventilation and Air Conditioning Systems

Reinforcement learning has emerged as a potentially disruptive technology for control and optimization of HVAC systems. A reinforcement learning agent takes actions, which can be direct HVAC actuator commands or setpoints for control loops in building automation systems. The actions are taken to optimize one or more targets, such as indoor air quality, energy consumption and energy cost. The agent receives feedback from the HVAC systems to quantify how well these targets have been achieved. The feedback is captured by a reward function designed by the developer of the reinforcement learning agent. A few reviews have focused on the reward aspect of reinforcement learning applications for HVAC. However, there is a lack of reviews that assess how the actions of the reinforcement learning agent have been formulated, and how this impacts the possibilities to achieve various optimization targets in single zone or multi-zone buildings. The aim of this review is to identify the action formulations in the literature and to assess how the choice of formulation impacts the level of abstraction at which the HVAC systems are considered. Our methodology involves a search string in the Web of Science database and a list of selection criteria applied to each article in the search results. For each selected article, a three-tier categorization of the selected articles has been performed. Firstly, the applicability of the approach to buildings with one or more zones is considered. Secondly, the articles are categorized by the type of action taken by the agent, such as a binary, discrete or continuous action. Thirdly, the articles are categorized by the aspects of the indoor environment being controlled, namely temperature, humidity or air quality. The main result of the review is this three-tier categorization that reveals the community’s emphasis on specific HVAC applications, as well as the readiness to interface the reinforcement learning solutions to HVAC systems. The article concludes with a discussion of trends in the field as well as challenges that require further research.

Thermal modeling of the occupied multi-zone buildings taking into account the uncertainties of occupant behavior

In this paper, the interaction between residential apartments in the same building is considered by including the occupant behavior. The study is applied to three-floor residential building. First, a multi-zone building model was implemented and validated by comparison with thermal simulation tools. Deterministic study allows us to determine the worst case of energy highest consumption. The interior insulation thickness is then progressively increased to obtain its optimal value with respect to heat loss and equilibrated energy consumption between apartments. Finally, a random variation of family absence number and duration has been applied to evaluate their influence on energy demand considering heat interaction between apartments.

Nonlinear Distributed Model Predictive Control for multi-zone building energy systems

This paper presents a distributed Model Predictive Control (MPC) approach for multi-zone building energy systems based on nonlinear Modelica controller models. The method considers both thermal and hydraulic coupling among different building zones. The iterative and parallel distributed optimization approach builds upon an uncooperative approach for thermal coupling using the Nash equilibrium approach and a cooperative approach for the hydraulic coupling using the Alternating Direction Method of Multipliers (ADMM). Apart from thermal coupling through walls, the modeling takes thermal coupling through doors into account using a data-driven approach, which calculates the inter-zone air exchanges based on temperature differences between door-coupled zones. The hydraulic coupling enables consideration of interactions between the zones introduced by a shared, central Heating, Ventilation and Air Conditioning (HVAC) system. The distributed MPC framework is structured in an easy-scalable, plug-and-play composition, where local systems are automatically assigned to the global coordination scheme. The distributed MPC method is applied to a simulative nonlinear case study, consisting of a six-room-building Modelica model considering both thermal and hydraulic interactions. The benefits of the proposed approach are demonstrated and compared against centralized and decentralized control concepts in terms of energy consumption, discomfort and computation time.

Development of an Integrated Simulation Platform for Modelling and Simulation of the Energy Consumption of a Multi-zone Building

This paper focuses on a simulation platform to model and simulate the energy consumption of a building with multiple zones under the influence of daily and seasonal environmental factors, as well as the impact of the HVAC system and human activities. By using model-based requirements engineering (MBRE), the requirements of the simulation platform are hierarchically derived and specified. On this basis, the simulation platform is designed and developed. With the help of the simulation platform, it is possible to obtain a mathematical model for a realistic description of the thermodynamics and energy consumption of the individual zones of a building, which can be used as a basis for the model-based development of the energy management system, e.g. intelligent energy dispatch strategy. In the following, the concept of the simulation platform and the process will be demonstrated and validated with an application example of the research building Open Mobility Lab (OML).

Investigating the flexibility of a novel multi-zone air heating and ventilation system using model predictive control

In forced air heating systems, there are basically two approaches to set the temperature of each zone within a multi-zone building: regulating the supply airflow rate and regulating the supply air temperature (or a combination of these two parameters). Conventional variable air volume (VAV) systems usually maintain the supply air temperature constant within all rooms and regulate only the supply airflow rate to the room. Recently, a novel multi-zone air heating and ventilation (MZHV) system has been developed, which is capable of regulating both the supply airflow rate and the supply air temperature to the rooms independently of each other. This paper investigates the flexibility of the novel MZHV system and compares it with the conventional VAV systems. The MZHV system as well as a multi-zone case study building are modeled in a high-fidelity simulation environment (IDA ICE). Instead of traditional rule-based control schemes, a model predictive control (MPC) scheme is developed to control real-time power consumption of the multi-zone building. For this purpose, a second-order state space model is identified for each zone and calibrated against the results from the high fidelity simulations. MPC is implemented in MATLAB environment by using the second-order state space models. Two co-simulations are performed to compare the flexibility of the novel MZHV system with a conventional VAV system. The results show that the MZHV system provides a significant flexibility in terms of the energy usage, without affecting the comfort levels. This flexibility can for example be exploited to reduce the overall energy usage or provide ancillary services for a smart electricity grid.

Hybrid Physics-based and Data-driven Model Predictive Control for Multi-Zone Building's Thermal Comfort Under Disjunctive Uncertainty

This paper develops a hybrid approach that utilizes the distributional information of the disjunctive uncertainty sets and incorporates them into the model predictive control (MPC). This approach aims at the multi-zone building control to the thermal comfort, and it's robust to the uncertain weather forecast errors. The control objective is to maintain each zone's temperature and relative humidity within the specified ranges using the minimum cost of energy of the underlying heating system. The hybrid model is constructed using a physics-based and regression method for the temperature and relative humidity of each zone in the building. The uncertainty space is based on historical weather forecast error data, which are captured by a group of disjunctive uncertainty sets using k-means clustering algorithm. Machine learning approaches based on principal component analysis and kernel density estimation are used to construct each basic uncertainty set and reduce the conservatism of resulting robust control action under disturbances. A robust MPC framework is developed based on the proposed hybrid model and data-driven disjunctive uncertainty set. An affine disturbance feedback rule is employed to obtain a tractable approximation of the robust MPC problem. A case study of controlling temperature and relative humidity of a multi-zone building in Ithaca, New York, USA, is presented. The results demonstrate that the proposed hybrid approach can reduce 9.8% to 17.9% of total energy consumption compared to conventional robust MPC approaches. Moreover, the proposed hybrid approach can essentially satisfy the thermal constraints that certainty equivalent MPC and robust MPC largely violate.

A mathematical model of the room temperature dynamic response in multi-zone buildings

There are many factors that affect the change of room temperature, especially outdoor environmental factors, which makes it difficult to establish a model to describe the dynamic changes of the real time room temperature. In response to this problem, this study proposes a RC network two-node wall structure method to describe and analyze the heat conduction process of the building envelope. With this method, we first divide building envelope such as walls and floors into two-story structures, and then use energy balance theory to establish the energy balance equation of each node. Finally, we obtained the dynamic response equations of room temperature and humidity in each room of the whole building. This method simplifies the complex heat transfer process of the building envelope to a simple mathematical problem. The experimental validation indicated the accuracy of the proposed model.

Data-Driven Aggregate Thermal Dynamic Model for Buildings: A Regression Approach

The thermal inertia of buildings brings considerable flexibility to the building heating and cooling loads, which is believed to be a promising demand response resource in energy systems. However, it is challenging to utilize the thermal inertia of buildings in the operation of energy systems because of the complicated thermal dynamics and high computational cost. This paper proposes a data-driven aggregate thermal dynamic model (ATDM) for the multi-zone building and building cluster, respectively, which offers an equivalent and low-complexity building model for the operation and control of energy systems. The ATDM consists of the aggregation equation and the state equation. The former projects the detailed real states of buildings into the characteristic state (i.e., aggregate state) using an affine function, and the latter describes the thermal dynamics of buildings using the aggregate state. The ATDM is formulated for two practical load control strategies for the building cluster, including direct load control and indirect load control. Then, the constrained nonlinear regression model is proposed to estimate the model parameters and occupant behavior, for which an efficient algorithm based on the block coordinate descent method is developed by exploiting the decomposable structure of the regression model. Simulation results based on real-world data show that the root mean square error and mean absolute percentage error for the multi-zone building (or building cluster) are below 0.72 °C and 1.44% (or 0.32°C and 1.39%), respectively, verifying the effectiveness of the proposed methods.

Using a calibrated multi-zone building model to analyse the impact of occupant behaviour on building performance

Building performance simulation has been commonly used for performance-based building design. However, the simulation accuracy is closely related to the model and input parameters regarding the building physics and occupant behaviour. To check and improve the accuracy of the simulation, the model usually needs to be calibrated using measured data. Uncertain parameters are adjusted in this process to reduce the discrepancy between simulation and measurement results. Although occupant behaviour has a significant impact on building performance, it has not been considered a key component of model calibration in previous studies. Model calibration using only indoor environmental and energy consumption data without considering occupant behaviour is unreliable. In this context, a step-by-step model calibration approach is proposed in this study to optimise the estimated parameters in typical operational scenarios where occupant behaviour can be conveniently determined. A case study shows that although the simulation results for apartment energy use are similar before and after model calibration, differences can be observed in the evaluation of indoor air quality and thermal environment. Simulation results using occupant behaviour based on design standards are significantly deviated from the measured values, compared to that using actual occupant behaviour.

An averaging approach for nearly-optimal control of multi-zone buildings

Controlling Heating, Ventilation and Air-Conditioning (HVAC) in a way that is both optimal (energy efficient) and adaptive (able to cope with uncertainties) is a recognized challenge in the building sector. This work presents a simple modelling approach to HVAC operation aiming for adaptive and optimal control. The proposed framework exhibits several advantages amenable for implementation in energy management systems: minimum number of parameters to be estimated; preserving the distributed nature of multiple zones in the estimation process; easiness of imposing positivity and controllability constraints on the estimated coefficients. Optimality is addressed using an adaptive linear quadratic control philosophy. Comparisons with Proportional-Integral-Derivative (PID) and non-adaptive controllers show the capability of the proposed framework of coping with uncertainty while delivering nearly optimal controls.

Potential evaluation and analysis of near-to-net zero energy building in hot and dry climate

• Near-to-net zero energy building in hot and dry climate has been investigated. • NZEBs model has been developed and simulated using TRNSYS models/components. • Space cooling, space heating, and electrical energy outcomes was developed. • Maximum reduction rate ratio in energy demand 48.2% was achieved. In a hot and dry climate like the weather of Iraq, near-to-net zero energy buildings (NNZEBs) can become one of the effective solutions to reduce energy demands in the residential buildings sector. Therefore, the current work aims to cornerstone of the developed solution combined the different scenarios for residential building with a total living area of 240 m 2 has been considered. The investigated scenarios include different building constructions, orientations, control strategies, and using solar PV panels as renewable energy techniques. The proposed scenarios have been developed and simulated using TRNSYS models/components. The energy demand for space cooling, space heating, and electrical energy outcomes from an NNZEB model was developed using the TRNSYS simulation workspace. The data input of the residential multi-zone building envelope is configured using TRNBuild (i.e., building simulation software) and then introduced into TRNSYS. Moreover, the solar PV panels tool is developed in TRNSYS to evaluate the impact of renewable energy on energy consumptions. A full cost analysis of the energy consumption has been carried out to see if this technology is viable. In terms of maximum energy demand, the preliminary building requires 340 kWh/m 2 per year as a cooling demand compared with 65 kWh/m 2 per year as a heating demand. The results showed that the maximum reduction rate in energy demand for cooling demand of 33.9% was achieved by changing the construction, ventilation, and control systems. The scenarios of using double glaze and insulating the building showed the most significant impact on the reducing of energy demand with a total reduction rate ratio of 34.40% compared with other scenarios. From economic analysis, it can be concluded that the increase in the energy tariff led to an increase in the yearly energy cost for the preliminary building without solar PV panels. The results highlighted that the effects of key investigated scenarios should be considered in combination through a parametric investigation.

Distributed Diagnosis of Sensor and Actuator Faults in Air Handling Units in Multi-Zone Buildings: A Model-Based Approach

The presence of sensing and actuating equipment faults in Heating, Ventilation and Air-Conditioning (HVAC) systems may result in unsatisfactory indoor thermal conditions and significant wasted energy in the building sector. The design of fault diagnosis algorithms for multi-zone buildings is a highly challenging procedure due to the overall complexity of the system. This paper proposes a distributed model-based fault diagnosis algorithm for detecting, isolating and identifying faults that can affect the actuating devices of Air Handling Units (AHUs) and the sensor devices. Based on modeling the temperature dynamics of each local AHU and its underlying zone with respect to its neighboring zones, a local fault diagnosis agent that collects local control signals and local and neighboring sensor measurements is designed. Each local diagnosis agent is designed to estimate on-line the states of each AHU. For each estimation, a residual signal is derived that corresponds to the difference between the measured and estimated value at each time step. Considering healthy conditions, the residual is bounded by its corresponding adaptive threshold that is calculated considering the effects of modeling uncertainties. Each corresponding pair of residual and adaptive threshold forms an analytical redundancy relation which indicates the possible presence of one or more faults. Exploiting the dependency of each actuator and sensor fault with each Analytical Redundancy Relation (ARR), a decision logic is designed, that can reveal under certain conditions the location and type of the fault during the operation of the system. The proposed algorithm is evaluated using a simulation example of a primary school building with 25 zones.

Detailed Thermodynamic Modeling of Multi-Zone Buildings with Resistive-Capacitive Method

Increased use of energy in buildings and HVAC systems requires advanced control schemes like model-based control to improve energy efficiency, which in turn requires accurate thermodynamic models of buildings. The Resistive-Capacitive (RC) method is a popular and versatile approach for thermal modeling of buildings. Despite this, it is not easy to find practical solutions of implementation of the RC method. It is the goal of this paper to clarify the RC method and demonstrate simple implementation of this method, especially for multi-zone buildings, which have more potential for energy savings from use of model-based control. This paper provides two contributions. First is a detailed explanation of the RC method, focusing on its use for developing a structure of a model and first-principles approach for estimation of parameters of a model. Second is a demonstration of an algorithm that enables automatic development of the structure of a model from basic information about a building (layout, construction elements) and its combination with data-based parameter estimation. Use of the algorithm is presented with a case-study on industrial multi-zone building, for which such a grey-box model is developed and analyzed. The resulting model is rapidly developed and used in a simulation with the measured data. The outputs of the model are compared with the measured temperatures and they show good fit.

Model order reduction of building energy simulation models using a convolutional neural network autoencoder

Building energy simulation (BES) tools are fundamental for predicting energy performance and comfort. However, detailed models are computationally complex and demand high simulation times. These lead to difficulties in parametric runs and numerical optimizations. Performing numerous retrofit scenarios is hardly feasible, especially in multi-zone buildings with complex geometries.This paper introduces a novel approach to model order reduction (MOR) of BES models. The approach utilizes a deep learning-based unsupervised convolutional neural network autoencoder (CNN-AE). The method decomposes complex time series data derived from detailed simulations into lower dimension features. The low-dimension representations can be grouped through clustering algorithms to build a reduced-order model (ROM). The approach in this study automatically finds archetype zones of the original model that represent the energy behavior of a group of rooms, and removes redundant ones. The energy demand of the whole building can be estimated through these archetype zones.Our investigation shows that CNN-AE can be efficiently applied to reduce complex building energy simulation models. As proof of concept, a detailed model of a multi-zone campus building with 889 thermal zones is compared to the ROM derived from the CNN-AE. Comprehensive autoencoder hyperparameter training to optimize the accuracy of the model is provided. The ROM supports different purposes, such as energy scenario developments, with a total error of less than 1% compared to the original model, and reduced simulation times by a factor of more than 16.

Factors affecting aerosol SARS-CoV-2 transmission via HVAC systems; a modeling study.

The role of heating, ventilation, and air-conditioning (HVAC) systems in the transmission of SARS-CoV-2 is unclear. To address this gap, we simulated the release of SARS-CoV-2 in a multistory office building and three social gathering settings (bar/restaurant, nightclub, wedding venue) using a well-mixed, multi-zone building model similar to those used by Wells, Riley, and others. We varied key factors of HVAC systems, such as the Air Changes Per Hour rate (ACH), Fraction of Outside Air (FOA), and Minimum Efficiency Reporting Values (MERV) to examine their effect on viral transmission, and additionally simulated the protective effects of in-unit ultraviolet light decontamination (UVC) and separate in-room air filtration. In all building types, increasing the ACH reduced simulated infections, and the effects were seen even with low aerosol emission rates. However, the benefits of increasing the fraction of outside air and filter efficiency rating were greatest when the aerosol emission rate was high. UVC filtration improved the performance of typical HVAC systems. In-room filtration in an office setting similarly reduced overall infections but worked better when placed in every room. Overall, we found little evidence that HVAC systems facilitate SARS-CoV-2 transmission; most infections in the simulated office occurred near the emission source, with some infections in individuals temporarily visiting the release zone. HVAC systems only increased infections in one scenario involving a marginal increase in airflow in a poorly ventilated space, which slightly increased the likelihood of transmission outside the release zone. We found that improving air circulation rates, increasing filter MERV rating, increasing the fraction of outside air, and applying UVC radiation and in-room filtration may reduce SARS-CoV-2 transmission indoors. However, these mitigation measures are unlikely to provide a protective benefit unless SARS-CoV-2 aerosol emission rates are high (>1,000 Plaque-forming units (PFU) / min).

Feasibility Study and Economic Analysis of a Fuel-Cell-Based CHP System for a Comprehensive Sports Center with an Indoor Swimming Pool

Unlike a general commercial building, heating for a building with an indoor swimming pool is highly energy-intensive due to the high energy demand for swimming water heating. In Korea, the conventional heating method for this kind of building is to use boilers and heat storage tanks that have high fuel costs and greenhouse gas emissions. In this study, a combined heat and power (CHP) system for such a building using the electricity and waste heat from a Phosphoric Acid Fuel Cell (PAFC) system was designed and analyzed in terms of its primary energy saving, CO2 reduction, fuel cell and CHP efficiency, and economic feasibility. The mathematical model of the thermal load evaluation was used with the 3D multi-zone building model in TRNSYS 18 software (Thermal Energy System Specialists, LLC, Madison, MI, USA) to determine the space heating demand and swimming pool heat losses. The energy efficiency of the fuel cell unit was evaluated as a function of the part-load ratio from the operating data. The fundamental components, such as the auxiliary boiler, thermal storage tank, and heat exchanger are also integrated for the simulation of the system’s operation. The result shows that the system has a high potential to improve the utilization efficiency of fuel cell energy production. Referring to the local condition of the energy market in Korea, an economic analysis was also carried out by using a specific FC-CHP capacity at 440 kW. The economic benefit is significant in comparison with a conventional heating system, especially for the full-time operating (FTO) mode. The net profit made by comparison with the conventional energy supply system is about 178,352 to 273,879 USD per year, and the payback period is expected to be 6.9 to 10.7 years under different market conditions.

Estimation of pollutant sources in multi-zone buildings through different deconvolution algorithms.

Effective identification of pollution sources is particularly important for indoor air quality. Accurate estimation of source strength is the basis for source effective identification. This paper proposes an optimization method for the deconvolution process in the source strength inverse calculation. In the scheme, the concept of time resolution was defined, and combined with different filtering positions and filtering algorithms. The measures to reduce effects of measurement noise were quantitatively analyzed. Additionally, the performances of nine deconvolution inverse algorithms under experimental and simulated conditions were evaluated and scored. The hybrid algorithms were proposed and compared with single algorithms including Tikhonov regularization and iterative methods. Results showed that for the filtering position and algorithm, Butterworth filtering performed better, and different filtering positions had little effect on the inverse calculation. For the calculation time step, the optimal Tr (time resolution) was 0.667% and 1.33% in the simulation and experiment, respectively. The hybrid algorithms were found to not perform better than the single algorithms, and the SART (simultaneous algebraic reconstruction technique) algorithm from CAT (computer assisted tomography) yielded better performances in the accuracy and stability of source strength identification. The relative errors of the inverse calculation for source strength were typically below 25% using the optimization scheme.