Occupancy data is a critical input parameter for building energy simulation since it has a big impact on the precision and accuracy of building energy model performance. However, current approaches ...to get such data through the conventional occupancy detection technology require either implementation of a large-scale sensor network and/or sophisticated and time-consuming computational algorithms, which to some degree limits the application of the real-time occupancy data for building energy simulation. In the era of the mobile internet, the massive people position data, which is generated by smartphone users and stored on cloud servers, offers a potential to solve this important problem. Such mobile data source is precisely monitored, real-time updated, and accessible with affordable time and labor cost upon customer's agreements in some regions, and therefore could be one of the alternatives to traditional occupancy detection methods.
This paper presents an investigation of whether and how the mobile-internet positioning data can benefit building energy simulation. This paper first summarizes the pros and cons of several mainstream occupancy detection methods. Then, the principle of the proposed mobile-internet-based occupancy detection method is introduced. The methodology of using such occupancy data for building energy simulation is developed. An energy performance model of a complex building in Shanghai with a whole building simulation software EnergyPlus is used as a pilot case study to demonstrate the effectiveness of the proposed methodology. A calibration is performed using the building automation system data and the mobile-internet-based occupancy data. The simulation results show that mobile-internet-based occupancy data can help improve the building model prediction accuracy.
•An overview of current occupancy detection methods is presented.•A mobile-internet-based occupancy detection method is proposed.•A case study building energy model is built to investigate the application of the proposed method in calibrated simulation.•This occupancy detection method can be further used in occupant-centered HVAC control researches.
Air-to-air energy recovery ventilators (AERVs) are widely applied in building heating, ventilation, and air conditioning (HVAC) systems to reduce energy consumption by reducing heating and cooling ...loads caused by ventilation. Its energy exchange efficiency data are typically acquired experimentally and are time consuming to be applied to engineering fields when the number of working scenarios to be tested is large. Instead of manual labor and time, detailed numerical calculation methods require the characteristic information of the membrane material, which is difficult for manufacturers to provide. Hence, an energy exchange efficiency prediction approach based on a multivariate polynomial regression model is proposed herein to predict the energy exchange efficiency of the membrane-based AERV (MERV) core. In the proposed approach, a simplified numerical efficiency calculation model of a cross-flow air-to-air enthalpy exchanger is trained with a small sample of experimentally measured efficiency data. Subsequently, the trained model is adopted to predict the energy exchange efficiency of the MERV core made of the same membrane, under other flowrates and structures. As verified by the experimental data, the proposed approach can predict the energy exchange efficiency with absolute deviation limits within ±8.0%. Compared to detailed numerical calculation methods, the proposed approach requires less membrane characteristic information and calculations; thus, the proposed approach is practical for engineering applications to simulate the equipment performance over different conditions and help sizing the equipment with less requirements.
•A combination of theoretical ε-NTU method and data-mining methods.•No requisition of known thermal properties of the membrane.•Rapid prediction based on a small sample of experimental test data.
Air conditioning water systems account for a large proportion of building energy consumption. In a pressure-controlled water system, one of the key measures to save energy is to adjust the ...differential pressure setpoints during operation. Typically, such adjustments are based either on certain rules, which rely on operator experience, or on complicated models that are not easy to calibrate. In this paper, a data-driven control method based on reinforcement learning is proposed. The main idea is to construct an agent model that adapts to the researched problem. Instead of directly being told how to react, the agent must rely on its own experiences to learn. Compared with traditional control strategies, reinforcement learning control (RLC) exhibits more accurate and steady performances while maintaining indoor air temperature within a limited range. A case study shows that the RLC strategy is able to save substantial amounts of energy.
Air-to-air membrane-energy exchangers have been widely used in green buildings as passive energy-recovery technologies for energy conservation and carbon reduction. An air-to-air membrane-energy ...exchanger conserves building energy by forcing the indoor exhaust air and unconditioned fresh air to exchange heat and moisture. As a critical performance indicator, the energy-exchange efficiency is primarily determined by the core. To improve energy-saving performance, many studies have been conducted to analyze and enhance the efficiency of air-to-air membrane-energy exchanger cores by manufacturing, testing, modeling, and comparing different cores. However, few studies have investigated the uncertainty introduced by the manufacturing and testing processes (most related research only studied the uncertainty of the sensible efficiency), which means that some positive results of those controlled trials are possibly owing to the manufacturing tolerance or testing errors, rather than real technical progress. This study addresses this problem by quantifying the efficiency uncertainty caused by manufacturing and testing processes. Three cores were produced using the same manufacturing process and materials, followed by identical experimental tests on their sensible, latent, and enthalpy efficiencies. Comparison results suggest that even when all variables are controlled, the differential enthalpy efficiency of two “identical” cores still vary from 0 % to 2 %. Variance of repetitive testing results of a certain core can afford an enthalpy efficiency of 4 % owing to a non-ideal experimental procedure. When manufacturing tolerances and experimental errors are considered simultaneously, the differential sensible efficiency among multiple “identical” cores is within 0–5%, whereas latent efficiency data in winter shows a significant variation of 5–10 % for “identical” cores. The contributions and conclusions of this study are as follows. The efficiency uncertainty caused by manufacturing and testing processes is non-negligible, particularly in latent efficiency. This phenomenon could alert researchers and engineers in this field to conduct repetitive manufacturing and testing while trying to modify/test/compare air-to-air membrane-energy exchanger cores. Because this study is based on a limited number of cores, the quantitative results of this study should be considered a non-universal reference, and a systematic benchmark must be established in the future.
Model-free reinforcement learning (RL) techniques are currently drawing attention in the control of heating, ventilation, and air-conditioning (HVAC) systems due to their minor pre-conditions and ...fast online optimization. The simultaneous optimal control of multiple HVAC appliances is a high-dimensional optimization problem, which single-agent RL schemes can barely handle. Hence, it is necessary to investigate how to address high-dimensional control problems with multiple agents. To realize this, different multi-agent reinforcement learning (MARL) mechanisms are available. This study intends to compare and evaluate three MARL mechanisms: Division, Multiplication, and Interaction. For comparison, quantitative simulations are conducted based on a virtual environment established using measured data of a real condenser water system. The system operation simulation results indicate that (1) Multiplication is not effective for high-dimensional RL-based control problems in HVAC systems due to its low learning speed and high training cost; (2) the performance of Division is close to that of the Interaction mechanism during the initial stage, while Division’s neglect of agent mutual inference limits its performance upper bound; (3) compared to the other two, Interaction is more suitable for multi-equipment HVAC control problems given its performance in both short-term (10% annual energy conservation compared to baseline) and long-term scenarios (over 11% energy conservation).
To conserve building energy, optimal operation of a building’s energy systems, especially heating, ventilation and air-conditioning (HVAC) systems, is important. This study focuses on the ...optimization of the central chiller plant, which accounts for a large portion of the HVAC system’s energy consumption. Classic optimal control methods for central chiller plants are mostly based on system performance models which takes much effort and cost to establish. In addition, inevitable model error could cause control risk to the applied system. To mitigate the model dependency of HVAC optimal control, reinforcement learning (RL) algorithms have been drawing attention in the HVAC control domain due to its model-free feature. Currently, the RL-based optimization of central chiller plants faces several challenges: (1) existing model-free control methods based on RL typically adopt single-agent scheme, which brings high training cost and long training period when optimizing multiple controllable variables for large-scaled systems; (2) multi-agent scheme could overcome the former problem, but it also requires a proper coordination mechanism to harmonize the potential conflicts among all involved RL agents; (3) previous agent coordination frameworks (identified by distributed control or decentralized control) are mainly designed for model-based control methods instead of model-free controllers. To tackle the problems above, this article proposes a multi-agent, model-free optimal control approach for central chiller plants. This approach utilizes game theory and the RL algorithm SARSA for agent coordination and learning, respectively. A data-driven system model is set up using measured field data of a real HVAC system for simulation. The simulation case study results suggest that the energy saving performance (both short- and long-term) of the proposed approach (over 10% in a cooling season compared to the rule-based baseline controller) is close to the classic multi-agent reinforcement learning (MARL) algorithm WoLF-PHC; moreover, the proposed approach’s nature of few pending parameters makes it more feasible and robust for engineering practices than the WoLF-PHC algorithm.
When modeling the energy performance of existing buildings, model calibration always serves as an essential and necessary step to ensure the accuracy and applicability of building models. Model ...calibration for detailed energy models, which is also known as calibrated simulation (CS), refers to the process of tuning model's input parameters to narrow down the mismatch between the simulation result and the real-monitored data of building energy consumptions. Two major problems of current CS are: (1) a successful calibration requires HVAC domain knowledge of modelers. (2) Traditional building calibration process is labor-intensive and time-consuming. To solve these problems, a normative energy modeling based quick auto-calibration approach is proposed in this study. This approach is able to reduce the modeling time and simplify the calibration procedures. Firstly, this paper demonstrates the working principles and the credibility of Normative Energy Modeling (NEM). Then, the methodology of the proposed quick auto-calibration approach is elaborated. The energy performance model of a large hotel located in Shanghai, China with NEM is used as a case study to investigate the effectiveness of the proposed quick auto-calibration approach. The simulation result of the case study suggests that the proposed approach can significantly simplify the procedures of model calibrations while still achieve a good accuracy specified by ASHRAE Guideline 14 1. Besides, the results further confirm that such method requires much less time and computational power for modeling, compared with other counterparts (e.g., “Autotune” calibration 2). With the advantage of speed and accuracy, the auto-calibration proposed in this paper is promising to be applied in both engineering and research field.
Chillers are responsible for nearly 30–40% of the energy consumption of an HVAC system. To enhance the efficiency of HVAC systems, it is necessary to optimize the operation of chillers. Conventional ...chiller operation strategies are mostly intended to save energy by suitably distributing the chiller cooling load in accordance with the measured cooling system data, which is typically considered noise-free. In other words, the uncertainty of the cooling system measurement and the necessity of customized optimization—as different buildings require different tradeoffs between energy-saving and indoor comfort—are usually not considered. This paper proposes a stochastic optimized chiller operation strategy based on multi-objective optimization and measurement uncertainty. The strategy consists of five steps: (1) Quantify the indoor comfort utility and energy consumption utility by using utility functions; (2) integrate these two utilities into one comprehensive utility using user-defined weights; (3) specify the measurement uncertainty distribution in the cooling system; (4) traverse each possible operation plan, and calculate the mean value (i.e., mathematical expectation) of its corresponding comprehensive utility (expected utility, EU); and (5) determine the optimal operation plan to be the one corresponding to the maximum EU value. The performance of the proposed strategy is validated by establishing a cooling system model on TRNSYS and comparing the findings with those of two conventional chiller operation strategies commonly adopted in Shanghai: the cooling load control strategy and coefficient of performance (COP) optimization strategy. Moreover, the robustness of the proposed strategy is validated by performing a comparison with the deterministic multi-objective strategy, which assumes that the measured system data is noise-free. The simulation results suggest that the proposed strategy can be customized to satisfy different optimization requirements such as less energy consumption or a high level of indoor comfort.
In the domain of optimal control for building HVAC systems, the performance of model-based control has been widely investigated and validated. However, the performance of model-based control highly ...depends on an accurate system performance model and sufficient sensors, which are difficult to obtain for certain buildings. To tackle this problem, a model-free optimal control method based on reinforcement learning is proposed to control the building cooling water system. In the proposed method, the wet bulb temperature and system cooling load are taken as the states, the frequencies of fans and pumps are the actions, and the reward is the system COP (i.e., the comprehensive COP of chillers, cooling water pumps, and cooling towers). The proposed method is based on Q-learning. Validated with the measured data from a real central chilled water system, a three-month measured data-based simulation is conducted under the supervision of four types of controllers: basic controller, local feedback controller, model-based controller, and the proposed model-free controller. Compared with the basic controller, the model-free controller can conserve 11% of the system energy in the first applied cooling season, which is greater than that of the local feedback controller (7%) but less than that of the model-based controller (14%). Moreover, the energy saving rate of the model-free controller could reach 12% in the second applied cooling season, after which the energy saving rate gets stabilized. Although the energy conservation performance of the model-free controller is inferior to that of the model-based controller, the model-free controller requires less a priori knowledge and sensors, which makes it promising for application in buildings for which the lack of accurate system performance models or sensors is an obstacle. Moreover, the results suggest that for a central chilled water system with a designed peak cooling load close to 2000 kW, three months of learning during the cooling season is sufficient to develop a good model-free controller with an acceptable performance.
Optimal operation of chillers could be realized by controlling chiller on–off status and resetting chilled water supply/return temperature. The chilled water supply temperature could influence both ...chiller power and indoor comfort simultaneously. Hence, it is necessary to optimize its set point. By combining the reinforcement learning (RL) technique and expertise knowledge, a hybrid model-free chilled water temperature resetting method for chillers is proposed, to improve the robustness and learning speed of RL control. The proposed method and the comparative expert manual control are applied to control an actual HVAC system in Shanghai. Engineering practice results suggest: (1) the energy-saving performance of the hybrid model-free method is close to that of expert manual control, with limited comfort sacrifice, acceptable robustness, learning speed and control stability; (2) without model establishing and persistent manual overwatch work, the proposed control method could conserve manual labor before and after the intervention of optimized control. Based on the engineering practice, a supplementary simulation study is conducted, which indicates: (1) the performance of the proposed method only takes one cooling season to converge; (2) the proposed hybrid method outperforms both pure RL-based control and pure rule-based control in the first training cooling season, while inferior to model predictive control; (3) the proposed method is an acceptable alternative for systems without sufficient data or certain equipment condition, due to its online self-learning mechanism and low pre-condition requirement on historical data and models.
