Heating, Ventilation, And Air Conditioning Control Research Articles

Real-time monitoring of occupancy activities and window opening within buildings using an integrated deep learning-based approach for reducing energy demand

Occupancy behaviour in buildings can impact the energy performance and operation of heating, ventilation and air-conditioning (HVAC) systems. HVAC, which uses conventional control strategies or “fixed” setpoint schedules, could not adjust to the conditioned spaces' actual requirements, resulting in building spaces being over or under-conditioned. While the unintended opening of windows can lead to substantial heat loss and consequently raises energy consumption. To optimise building operations, it is necessary to employ solutions such as demand-driven controls, which can monitor the utilisation of indoor spaces and provide the actual thermal comfort requirements of occupants. This study presents a novel vision-based deep learning framework for occupancy activity detection and recognition including the manual window operations in buildings. A region-based Convolutional Neural Network (R-CNN) model was trained and deployed to a camera for real-time detection and recognition. Based on the field experiments conducted within a case study University building, overall accuracy of 85.63% was achieved for occupancy activity detection and 92.20% for window operation detection. Building energy simulation and various scenario-based cases were used to assess the impact of such an approach on the building energy demand and provide insights into how the proposed detection method can enable HVAC systems to respond to dynamic changes within indoor spaces. Results showed that the proposed approach could reduce the over-or under-estimation of occupancy heat gains compared with the use of “fixed” or static profiles. In addition, the approach can help alert building users or managers about windows left open unintentionally, which can reduce unnecessary ventilation heat losses. Furthermore, the approach can also predict the room CO2 concentration and advise occupants about a suitable natural ventilation strategy. The study highlighted the potential of the multi-purpose detection approach, but further development is necessary, including optimisation of the deep learning model, full integration with HVAC controls and further model training and field testing.

AI-powered virtual assistants nudging occupants for energy saving: proactive smart speakers for HVAC control

ABSTRACT Virtual assistants powered by Artificial Intelligence (AI) and integrated into the smart home ecosystems facilitate human–building interactions. We have envisioned that the proactive virtual assistant capabilities could be designed to encourage energy conservation behaviours by relying on their nudging effect through conversational interactions, autonomous actuation and guiding users’ decision-making. To this end, we investigated how proactive virtual assistants, in a simulated smart home ecosystem, influence occupants to take energy-saving, adaptive actions for Heating, Ventilation and Air Conditioning (HVAC) operations and how participants’ personal characteristics affect their responses. Through an interactive online experiment, we collected data from 307 participants from diverse backgrounds across the United States. It was found that proactive communications with follow-up conversations can significantly increase the likelihood of accepting virtual assistance recommendations. This improvement was reflected in an increased number of participants (by 16%) who accepted energy-saving suggestions by comparing initial versus final responses during proactive conversations. Characterizing groups of participants based on their personal features and individual differences showed that user experience (with ∼30% increase), pro-environmental values/beliefs (with ∼24% to 35% increase) and forgiving thermal preferences (with ∼12% increase) had a significant influence on participants’ stated likelihood to accept virtual assistants’ recommendations and their evaluation of the general concept of proactive communication from virtual assistants.

From occupants to occupants: A review of the occupant information understanding for building HVAC occupant-centric control.

Occupants are the core of the built environment. Traditional heating, ventilation, and air-conditioning (HVAC) systems operate with predefined schedules and maximum occupancy assumptions with no consideration of specific occupant information. These generalized assumptions usually do not align with the actual demand and result in over-conditioning and occupant discomfort. In recent years, with the aid of Information & Communication Technology (ICT) and Computer Science (CS), it is possible to acquire real-time and accurate occupant information to satisfy the exact thermal requirement through specific HVAC control in one particular built environment. This mechanism is called HVAC “Occupant-centric Control (OCC).” HVAC OCC strategy starts with collecting the occupant’s information (e.g., presence/absence) and then applies it to meet the occupant’s requirement (e.g., thermal comfort). However, even though some research studies and field pilot demonstrations have been devoted to the field of OCC, there is a lack of systematic knowledge about occupant data, which is the principal component of OCC for HVAC researchers and practitioners. To fill this gap, this review paper discusses OCC with a particular emphasis on occupant information and investigates how this information can assist HVAC operation in providing an acceptable built environment in required spaces during the required time. We provide a fine-grained, comprehensive picture of occupant information, discuss its features, the modalities of information feed-in into the HVAC control, and the application of commonly utilized occupant information for OCC.

Energy Management Model for HVAC Control Supported by Reinforcement Learning

Heating, ventilating, and air conditioning (HVAC) units account for a significant consumption share in buildings, namely office buildings. Therefore, this paper addresses the possibility of having an intelligent and more cost-effective solution for the management of HVAC units in office buildings. The method applied in this paper divides the addressed problem into three steps: (i) the continuous acquisition of data provided by an open-source building energy management systems, (ii) the proposed learning and predictive model able to predict if users will be working in a given location, and (iii) the proposed decision model to manage the HVAC units according to the prediction of users, current environmental context, and current energy prices. The results show that the proposed predictive model was able to achieve a 93.8% accuracy and that the proposed decision tree enabled the maintenance of users’ comfort. The results demonstrate that the proposed solution is able to run in real-time in a real office building, making it a possible solution for smart buildings.

Hardware-Based Emulator with Deep Learning Model for Building Energy Control and Prediction Based on Occupancy Sensors’ Data

Heating, ventilation, and air conditioning (HVAC) is the largest source of residential energy consumption. Occupancy sensors’ data can be used for HVAC control since it indicates the number of people in the building. HVAC and sensors form a typical cyber-physical system (CPS). In this paper, we aim to build a hardware-based emulation platform to study the occupancy data’s features, which can be further extracted by using machine learning models. In particular, we propose two hardware-based emulators to investigate the use of wired/wireless communication interfaces for occupancy sensor-based building CPS control, and the use of deep learning to predict the building energy consumption with the sensor data. We hypothesize is that the building energy consumption may be predicted by using the occupancy data collected by the sensors, and question what type of prediction model should be used to accurately predict the energy load. Another hypothesis is that an in-lab hardware/software platform could be built to emulate the occupancy sensing process. The machine learning algorithms can then be used to analyze the energy load based on the sensing data. To test the emulator, the occupancy data from the sensors is used to predict energy consumption. The synchronization scheme between sensors and the HVAC server will be discussed. We have built two hardware/software emulation platforms to investigate the sensor/HVAC integration strategies, and used an enhanced deep learning model—which has sequence-to-sequence long short-term memory (Seq2Seq LSTM)—with an attention model to predict the building energy consumption with the preservation of the intrinsic patterns. Because the long-range temporal dependencies are captured, the Seq2Seq models may provide a higher accuracy by using LSTM architectures with encoder and decoder. Meanwhile, LSTMs can capture the temporal and spatial patterns of time series data. The attention model can highlight the most relevant input information in the energy prediction by allocating the attention weights. The communication overhead between the sensors and the HVAC control server can also be alleviated via the attention mechanism, which can automatically ignore the irrelevant information and amplify the relevant information during CNN training. Our experiments and performance analysis show that, compared with the traditional LSTM neural network, the performance of the proposed method has a 30% higher prediction accuracy.

Development of a testing and evaluation protocol for occupancy sensing technologies in building HVAC controls: A case study of representative people counting sensors

Occupancy-based control (OBC) in smart buildings provides numerous potentials to improve building energy efficiency. To achieve effective OBC, the occupancy status of a space or a building needs to be well understood in terms of whether the space is occupied and how many people are in the space. Well-designed and selected people counting technologies are cornerstones for OBC. However, the study and application of such technologies in OBC are relatively new. There are few mature people counting sensors available on the market. In addition, there is a lack of standardized guidance to comprehensively test, evaluate and compare the performance of people counting sensors, which further limits the selection and application of them in OBC. Hence, an innovative testing protocol was developed to fill in the gap. This paper introduces the design of the protocol with eight diversities and discusses case studies that evaluated four representative types of people counting sensors following the proposed protocol. It is found that the protocol can effectively guide a comprehensive evaluation of the selected sensors and lead to informative findings on the sensor performance. The test results can provide insights and suggestions not only for sensor developers to improve the sensor hardware and software design but also for heating, ventilation, and air-conditioning (HVAC) system designers and building managers to select the proper people counting sensors for OBC design. The test results on the sensor accuracy can also support additional evaluations of integrated OBC and building system operation. It is anticipated that the developed methodology can offer guidance on the test and evaluation of other occupancy sensing technologies.

Energy‐efficient cabin climate control of electric vehicles using linear time‐varying model predictive control

AbstractA cabin climate control system, often referred to as a heating, ventilation, and air conditioning (HVAC) system, is one of the largest auxiliary loads of an electric vehicle (EV), and the real‐time optimal control of HVAC brings a significant energy‐saving potential. In this article, a linear‐time‐varying (LTV) model predictive control (MPC)‐based approach is presented for energy‐efficient cabin climate control of EVs. A modification is made to the cost function in the considered MPC problem to simplify the Hessian matrix in utilizing quadratic programming for real‐time computation. A rigorous parametric study is conducted to determine optimal weighting factors that work robustly under various operating conditions. Then, the performance of the proposed LTV‐MPC controller is compared against a rule‐based (RB) controller and a nonlinear economic MPC (NEMPC) benchmark. Compared with the RB controller benchmark, the LTV‐MPC reaches the target cabin temperature at least 69 s faster with 3.2% to 15% less HVAC system energy consumption, and the averaged cabin temperature difference is 0.7°C at most. Compared with the NEMPC, the LTV‐MPC controller can achieve comparable performance in temperature regulation and energy consumption with fast computation time: the maximum differences in temperature and energy consumption are 0.4°C and 2.6%, respectively, and the computational time is reduced 72.4% on average with the LTV‐MPC.

Load shifting and energy conservation using smart thermostats in contemporary high-rise residential buildings: Estimation of runtime changes using field data

As the proportion of the population who resides in urban centers grows, the number of dwellings in high-rise residential buildings is also increasing. In Canada, heating, ventilation, and air conditioning (HVAC) typically accounts for over half of all energy use in this building type and, as such, consideration of these loads is key to effective energy demand-side management. Smart thermostats present a new opportunity conserve energy and shift electricity load to off-peak periods in these buildings through modifications to HVAC control strategies. In this study, three thermostat control strategies were tested in 49 suites across two contemporary high-rise residential buildings: 1) baseline operation (thermostat operating as a standard programmable thermostat); 2) occupancy-based control (thermostat operating using an occupancy-reactive control strategy); 3) load shifting control (thermostat operating using a pre-conditioning strategy). HVAC runtime data was collected from smart thermostats installed in the suites and was used to develop data-driven regression models to estimate baseline suite HVAC runtime based on weather conditions and suite characteristics. Thermostat-measured HVAC runtime data for periods in which occupancy-based or load shifting control strategies were in use in the suite was then compared with estimated baseline runtime for similar outdoor conditions. Occupancy-based control was found to reduce HVAC system runtime during the cooling seasonby 5.9% ± 46% on average. On aggregate, the load shifting control strategy was not effective at reducing peak period HVAC load during the cooling season. The strategy was able to reduce peak period loads in a subset of suites substantially, which may be leveraged to reduce electricity demand during capacity-constrained periods of grid operation, however, based on marginal emissions factors for Ontario’s electricity grid, even in these suites, the strategy was not effective in reducing GHG emissions overall.

Технология программной имплементации цифрового двойника в контур IoT управления HVAC-системой

The development of application software for cyber-physical systems of buildings involves the widespread use of Internet of Things (IoT) integration platforms. In practice, the flexible functionality of IoT platforms often leads to additional costs for software enhancement of existing and connection of new units, in particular digital twins. The paper proposes a technological solution for the implementation of a digital twin of the ventilation process in the IoT control loop of heating, ventilation and air conditioning (HVAC) systems for buildings and industrial facilities. The implementation and execution of the digital twin in the form of a dynamic simulation model in the object-oriented modelling language Modelica in the OpenModelica environment is considered. The IoT platform InfluxData, based on the TICK stack, is considered as an example of an integration environment. It is a horizontally-oriented IoT platform that contains the mechanism for collecting data from devices and the InfluxDB time-series database for storing metrics. To integrate simulation models on Modelica with InfluxDB, an OMPython server is proposed. In this case, the integration scripts are executed in the Python language, which as a result extends the traditional capabilities of the IoT platform significantly to the level of a digitally twinned control system. This HVAC control involves adapting control loops by taking into account the dynamics of the air distribution process over the ventilation network, evaluating and compensating for process inertia. The publication was prepared within the framework of the Academic Fund Program at the HSE University in 2020–2021 (grant № 21-04-039).

Thermal Imagery Feature Extraction Techniques and the Effects on Machine Learning Models for Smart HVAC Efficiency in Building Energy

The control of thermostats of a heating, ventilation, and air-conditioning (HVAC) system installed in commercial and residential buildings remains a pertinent problem in building energy efficiency and thermal comfort research. The ability to determine the number of people at a particular time in an area is imperative for energy efficiency in order to condition only occupied regions and thermally deficient regions. In this study of the best features comparison for detecting the number of people in an area, feature extraction techniques including wavelet scattering, wavelet decomposition, grey-level co-occurrence matrix (GLCM) and feature maps convolution neural network (CNN) layers were explored using thermal camera imagery. Specifically, the pretrained CNN networks explored are the deep residual (Resnet-50) and visual geometry group (VGG-16) networks. The discriminating potential of Haar, Daubechies and Symlets wavelet statistics on different distributions of data were investigated. The performance of VGG-16 and ResNet-50 in an end-to-end manner utilizing transfer learning approach was investigated. Experimental results showed the classification and regression trees (CART) model trained on only GLCM and Haar wavelet statistic features, individually achieved accuracies of approximately 80% and 84%, respectively, in the detection problem. Moreover, k-nearest neighbors (KNN) trained on the combined features of GLCM and Haar wavelet statistics achieved an accuracy of approximately 86%. In addition, the performance accuracy of the multi classification support vector machine (SVM) trained on deep features obtained from layers of pretrained ResNet-50 and VGG-16 was between 96% and 97%. Furthermore, ResNet-50 transfer learning outperformed the VGG-16 transfer learning model for occupancy detection using thermal imagery. Overall, the SVM model trained on features extracted from wavelet scattering emerged as the best performing classifier with an accuracy of 100%. A principal component analysis (PCA) on the wavelet scattering features proved that the first twenty (20) principal components achieved a similar accuracy level instead of training on the whole feature set to reduce the execution time. The occupancy detection models can be integrated into HVAC control systems for energy efficiency and security systems, and aid in the distribution of resources to people in an area.

Occupancy-based HVAC control using deep learning algorithms for estimating online preconditioning time in residential buildings

This paper presents a rule-based (RB) heating, ventilation, and air-conditioning (HVAC) control system using a multi-layer perceptron network, a deep learning algorithm, for estimating dynamic preconditioning time in residential buildings. The proposed system takes advantage of occupancy, indoor temperature, and weather data to make control decisions in buildings. The system performance is evaluated in terms of financial, demand-side management, energy-efficiency, and occupants’ thermal comfort. The proposed approach considers the perfect occupancy prediction assumption to remove the impact of the uncertainty associated with occupancy prediction and to estimate an upper bound limit for the system performance. The system performance is compared with that of conventional rule-based control approaches to show its effectiveness. To select the optimal control system, the TOPSIS method, as a multi-criteria decision-making approach, is employed. The sensitivity of the proposed system to the temperature setback is also assessed by considering conservative, medium, and deep setback bounds. It is demonstrated that the proposed system outperforms other alternatives when the deep and medium bounds are utilized. This study reveals two limitations of occupancy-based control systems by investigating their performance from financial and peak-demand points of view. First, these systems can cause peak-demand issues and increase the on-peak energy consumption by up to 10%. Secondly, employing a conservative setback can significantly decrease the financial merits of the system, leading to a discounted payback period of 10.87 years for implementing smart thermostats.

Application of an occupant voting system for continuous occupant feedback on thermal and indoor air quality – Case studies in office spaces

Smartphone applications or electronic devices in the form of occupant voting systems (OVS) have demonstrated to be feasible for acquiring feedback from occupants to support the building operation manager to identify indoor environmental quality (IEQ) problems and inappropriate control settings of heating, ventilation and air-conditioning (HVAC). The present paper aims to contribute to the growing research on OVS as a tool to support building operation management. The paper presents a study about a tangible OVS, denoted TiAQ, that occupants could use to vote on their here-and-now experiences regarding the thermal environment and indoor air quality. The main objectives were to identify whether there was an alignment between the votes collected with TiAQ and the monitored IEQ variables as well as to demonstrate the use of occupant votes to identify problematic IEQ conditions and appropriate strategies for HVAC control. The study demonstrated that thermal votes collected with TiAQ could be related to the variations in the indoor temperature and ventilation airflow. Additionally, collected votes could identify i.a. poorly set indoor temperature setpoint and appropriate control strategies that could reduce energy use by 46% and increase thermal comfort and cold complaints by 6% if current airflow and indoor temperature setpoint were lowered.

Development and Evaluation of Smart Home IoT Systems applied to HVAC Monitoring and Control

This work describes the development and evaluation of two smart home-based Internet of Things (IoT) systems applied to HVAC (Heating, Ventilation and Air Conditioning) monitoring and control, including parameters such as temperature, humidity, air quality, smoke detection, and

A hybrid deep transfer learning strategy for thermal comfort prediction in buildings

Since the thermal condition of living spaces affects the occupants' productivity and their quality of life, it is important to design effective heating, ventilation and air conditioning (HVAC) control strategies for better energy efficiency and thermal comfort. An essential step in HVAC control and energy optimization is thermal comfort modeling. Recently, data-driven thermal comfort models have been preferred over the Fanger's Predicted Mean Vote (PMV) model due to higher accuracy and ease of use. However, the unavailability of comprehensive labelled thermal comfort data from the occupants poses a significant modeling challenge. This paper addresses data inadequacy issues by adopting ‘transfer learning’ to leverage well learned knowledge from source domain (same climate zones) to target domain (different climate zone) where modeling data is sparse. Specifically, a Transfer Learning based Convolutional Neural Networks-Long Short Term Memory neural networks (TL CNN-LSTM) is designed for effective thermal comfort modeling that exploits the spatio-temporal relations in the thermal comfort data. The significant modeling parameters for TL CNN-LSTM are identified using the Chi-squared test. Further, the lack of sufficient samples across all thermal conditions in the available thermal comfort datasets was handled by Synthetic Minority Oversampling Technique (SMOTE). Experiments with two source (ASHRAE RP-884 and Scales Project) and one target (Medium US office) datasets demonstrate the ability of TL CNN-LSTM in achieving an accuracy of >55% with limited data in target buildings. The limitation of TL CNN-LSTM is its continued dependence on intrusive parameters and the challenges in assessing its adaptability to different climate zones.

A framework for a multi-source, data-driven building energy management toolkit

Poor building energy performance often arises from suboptimal operations such as inappropriate control sequences or hard faults in heating, ventilation, and air-conditioning (HVAC) systems. Furthermore, such deficiencies are often left unaddressed due to a lack of accessible analytical tools that can derive insights which identify energy-saving measures using multiple data resources. This paper presents a novel multi-source, data-driven building energy management toolkit as a synthesis of established inverse energy modelling, anomaly detection and diagnostics, load disaggregation, and occupancy and occupant complaint analytics methods in the literature. The toolkit contains seven functions that input HVAC controls, energy meter, Wi-Fi-based occupancy, and work order log data to detect hard and soft faults, improve sequences of operation, and monitor energy flows, occupancy patterns, and occupant satisfaction. The toolkit’s unique multifaceted analytical approach was demonstrated on a case study building as a proof of concept. Five faults pertaining to the air handling units’ mode of operation and heating coil valves were identified and the generated insights were used to pinpoint operational deficiencies stemming from inappropriate zone temperature overheating thresholds and perimeter heating devices. The toolkit, along with the data from the case study, is open-source and accessible through GitHub (Markus, 2021) to initiate and facilitate future development.

Energy Intensity Reduction in Large-Scale Non-Residential Buildings by Dynamic Control of HVAC with Heat Pumps

One of the main elements for increasing energy efficiency in large-scale buildings is identified in the correct management and control of the Heating Ventilation and Air Conditioning (HVAC) systems, particularly those with Heat Pumps (HPs). The present study aimed to evaluate the perspective of energy savings achievable with the implementation of an optimal control of the HVAC with HPs. The proposed measures involve the use of a variable air volume system, demand-controlled ventilation, an energy-aware control of the heat recovery equipment, and an improved control of the heat pump and chiller supply water temperature. The analysis has been applied to an academic building located in Pisa and is carried out by means of dynamic simulation. The achieved energy saving can approach values of more than 80% if compared with actual plants based on fossil fuel technologies. A major part of this energy saving is linked to the use of heat pumps as thermal generators as well as to the implementation of an energy efficient ventilation, emphasizing the importance of such straightforward measures in reducing the energy intensity of large-scale buildings.

Development and evaluation of data-driven controls for residential smart thermostats

The advent of smart thermostats with real-time sensing raises the question of how to preemptively control heating, ventilation, and air conditioning (HVAC) systems to minimize energy usage while maintaining occupant comfort. To this end, we empirically compare a standard reactive deadband control to two new smart thermostat HVAC control methods: (1) a model-free reinforcement learning (RL) approach and (2) a novel model predictive control (MPC) method, whose solution is optimal with respect to its data-driven linear model. We evaluated the controls with 500 unique energy models of houses located in the United States. The models were modified to facilitate the short-term performance simulation required for residential HVAC systems. Overall, we found the MPC controller offers three distinct advantages over the RL and deadband methods: (1) MPC had the lowest average cost (defined as a custom weighted combination of runtime and comfort) of the evaluated controllers; (2) the MPC control’s linear model was able to reliably extrapolate from the sparse sample of training observations, thus enabling it to adapt quickly to recent data; and (3) in contrast to RL methods, MPC did not subject the houses or occupants to the discomfort of system exploration.

Experimental evaluation of model-free reinforcement learning algorithms for continuous HVAC control

Controlling heating, ventilation and air-conditioning (HVAC) systems is crucial to improving demand-side energy efficiency. At the same time, the thermodynamics of buildings and uncertainties regarding human activities make effective management challenging. While the concept of model-free reinforcement learning demonstrates various advantages over existing strategies, the literature relies heavily on value-based methods that can hardly handle complex HVAC systems. This paper conducts experiments to evaluate four actor-critic algorithms in a simulated data centre. The performance evaluation is based on their ability to maintain thermal stability while increasing energy efficiency and on their adaptability to weather dynamics. Because of the enormous significance of practical use, special attention is paid to data efficiency. Compared to the model-based controller implemented into EnergyPlus, all applied algorithms can reduce energy consumption by at least 10% by simultaneously keeping the hourly average temperature in the desired range. Robustness tests in terms of different reward functions and weather conditions verify these results. With increasing training, we also see a smaller trade-off between thermal stability and energy reduction. Thus, the Soft Actor Critic algorithm achieves a stable performance with ten times less data than on-policy methods. In this regard, we recommend using this algorithm in future experiments, due to both its interesting theoretical properties and its practical results.

Formalized control logic fault definition with ontological reasoning for air handling units

Control logic programs determine the behavior of Heating, Ventilation, and Air Conditioning (HVAC) systems under different operating conditions. Faults in control logic account for more than 15% of all HVAC system problems, causing energy waste and occupancy discomfort. The first step towards systematically detecting and diagnosing control logic faults is to have an unambiguous control logic fault definition so that the existence of control logic faults in an HVAC system can be discovered from its operational data. In order to have high level of fault detection accuracy, the control logic fault definition needs to be customized for different HVAC systems as they have different components and control sequences. In this paper, we propose an object-oriented classification approach to systematically define customized control logic faults in terms of control logic input/output variable expressions. Focusing on air handling unit (AHU) systems, and their general operation objectives of energy efficiency and occupancy comfort, we elaborated four control goals and developed corresponding reasoning mechanisms to derive fault definitions. We also developed an HVAC component and control information ontology to be used in these reasoning mechanisms by extending existing HVAC information models. The prototype of the developed approach was tested with 27 common AHUs specified by the American Society of Heating, Refrigerating and Air-Conditioning Engineers (ASHRAE), and the results show that using the developed approach, it is possible to define a customized set of control logic faults applicable to each specific AHU with an overall precision of 94.2% and recall of 83.0%. This demonstrates the generality of our proposed approach in providing customized control logic fault definitions for different types of AHUs.

Evaluating Meta-Reinforcement Learning through a HVAC Control Benchmark (Student Abstract)

Meta-Reinforcement Learning (RL) algorithms promise to leverage prior task experience to quickly learn new unseen tasks. Unfortunately, evaluating meta-RL algorithms is complicated by a lack of suitable benchmarks. In this paper we propose adapting a challenging real-world heating, ventilation and air-conditioning (HVAC) control benchmark for meta-RL. Unlike existing benchmark problems, HVAC control has a broader task distribution, and sources of exogenous stochasticity from price and weather predictions which can be shared across task definitions. This can enable greater differentiation between the performance of current meta-RL approaches, and open the way for future research into algorithms that can adapt to entirely new tasks not sampled from the current task distribution.