• Energy Research

Abstract The integration of renewable energy sources into building energy systems and the progressive coupling between the thermal and electrical domains makes the analysis of these systems increasingly complex. At the same time, however, more and more building monitoring data is being collected. The manual evaluation of this data is time-consuming and requires expert knowledge. Hence, there is a strong need for tools that enable the automatic knowledge extraction from these huge data sets to support system integrators and favor the development of smart energy services, e.g., predictive maintenance. One crucial step in knowledge extraction is the detection of change points and hidden states in measurements. In this work, we present a tool for automated detection of hidden operation modes based on multivariate time series data deploying motif-aware state assignment (MASA). The tool is evaluated utilizing measurements of a heat pump and compared to two baseline algorithms, namely k-Means and k-Medoids. MASA performs particularly well on noisy data, where it shows only a small deviation in the number of detected change points compared to the ground truth with up to 77% accuracy. Furthermore, it almost always outperforms the baseline algorithms, which in turn require extensive preprocessing.

This paper describes the methodology developed and the calculation steps used to evaluate the energy efficiency potential of office buildings. The methodology enables a detailed analysis of retrofit options for the building envelope and its energy supply system. Different simplification measures accelerate the data acquisition process for office building stock owners and allow a data handling according to the existing building information, thus enabling office building structures to be prompted to design typical building constructions. We implement solutions enabling both a time-saving accelerated data input for office buildings and the handling of incomplete data. An automated calculation of the most common refurbishment measures allows a comparison of up to 64 combinations of measures, the illustration of energy and CO2 savings, and an economic evaluation. The latter takes into account the time value of money, the uncertainty of future energy prices, and the possibility of delaying an investment. To this end, a net present value analysis and a real options analysis are implemented, enabling a comparison of retrofit alternatives with different initial and future cash flows both for buildings occupied by the investor (owner-occupier perspective) and for rented buildings (tenant perspective). Energy price scenarios as well as a Monte Carlo simulation account for the uncertainty in energy price trends. For a university building used as a test case, the simplified and time-saving data input methods were successfully tested and an automated evaluation of 64 typical retrofit combinations carried out. The results of the energy, ecological and economic efficiency evaluation shows that a generally preferred retrofit option cannot always be identified. Specifically, for the test case, the best-rated economic refurbishment possibility leads to the largest increase in final energy demand amongst all options considered, which points out the necessity of a multi-criteria evaluation.

AbstractSolar thermal systems without appropriate control strategies are often working inefficiently despite high quality components and suitable hydraulic concepts. In this paper different control strategies for solar cooling systems are compared and evaluated according to their exergetic efficiency. Therefore, a Software-In-The-Loop test bench is used. The original controller of a solar thermal system is connected to a building simulation model. The simulation model is created with the modeling language Modelica using the simulation environment Dymola. This enables testing and monitoring of various control strategies. These controlling strategies are evaluated according to the energetic as well as exergetic output. Models for different parts of the hydraulic system are set up and validated. These different components are combined to a virtual solar test bench. The virtual test bench is connected to the system controller using Ethernet connection and an adapting program which has been created in C- Code (Software-In-The-Loop). Thus it is possible to use the advantages of a simulation environment, like performing repeatable tests with specified weather conditions, without implementing the control algorithms of the devices into Dymola. In a test scenario, two different control strategies for a solar cooling system have been implemented. First, a strategy to maximize the energetic output of the solar collectors is applied (Maximum Yield Strategy). Thereby the flow temperature of the solar system is kept as low as possible in order to increase the collectors’ efficiency. Second, the flow temperature is adapted to the required temperature of the absorption chiller, which leads to a lower energetic output of the collector and to a higher temperature level of storage (High Temperature Strategy). This paper shows that Software-In-The-Loop tests can be used for the optimization of control strategies. In the test scenario, the High Temperature Strategy leads to lower energy yields due to lower solar collector efficiency, whilst the exergy output and the output of cooling energy of the absorption chiller can be significantly increased.

Abstract A variety of novel, recyclable and reusable, construction materials has already been studied within literature during the past years, aiming at improving the overall energy efficiency ranking of the building envelope. However, several studies show that a delicate control of indoor climating elements can lead to a significant performance improvement by exploiting the building’s savings potential via smart adaptive HVAC regulation to exogenous uncertain disturbances (e.g. weather, occupancy). Building Optimization and Control (BOC) systems can be categorized into two different groups: centralized (requiring high data transmission rates at a central node from every corner of the overall system) and decentralized 1 (assuming an intercommunication among neighboring constituent systems). Moreover, both approaches can be further divided into two subcategories, respectively: model-assisted (usually introducing modeling oversimplifications) and model-free (typically presenting poor stability and very slow convergence rates). This paper presents the application of a novel, decentralized, agent-based , model-free BOC methodology (abbreviated as L4GPCAO) to a modern non-residential building (E.ON. Energy Research Center’s main building), equipped with controllable HVAC systems and renewable energy sources by utilizing the existing Building Management System (BES). The building testbed is located inside the RWTH Aachen University campus in Aachen, Germany. A combined rule criterion composed of the non-renewable energy consumption (NREC) and the thermal comfort index – aligned to international comfort standards – was adopted in all cases presented herein. Besides the limited availability of the specified building testbed, real-life experiments demonstrated operational effectiveness of the proposed approach in BOC applications with complex, emerging dynamics arising from the building’s occupancy and thermal characteristics. L4GPCAO outperformed the control strategy that was designed by the planers and system provider, in a conventional manner, requiring no more than five test days.

Abstract With increasing complexity of building energy systems and rising shares of renewable energies in the grids, the requirements for building automation and control systems (BACS) are increasing. The use of storage systems enables the decoupling of energy demand and supply and to consider dynamic constraints in the control of the systems. The resulting optimization problem is very challenging to solve with the state-of-the-art rule-based-control (RBC) approach. Model Predictive Control (MPC) on the other hand allows a nearly optimal operation but comes with expensive modeling efforts and high computational costs. These drawbacks are contrasted by promising results from the field of Reinforcement Learning (RL). RL can be model-free, is highly adaptive and learns a policy by interacting with the controlled system. However, the literature also addresses a number of questions, to be answered before RL for BACS can be realized. One is the slow convergence of the training process, which makes the application of a pre-training strategy necessary. Therefore, we design and compare different pre-training work-flows for a real-world energy system, in a demand response scenario. We apply a data-driven approach, covering all aspects from raw monitoring data to the trained algorithm. The considered energy system consists of two compression chillers and an ice storage. The objective of the control task is to charge and discharge the storage with respect to dynamic constraints. We use machine learning models of the energy system to train and evaluate a state-of-the-art RL algorithm (DQN) under five different pre-training strategies. We compare, online and offline training and initialization of the RL controller together with a guiding RBC. We demonstrate that offline training with a guiding RBC provides stable learning and a RL controller that always outperforms this guiding RBC. Unguided exploration on the other hand leads to higher accumulated cost savings. Based on our findings, we derive recommendations for practical application and future research questions.

AbstractModel predictive control (MPC) is a promising approach for optimizing the performance of borehole heat exchangers (BHEs) in ground-source heat pump systems. The central element of MPC is the forward model that predicts the thermal dynamics in the ground. In this work, we validate the prediction accuracy of four BHE modeling approaches against real-world measurement data across various operational events and timescales. We simulate the fluid temperature leaving a BHE using a fully discretized 3-D numerical model, a resistance–capacitance model, a g-function model, and a hybrid model. The simulated temperatures are compared to measured temperatures using three validation metrics that quantify temperature offset, noise, and accuracy. The main reason for a mismatch between measured and modeled temperatures is a temperature offset of the simulated temperature. To remove this effect, the models were calibrated for their most sensitive parameter, the ground temperature, and their prediction accuracy over 4 years was evaluated. Thereby, model calibration seems to be a viable solution to account for an unknown load history. The results show that the resistance–capacitance model provides decent predictions in the short term and the g-function model in the long term. However, both models are strongly dependent on accurate calibration. The hybrid model provides the most accurate short and long-term predictions and is less dependent on calibration. Still, its integration into optimization syntax poses challenges compared to the other models. Although not yet applied in model predictive control, the hybrid model stands out as a promising choice for optimizing BHE field operations across various timescales.

Modern literature exhibits numerous centralized control approaches—event-based or model assisted—for tackling poor energy performance in buildings. Unfortunately, even novel building optimization and control (BOC) strategies commonly suffer from complexity and scalability issues as well as uncertain behavior as concerns large-scale building ecosystems—a fact that hinders their practical compatibility and broader applicability. Moreover, decentralized optimization and control approaches trying to resolve scalability and complexity issues have also been proposed in literature. Those approaches usually suffer from modeling issues, utilizing an analytically available formula for the overall performance index. Motivated by the complications in existing strategies for BOC applications, a novel, decentralized, optimization and control approach—referred to as Local for Global Parameterized Cognitive Adaptive Optimization (L4GPCAO)—has been extensively evaluated in a simulative environment, contrary to previous constrained real-life studies. The current study utilizes an elaborate simulative environment for evaluating the efficiency of L4GPCAO; extensive simulation tests exposed the efficiency of L4GPCAO compared to the already evaluated centralized optimization strategy (PCAO) and the commercial control strategy that is adopted in the BOC practice (common reference case). L4GPCAO achieved a quite similar performance in comparison to PCAO (with 25% less control parameters at a local scale), while both PCAO and L4GPCAO significantly outperformed the reference BOC practice.

36. International Conference on Efficiency, Cost, Optimization, Simulation and Environmental Impact of Energy System, ECOS 2023, Las Palmas de Gran Canaria, Spain, 25 Jun 2023 - 30 Jun 2023; Renewable energy 236, 121409 (2024). doi:10.1016/j.renene.2024.121409 special issue: "Selected Papers from ECOS 2023 in Renewable Energy : special issue / edited by Dr. Ana María Blanco Marigorta, Dr. Noemi Melián Martel, Dr.-Ing. Beatriz Del Rio Gamero" Published by Elsevier Science, Amsterdam [u.a.]

Abstract With the increasing use of volatile renewable energies, the requirements for building automation and control systems (BACS) are increasing. Load shifting within local energy systems stabilizes fluctuations in the grid and can be triggered by price signals. The energy purchase can thus be considered and solved as an optimal control problem. Classical approaches, often based on the optimization of mathematical models, are uneconomical in many cases, due to the high effort involved in the model creation. Algorithms from the field of Reinforcement Learning (RL), on the other hand, have a high potential for the automation of energy system optimization, due to their model-free and data-driven characteristics. However, there is still a lack of studies that examine algorithms for BACS-related applications in a structured way. Therefore, we present a study, investigating the potential of two different RL algorithms for load shifting in a cooling supply system. We combine the benefits of Modelica, a powerful modeling language, with RL algorithms and demonstrate how generalized relationships and control decisions can be learned. The case study is modeled according to a cooling supply system in Berlin, Germany. The two different algorithms (DQN and DDPG) are used to control the operation parameters of a central compression chiller, with respect to a price signal. While real monitoring data are used as exogenous influences, the thermal dynamics of the cooling network are simulated. With the learned policies, flexibility in the network is used which leads on average to weekly cost savings of 14 %, compared to direct load coverage. Our results suggest that, under certain conditions, RL is a suitable alternative to established methods. However, we also acknowledge that there are still research questions to address before RL can be applied in real BACS.