CS Paper Radar

Physics-informed neural control of heat flow rate in a counter-flow heat exchanger: modeling, identification and experimental validation

Konstantinos Skantzikas, Emmanuel Witrant, Bojan Mavkov

eess.SY ·submitted 2026-10-04 17:07 UTC · v1

Abstract

The advanced operation of heat exchangers plays a critical role in the development of efficient and sustainable multivector energy systems. This paper proposes a hybrid physics-informed and data-driven framework for the analysis, identification, and regulation of transient heat transfer in counter-flow heat exchangers. A parameterized physical model combines classical heat-transfer relations with transport equations to capture convective propagation, transport-induced delays, and nonlinear effects associated with flow-rate variations. The heat-transfer coefficient is identified from experimental data, enabling the characterization of flow-dependent heat-transfer dynamics over a broad operating range. The identified model is then used to develop a hierarchical real-time control strategy. An inner loop regulates the volumetric flow rate using a high-order data-driven linear model, while an outer loop regulates the exchanged heat power using a machine-learning controller initially trained on the physical model and subsequently refined with experimental data. Experimental results demonstrate accurate reproduction of transient heat-transfer dynamics and effective heat-flow-rate regulation under varying thermal boundary conditions (identification fit above 94\% in normalized root-mean-square error and total variation of the flow rate reduced by half compared to a proportional-integral controller). The proposed framework combines physical interpretability with data-driven adaptation for the analysis and real-time operation of dynamic heat-transfer systems.

Closest papers

Nearest neighbours by meaning across the whole index.

Control-Oriented Modeling of Heat Exchangers with Improved Low-Flow Dynamics

eess.SY ·2026-10-30 ·Noah Hofmann, Stefan L. Hölzl, Robert Bauer, Martin Steinberger
similarity 0.70
Abstract · 6 pages

Existing low-order models of heat exchangers can lose dynamic accuracy at low flow rates, when fluid residence times become significant. This paper proposes two control-oriented models that improve the low-flow performance. The first extends a finite-volume formulation with the first spatial temperature moment, whereas the second uses Chebyshev--Lobatto collocation. Both capture the coupled heat-transfer and transport dynamics and improve the performance when the system behavior is dominated by the inlet-temperature dynamics. At equal state dimension, comparisons with a high-order reference show more accurate delayed outlet-temperature responses while retaining steady-state accuracy.

0.22

Model Predictive Control for Dynamic Hydraulic Balancing in Building Radiator Heating Systems: Modeling, Design, and Experimental Validation

eess.SY math.OC ·2026-09-08 ·Erfan Shakhesi, Maurice Heemels, Paula Chanfreut
similarity 0.61
Abstract

Hydronic radiator systems are among the most widely used heating systems in buildings. In the literature, radiator heat outputs are often assumed to be independent of one another and arbitrarily adjustable in time when designing controllers. However, in practice, radiators are supplied by one or multiple common heat sources and are hydraulically coupled through the water circulation system. Maintaining occupant comfort in multi-zone buildings therefore requires not only an appropriate supply temperature but also proper hydraulic balancing to distribute the available water flow according to the heating demand of each zone. To this end, we develop a grey-box thermal model that captures the hydraulic interactions among radiators and the effects of radiator valves, circulation pumps, and heat sources. In a real building, we show that accounting for hydraulic interactions reduces the root mean square error (RMSE) between the measured and modeled zone temperatures by approximately 11% compared with a model that neglects these interactions. Additionally, we integrate the developed model into a model predictive control (MPC) framework for dynamic hydraulic balancing that jointly optimizes valve openings and the supply temperature to maintain thermal comfort in each zone while reducing energy consumption. Through both real-world experiments and numerical case studies, we demonstrate that, compared with existing MPC formulations that neglect hydraulic interactions or do not control radiator valves, the proposed MPC reduces comfort-range violations by at least 27% while requiring a similar or even lower cumulative supply temperature.

0.40

A sensor-driven physics-informed neural network for real-time occupant-centric space heating control

cs.LG eess.SY ·2026-11-01 ·Yunfei Bai, Chenghao Li, Sunku Prasad Jenne, Hao Wang, Sizhe Zhang +1 more
similarity 0.60
Abstract

Conventional space heating systems typically rely on single or sparse fixed-point temperature measurements, overlooking non-uniform indoor temperature distributions, transient thermal dynamics, and occupant-specific comfort requirements. This study explores an occupant-centric control strategy that aims to maintain the desired temperature around the occupant location. Spatially and temporally varying temperature fields, together with occupant mobility, make real-time local temperature estimation challenging. To address this challenge, we propose a control-oriented Physics-Informed Neural Network (PINN) model for indoor temperature-field prediction. Conventional PINNs rely on explicit time inputs and assume consistent temporal patterns across samples, which limits adaptability under variable initial conditions and heating sequences. The proposed model infers temporal dynamics implicitly from spatial coordinates and limited sensor measurements, improving robustness to changes in initial conditions and external disturbances. The method was validated across nine horizontal sensing planes in three controlled room-scale configurations. Against a validated COMSOL Multiphysics simulation, the PINN predicted the evaluated temperature fields with a maximum relative error below 1.85% across validation cases. These predictions supported an occupant-centric space heating control scheme. Compared with COMSOL-based offline control, the PINN-based system achieved an average prediction error of around 0.67%, and its rapid inference supports real-time occupant-centric control implementation. This work represents a move away from reliance on fixed-point temperature measurements in space heating control to an occupant-centric control strategy informed by real-time environmental feedback.

0.24

Virtual Temperature Sensors in Power Transformers Using Neural Ordinary Differential Equations

cs.LG ·2026-08-13 ·Berk Hadzhamolla, Alexander Johannes Stasik, Signe Riemer-Sørensen
published similarity 0.60
Abstract

Accurate modeling and forecasting of power transformer thermal behavior are critical for reliability, asset lifetime, and optimized power system operation. Numerical approaches such as finite element methods (FEM) and computational fluid dynamics (CFD) offer high fidelity but are computationally expensive, require complex mesh generation, and are often impractical for real-time or large-scale applications, particularly when transformer geometries are unknown. Lumped-parameter thermal models are more practical but depend on transformer-specific thermal constants and may fail to capture dynamic responses under varying operating and environmental conditions. Purely data-driven machine learning methods, including artificial neural networks, convolutional neural networks, and long short-term memory (LSTM) networks, have shown success in forecasting transformer temperatures but typically require large volumes of high-quality training data and may produce physically inconsistent or uninterpretable results. This paper develops a physics-aware Neural Ordinary Differential Equation (Neural ODE) framework for forecasting transformer thermal behavior from real-world time-series data. Neural ODEs model system dynamics in continuous time, providing smooth trajectory prediction and a natural representation of continuously evolving thermal dynamics. A key contribution is the integration of simplified heat-transfer equations directly into the Neural ODE formulation. The model is evaluated across datasets from fifteen transformers in different regions of Norway with varying designs and cooling mechanisms. The results demonstrate that the developed Neural ODE framework provides a standardized, physics-aware, and robust forecasting approach for heterogeneous transformer units.

0.58

Model Predictive Control of Heating Systems via Koopman-Operator Bilinearization

eess.SY ·2026-11-02 ·Paula Chanfreut, Erfan Shakhesi, Mircea Lazar, Maurice Heemels
similarity 0.57
Abstract

Hydronic heat emitters are widely used in buildings, but their control is challenging due to nonlinear dynamics, arising from both the heat transfer in each emitter and the hydraulic coupling among those that share a circulating pump. These nonlinearities are often simplified or neglected in the literature. In this paper, we use Koopman theory to make the resulting model more tractable for control. In particular, we propose an input reparameterization and construct an infinite-dimensional set of observables that yields an exact bilinear representation of the system in the chosen lifted space. We then show that truncating the infinite-dimensional representation leads to bounded approximation errors that decrease as the truncation order increases, as well as bounded trajectory deviations over a finite horizon. The resulting bilinear model is incorporated into a model predictive control (MPC) framework to provide thermal comfort in multi-zone buildings while reducing energy use. Using a numerical case study of a five-story building with 20 zones, we show that the proposed Koopman-based MPC achieves a realized performance cost close to that of nonlinear MPC with faster solve times, and outperforms MPC formulations based on time-invariant and time-varying linearizations.

0.27

Model-Free Control for Residential Heating: Deployment and Simulation of Nonlinear Data-Enabled Predictive Control

eess.SY math.OC ·2026-09-18 ·Sebastian Zieglmeier, Chris Verhoek, Jaap Eising, Mathias Hudoba de Badyn
similarity 0.57
Abstract · 19 pages, 8 figures, 7 tables

Residential heating accounts for a large share of building energy use, and predictive control strategies can reduce it by anticipating rather than reacting to the room temperature alone. The usual manner of implementing predictive control, model predictive control (MPC), requires an accurate model of each individual residential unit for heating control. Obtaining this model is time-consuming to obtain manually or even impossible, and such models do not transfer across a heterogeneous building stock. Data-enabled predictive control (DeePC) removes this modeling step by designing the controller directly using measured trajectories of the system. The foundations of DeePC, however, are built on the class of deterministic linear time-invariant systems, which is an unrealistic assumption for residential heating. This work applies three recently proposed nonlinear extensions of DeePC, collectively referred to as data-driven predictive control (DPC): Select-DPC, gain-scheduling DPC, and linear parameter-varying DPC. These are compared against standard DeePC and the hysteresis controller, the industry standard in residential heating. The comparison runs over a full heating season on a calibrated digital twin of an occupied research unit, the NEST research building in Switzerland, assessed by heating energy and comfort-band violation. Moreover, gain-scheduling DPC was deployed on the real residential apartment and validated the conclusions from the simulations. Each of the DPC controllers consumes significantly less energy than the hysteresis controller, amounting to roughly $11\%$ over the season, validating the use of DPC. The nonlinear methods further outperform linear DeePC in all assessed scenarios, and we analyze the relative costs and benefits of each nonlinear method.

0.41

Temperature-Dependent Multiphysics Modeling of Additive Friction Stir Deposition Using Multi-Task Coupled Physics-Informed Neural Networks

cs.CE physics.comp-ph ·2026-10-02 ·Dhrubajyoti Gupta, Nikhil Gotawala, Raghav Gnanasambandam, Rohit Kannan, Jian Yu +2 more
benchmark similarity 0.56
Abstract · 14 pages, 12 figures, 7 tables

Additive friction stir deposition (AFSD) involves coupled thermal and material-flow fields governed by frictional heating, plastic deformation, and tool-imposed boundary conditions. High-fidelity finite-volume methods (FVMs) can resolve these fields but are computationally expensive, while constant-property approximations can distort the predicted thermo-mechanical response. This work develops a steady-state multi-task coupled physics-informed neural network (MCoPINN) to predict three-dimensional temperature and material-flow fields while reconstructing temperature-dependent thermal conductivity, density, and specific heat from sparse property measurements. A theoretical analysis decomposes prediction error into property reconstruction and neural field-solver contributions and characterizes the discrepancy associated with constant-property models. A controlled one-dimensional nonlinear-conduction study evaluates property reconstruction under sparse measurements. The framework is then evaluated for AFSD against a temperature-dependent FVM benchmark and experimental thermocouple measurements. MCoPINN reproduces the reference thermal and material-flow fields and reduces average temperature root-mean-square error from 50.13 K for the constant-property CoPINN to 30.28 K across four traverse speeds. MCoPINN requires approximately 8.5 hours of training per condition, compared with 52 hours for the FVM benchmark. These results demonstrate computationally efficient, physics-coupled prediction with temperature-dependent material properties.

0.34

Differentiable Hybrid-Action Neural Feedback Control for District Heating Networks

eess.SY ·2026-10-01 ·Nicolas Kirsch, Corrado Sgadari, Alessio La Bella, Giancarlo Ferrari-Trecate
similarity 0.56
Abstract

Many cyber-physical systems require control policies that combine continuous setpoints with discrete operational de- cisions, such as equipment switching, mode selection, or resource scheduling. Discrete actions are not differentiable, which ob- structs gradient-based policy training, while conventional mixed- integer formulations remain costly to solve online. This paper proposes a hybrid-action neural controller (HANC) in which a continuous branch, a categorical branch and a differentiable assembly layer jointly generate commands that satisfy complex actuator constraints by construction. Categorical decisions are handled using a straight-through Gumbel estimator, enabling the policy to be trained by backpropagation through time over full closed-loop rollouts. The proposed framework is deployed on a district heating network (DHN) featuring multiple heat generation units and stratified thermal energy storage. Its performance is evaluated on a simulation of a real DHN located at RSE SpA in Italy. The resulting policy jointly learns switching decisions and continu- ous operating setpoints. Under dynamic electricity pricing, the learned controller reduces operating cost by 30% compared to a rule-based industrial baseline. We also show that, compared with a deterministic straight-through relaxation, injecting noise during training achieves similar cost while reducing hard switching by an order of magnitude, and attribute this difference to the wider decision margins of the resulting policy.

0.24