Modeling of a sorption heat storage reactor using nonlinear autoregressive neural networks

Luca Scapino, Herbert Zondag, Jan Diriken, Camilo Rindt, Adriano Sciacovelli

Research output: Chapter in Book/Report/Conference proceedingConference contributionAcademicpeer-review

Abstract

Sorption thermal energy storage has the potential to store thermal energy over a long time with a higher energy density and less thermal losses compared to other technologies. In terms of modeling, sorption reactors are commonly described by physics-based models encompassing complex nonlinear phenomena occurring in the reactor. From a system modeling perspective, the use of data-driven models can be beneficial in cases where experimental data or high-fidelity data from more complex models are available, and a low computational cost with an acceptable accuracy is desired. The aim of this work is to investigate the capabilities of data-driven models based on two neural networks for modeling an open sorption reactor. The model takes as inputs the inlet temperature and sorbate concentration, and gives as outputs the reactor state of charge (SOC) and outlet temperature (TOUT). To account also for the thermal inertia of heat taking place in the reactor, both outputs are estimated with nonlinear autoregressive neural networks with exogenous inputs (NARXn), which account for the past n model outputs to determine the next output. Three neural network models are analyzed and several test cases are investigated to compare the performance of these neural network models with a high-fidelity CFD model. The results show that, for the SOC estimation, the NARX10 mean squared error (MSE) with respect to the high-fidelity CFD model was approximately two orders of magnitude smaller compared to the NARX1 MSE, resulting in a higher prediction accuracy. On the other hand, using the NARX10 architecture also for the TOUT estimation decreased the accuracy of TOUT estimations compared to a simpler FFNN neural network architecture considered in this work.
LanguageEnglish
Title of host publicationEurotherm Seminar #112
Subtitle of host publicationAdvances in Thermal Energy Storage
Place of PublicationLleida
Pages849-859
Number of pages11
ISBN (Electronic)978-84-9144-155-7
StatePublished - 17 May 2019
EventEurotherm Seminar #112: Advances in Thermal Energy Storage - University of Lleida, Lleida, Spain
Duration: 15 May 201917 May 2019
http://eurotherm.udl.cat/#Home

Conference

ConferenceEurotherm Seminar #112
CountrySpain
CityLleida
Period15/05/1917/05/19
Internet address

Fingerprint

Heat storage
Sorption
Neural networks
Thermal energy
Computational fluid dynamics
Network architecture
Energy storage
Physics

Keywords

  • Sorption heat storage; Artificial Neural Networks; Energy Efficiency

Cite this

Scapino, L., Zondag, H., Diriken, J., Rindt, C., & Sciacovelli, A. (2019). Modeling of a sorption heat storage reactor using nonlinear autoregressive neural networks. In Eurotherm Seminar #112: Advances in Thermal Energy Storage (pp. 849-859). Lleida.
Scapino, Luca ; Zondag, Herbert ; Diriken, Jan ; Rindt, Camilo ; Sciacovelli, Adriano. / Modeling of a sorption heat storage reactor using nonlinear autoregressive neural networks. Eurotherm Seminar #112: Advances in Thermal Energy Storage. Lleida, 2019. pp. 849-859
@inproceedings{fc7de264d69a4ff19bb3c8af880b1aa6,
title = "Modeling of a sorption heat storage reactor using nonlinear autoregressive neural networks",
abstract = "Sorption thermal energy storage has the potential to store thermal energy over a long time with a higher energy density and less thermal losses compared to other technologies. In terms of modeling, sorption reactors are commonly described by physics-based models encompassing complex nonlinear phenomena occurring in the reactor. From a system modeling perspective, the use of data-driven models can be beneficial in cases where experimental data or high-fidelity data from more complex models are available, and a low computational cost with an acceptable accuracy is desired. The aim of this work is to investigate the capabilities of data-driven models based on two neural networks for modeling an open sorption reactor. The model takes as inputs the inlet temperature and sorbate concentration, and gives as outputs the reactor state of charge (SOC) and outlet temperature (TOUT). To account also for the thermal inertia of heat taking place in the reactor, both outputs are estimated with nonlinear autoregressive neural networks with exogenous inputs (NARXn), which account for the past n model outputs to determine the next output. Three neural network models are analyzed and several test cases are investigated to compare the performance of these neural network models with a high-fidelity CFD model. The results show that, for the SOC estimation, the NARX10 mean squared error (MSE) with respect to the high-fidelity CFD model was approximately two orders of magnitude smaller compared to the NARX1 MSE, resulting in a higher prediction accuracy. On the other hand, using the NARX10 architecture also for the TOUT estimation decreased the accuracy of TOUT estimations compared to a simpler FFNN neural network architecture considered in this work.",
keywords = "Sorption heat storage; Artificial Neural Networks; Energy Efficiency",
author = "Luca Scapino and Herbert Zondag and Jan Diriken and Camilo Rindt and Adriano Sciacovelli",
year = "2019",
month = "5",
day = "17",
language = "English",
isbn = "978-84-9144-155-7",
pages = "849--859",
booktitle = "Eurotherm Seminar #112",

}

Scapino, L, Zondag, H, Diriken, J, Rindt, C & Sciacovelli, A 2019, Modeling of a sorption heat storage reactor using nonlinear autoregressive neural networks. in Eurotherm Seminar #112: Advances in Thermal Energy Storage. Lleida, pp. 849-859, Eurotherm Seminar #112, Lleida, Spain, 15/05/19.

Modeling of a sorption heat storage reactor using nonlinear autoregressive neural networks. / Scapino, Luca; Zondag, Herbert; Diriken, Jan; Rindt, Camilo; Sciacovelli, Adriano.

Eurotherm Seminar #112: Advances in Thermal Energy Storage. Lleida, 2019. p. 849-859.

Research output: Chapter in Book/Report/Conference proceedingConference contributionAcademicpeer-review

TY - GEN

T1 - Modeling of a sorption heat storage reactor using nonlinear autoregressive neural networks

AU - Scapino,Luca

AU - Zondag,Herbert

AU - Diriken,Jan

AU - Rindt,Camilo

AU - Sciacovelli,Adriano

PY - 2019/5/17

Y1 - 2019/5/17

N2 - Sorption thermal energy storage has the potential to store thermal energy over a long time with a higher energy density and less thermal losses compared to other technologies. In terms of modeling, sorption reactors are commonly described by physics-based models encompassing complex nonlinear phenomena occurring in the reactor. From a system modeling perspective, the use of data-driven models can be beneficial in cases where experimental data or high-fidelity data from more complex models are available, and a low computational cost with an acceptable accuracy is desired. The aim of this work is to investigate the capabilities of data-driven models based on two neural networks for modeling an open sorption reactor. The model takes as inputs the inlet temperature and sorbate concentration, and gives as outputs the reactor state of charge (SOC) and outlet temperature (TOUT). To account also for the thermal inertia of heat taking place in the reactor, both outputs are estimated with nonlinear autoregressive neural networks with exogenous inputs (NARXn), which account for the past n model outputs to determine the next output. Three neural network models are analyzed and several test cases are investigated to compare the performance of these neural network models with a high-fidelity CFD model. The results show that, for the SOC estimation, the NARX10 mean squared error (MSE) with respect to the high-fidelity CFD model was approximately two orders of magnitude smaller compared to the NARX1 MSE, resulting in a higher prediction accuracy. On the other hand, using the NARX10 architecture also for the TOUT estimation decreased the accuracy of TOUT estimations compared to a simpler FFNN neural network architecture considered in this work.

AB - Sorption thermal energy storage has the potential to store thermal energy over a long time with a higher energy density and less thermal losses compared to other technologies. In terms of modeling, sorption reactors are commonly described by physics-based models encompassing complex nonlinear phenomena occurring in the reactor. From a system modeling perspective, the use of data-driven models can be beneficial in cases where experimental data or high-fidelity data from more complex models are available, and a low computational cost with an acceptable accuracy is desired. The aim of this work is to investigate the capabilities of data-driven models based on two neural networks for modeling an open sorption reactor. The model takes as inputs the inlet temperature and sorbate concentration, and gives as outputs the reactor state of charge (SOC) and outlet temperature (TOUT). To account also for the thermal inertia of heat taking place in the reactor, both outputs are estimated with nonlinear autoregressive neural networks with exogenous inputs (NARXn), which account for the past n model outputs to determine the next output. Three neural network models are analyzed and several test cases are investigated to compare the performance of these neural network models with a high-fidelity CFD model. The results show that, for the SOC estimation, the NARX10 mean squared error (MSE) with respect to the high-fidelity CFD model was approximately two orders of magnitude smaller compared to the NARX1 MSE, resulting in a higher prediction accuracy. On the other hand, using the NARX10 architecture also for the TOUT estimation decreased the accuracy of TOUT estimations compared to a simpler FFNN neural network architecture considered in this work.

KW - Sorption heat storage; Artificial Neural Networks; Energy Efficiency

M3 - Conference contribution

SN - 978-84-9144-155-7

SP - 849

EP - 859

BT - Eurotherm Seminar #112

CY - Lleida

ER -

Scapino L, Zondag H, Diriken J, Rindt C, Sciacovelli A. Modeling of a sorption heat storage reactor using nonlinear autoregressive neural networks. In Eurotherm Seminar #112: Advances in Thermal Energy Storage. Lleida. 2019. p. 849-859.