Hesam Ahmadian Behrooz
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Hesam Ahmadian Behrooz has a total of 5 co-authors. Here are the top co-authors by number of shared publications:
R. Bozorgmehry Boozarjomehry
2 shared publications- Ramin Bozorgmehry Boozarjomehry
Department of Chemical and Petroleum Engineering, Sharif University of Technology, Tehran, Iran
2 shared publications Reza Kamalian Jahromi
2 shared publicationsMohammad Hoseini
1 shared publicationZohre Sattari
Chemical Engineering Faculty , 114739 Sahand University of Technology , Tabriz , Iran
1 shared publication
Latest Publications
- Open Access
- 6 July 2026
- Published by Elsevier in SSRN Electronic Journal
A thermodynamically consistent physics informed neural network framework for direct generation of binary Pxy VLE diagrams and automatic azeotrope identification
A framework of physics-informed neural network (PINN) for constructing Pxy vapor-liquid equilibrium (VLE) diagram and identifying azeotropes inherently without any flash calculations is proposed. Two networks – an Equilibrium State Network (ESN) for phase compositions and a Fugacity Coefficient Network (FCN) for vapor-phase thermodynamics – are included in the model structure and sequentially trained to ensure numerical stability. Vapor fugacity coefficients are used as primary neural outputs; on the other hand, residual Gibbs free energy, molar volume, and pressure are computed precisely using thermodynamic relations and automatic differentiation. Feasibility-preserving parameterizations for bounded mole fractions and material balances, together with azeotropic compositions and pressures as learning parameters, are considered in the design process. The Gibbs-Duhem consistency and volumetric feasibility are automatically satisfied by considering physics-based residuals including summation form fugacity equality, pure component relations, and ideal gas limits in the loss function. The model is successfully applied to several binary mixture systems, which include benzene/toluene, acetone/chloroform, ethanol/water, and methanol/n-C5, which involve ideal, strongly nonideal, and minimum/maximum pressure azeotropes. Smooth globally consistent Pxy diagrams are created, and azeotropes are recognized intrinsically without being predefined. High accuracy equivalent to that of the simulator is achieved, while the difference in the composition of solutions is minimal, proving the effectiveness of generalization using a physically guided thermodynamically-consistent structure.Moreover, the impact of the network capacity and the density of training samples was studied. Accuracy is consistently improved by network expansion and a non-monotonic effect is observed as data density is increased. While physical consistency and phase classification can be provided with enough training samples, overconstraint of the model near phase boundaries due to an excess number of samples reduces prediction accuracy for sensitive properties such as vapor fractions. Thus, the optimal number of training data samples is essential.
- 6 July 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessA thermodynamically consistent physics informed neural network framework for direct generation of binary Pxy VLE diagrams and automatic azeotrope identification
A framework of physics-informed neural network (PINN) for constructing Pxy vapor-liquid equilibrium (VLE) diagram and identifying azeotropes inherently without any flash calculations is proposed. Two networks – an Equilibrium State Network (ESN) for phase compositions and a Fugacity Coefficient Network (FCN) for vapor-phase thermodynamics – are included in the model structure and sequentially trained to ensure numerical stability. Vapor fugacity coefficients are used as primary neural outputs; on the other hand, residual Gibbs free energy, molar volume, and pressure are computed precisely using thermodynamic relations and automatic differentiation. Feasibility-preserving parameterizations for bounded mole fractions and material balances, together with azeotropic compositions and pressures as learning parameters, are considered in the design process. The Gibbs-Duhem consistency and volumetric feasibility are automatically satisfied by considering physics-based residuals including summation form fugacity equality, pure component relations, and ideal gas limits in the loss function. The model is successfully applied to several binary mixture systems, which include benzene/toluene, acetone/chloroform, ethanol/water, and methanol/n-C5, which involve ideal, strongly nonideal, and minimum/maximum pressure azeotropes. Smooth globally consistent Pxy diagrams are created, and azeotropes are recognized intrinsically without being predefined. High accuracy equivalent to that of the simulator is achieved, while the difference in the composition of solutions is minimal, proving the effectiveness of generalization using a physically guided thermodynamically-consistent structure.Moreover, the impact of the network capacity and the density of training samples was studied. Accuracy is consistently improved by network expansion and a non-monotonic effect is observed as data density is increased. While physical consistency and phase classification can be provided with enough training samples, overconstraint of the model near phase boundaries due to an excess number of samples reduces prediction accuracy for sensitive properties such as vapor fractions. Thus, the optimal number of training data samples is essential.
- Open Access
- 23 February 2026
- Published by Elsevier in SSRN Electronic Journal
Robust Design of Post-Combustion CO₂ Absorbers: An SVD-Based Controllability Analysis of MEA and PZ/MEA Solvents
Post-combustion CO2 capture with aqueous amines is recognized as technically mature, but high energy demand and sensitivity to disturbances remain important challenges. To address these challenges, a steady-state robustness framework was developed for an industrial absorber, and robustness was quantified via singular value decomposition (SVD) of the scaled steady-state gain matrix. The largest singular value (σ1) was used as a robustness index, and the associated right singular vector defined the worst-case disturbance direction; together, these metrics were used to determine the maximum allowable combined disturbances that preserve a specified minimum CO2 recovery. The disturbance set included flue-gas component flow rates and temperature, solvent circulation rate and composition, lean CO2 loading, solvent temperature, and column pressure. Across the operating space, the worst-case disturbance direction was dominated, in this order, by flue-gas CO2 flow rate, solvent flow rate, solvent MEA mass fraction, and lean CO2 loading.Parametric studies were performed over MEA concentration, lean loading, number of stages, pressure, and piperazine (PZ) addition, and clear trade-offs among separation efficiency, energy use, and robustness were identified. For MEA, an operating window near 30–35 wt% with lean loading of 0.20–0.25 mol CO2 per mol MEA was found to deliver 90–95% recovery with relatively low σ1. Increases in pressure or stage count reduced the required circulation rate but increased σ1, whereas adding approximately 2.5–10 wt% PZ reduced both circulation and σ1, enhancing disturbance tolerance. The proposed SVD-based procedure provides practical guidance for embedding robustness targets directly into steady-state design of amine absorbers.
- 23 February 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessRobust Design of Post-Combustion CO₂ Absorbers: An SVD-Based Controllability Analysis of MEA and PZ/MEA Solvents
Post-combustion CO2 capture with aqueous amines is recognized as technically mature, but high energy demand and sensitivity to disturbances remain important challenges. To address these challenges, a steady-state robustness framework was developed for an industrial absorber, and robustness was quantified via singular value decomposition (SVD) of the scaled steady-state gain matrix. The largest singular value (σ1) was used as a robustness index, and the associated right singular vector defined the worst-case disturbance direction; together, these metrics were used to determine the maximum allowable combined disturbances that preserve a specified minimum CO2 recovery. The disturbance set included flue-gas component flow rates and temperature, solvent circulation rate and composition, lean CO2 loading, solvent temperature, and column pressure. Across the operating space, the worst-case disturbance direction was dominated, in this order, by flue-gas CO2 flow rate, solvent flow rate, solvent MEA mass fraction, and lean CO2 loading.Parametric studies were performed over MEA concentration, lean loading, number of stages, pressure, and piperazine (PZ) addition, and clear trade-offs among separation efficiency, energy use, and robustness were identified. For MEA, an operating window near 30–35 wt% with lean loading of 0.20–0.25 mol CO2 per mol MEA was found to deliver 90–95% recovery with relatively low σ1. Increases in pressure or stage count reduced the required circulation rate but increased σ1, whereas adding approximately 2.5–10 wt% PZ reduced both circulation and σ1, enhancing disturbance tolerance. The proposed SVD-based procedure provides practical guidance for embedding robustness targets directly into steady-state design of amine absorbers.
- 1 December 2024
- Published by Walter de Gruyter GmbH in Chemical Product and Process Modeling
- Vol. 19 (6) , 989-1012
A control perspective on hybrid membrane/distillation propane/propylene separation process
This study explores the design, dynamic behavior, and controllability of a hybrid membrane-distillation process for propane/propylene separation, offering a novel comparison with traditional single distillation columns. Our key innovation lies in integrating hollow fiber membrane technology with distillation to significantly improve process efficiency, reducing energy consumption and operational costs. Using rigorous economic and dynamic simulations, the hybrid process demonstrates superior performance, achieving a 12.6 % reduction in total annual cost (TAC) compared to the conventional distillation system. Advanced control methodologies, including Singular Value Decomposition (SVD) and condition number analysis, are applied to assess and optimize the controllability of the hybrid structure under disturbances such as fluctuations in feed composition and flow rates. The results highlight the hybrid process’s ability to maintain high product purity and faster dynamic response, offering a robust solution for large-scale propane/propylene separation in the petrochemical industry. This work provides crucial insights into improving both operational efficiency and control robustness in industrial separation processes.
- 1 December 2024
- Published by Walter de Gruyter GmbH in Chemical Product and Process Modeling
- Vol. 19 (6) , 989-1012
A control perspective on hybrid membrane/distillation propane/propylene separation process
This study explores the design, dynamic behavior, and controllability of a hybrid membrane-distillation process for propane/propylene separation, offering a novel comparison with traditional single distillation columns. Our key innovation lies in integrating hollow fiber membrane technology with distillation to significantly improve process efficiency, reducing energy consumption and operational costs. Using rigorous economic and dynamic simulations, the hybrid process demonstrates superior performance, achieving a 12.6 % reduction in total annual cost (TAC) compared to the conventional distillation system. Advanced control methodologies, including Singular Value Decomposition (SVD) and condition number analysis, are applied to assess and optimize the controllability of the hybrid structure under disturbances such as fluctuations in feed composition and flow rates. The results highlight the hybrid process’s ability to maintain high product purity and faster dynamic response, offering a robust solution for large-scale propane/propylene separation in the petrochemical industry. This work provides crucial insights into improving both operational efficiency and control robustness in industrial separation processes.
- 1 December 2019
- Published by Elsevier in ISA Transactions
- Vol. 95, 93-109
Robust set-point optimization of inferential control system of crude oil distillation units
- 1 December 2019
- Published by Elsevier in ISA Transactions
- Vol. 95, 93-109
Robust set-point optimization of inferential control system of crude oil distillation units
- 7 June 2019
- Published by Walter de Gruyter GmbH in Chemical Product and Process Modeling
- Vol. 14 (4)
A Multiobjective Robust Approach for the Design of Natural Gas Transmission Pipelines
This paper presents a chance-constrained multiobjective optimization framework for the optimal design of gas transmission lines (GTL) in which the total annual cost (TAC) and operability index as the two conflicting objectives are minimized and maximized, respectively. The delivery flow rates are assumed to be random Gaussian variables since the customer demand can be uncertain in the design phase. Accordingly, a robustness measure is introduced in order to quantify the risk of the final design against delivery uncertainties. The proposed model is capable of determining the optimum design variables including the pipeline diameter and thickness along with the location of the compressor stations and their capacities. The effects of design parameters including pipe thickness, pipe diameter and customer demand uncertainty level on the design of a GTL are investigated from economic as well as the robustness point of view. It is shown that the design with the highest nominal pipe size (NPS) and least thickness possible is the optimal design when TAC is considered as the objective function while the design with both the highest NPS and thickness possible shows the highest robustness and flexibility against delivery demand fluctuations. The final decision about the optimal NPS and thickness should be made on the basis of the expected uncertainty of the customer future demand. The results also can quantitatively suggest the required over-design factor that must be considered in the design phase of the GTL for various compressor stations.
- 7 June 2019
- Published by Walter de Gruyter GmbH in Chemical Product and Process Modeling
- Vol. 14 (4)
A Multiobjective Robust Approach for the Design of Natural Gas Transmission Pipelines
This paper presents a chance-constrained multiobjective optimization framework for the optimal design of gas transmission lines (GTL) in which the total annual cost (TAC) and operability index as the two conflicting objectives are minimized and maximized, respectively. The delivery flow rates are assumed to be random Gaussian variables since the customer demand can be uncertain in the design phase. Accordingly, a robustness measure is introduced in order to quantify the risk of the final design against delivery uncertainties. The proposed model is capable of determining the optimum design variables including the pipeline diameter and thickness along with the location of the compressor stations and their capacities. The effects of design parameters including pipe thickness, pipe diameter and customer demand uncertainty level on the design of a GTL are investigated from economic as well as the robustness point of view. It is shown that the design with the highest nominal pipe size (NPS) and least thickness possible is the optimal design when TAC is considered as the objective function while the design with both the highest NPS and thickness possible shows the highest robustness and flexibility against delivery demand fluctuations. The final decision about the optimal NPS and thickness should be made on the basis of the expected uncertainty of the customer future demand. The results also can quantitatively suggest the required over-design factor that must be considered in the design phase of the GTL for various compressor stations.
- 1 November 2018
- Published by Elsevier in Fluid Phase Equilibria
- Vol. 475, 64-76
Application of the unscented transform in the uncertainty propagation of thermodynamic model parameters
- 1 November 2018
- Published by Elsevier in Fluid Phase Equilibria
- Vol. 475, 64-76
Application of the unscented transform in the uncertainty propagation of thermodynamic model parameters
- 1 September 2017
- Published by Elsevier in Computers & Chemical Engineering
- Vol. 104, 211-230
Robust synthesis of the pressure-swing distillation process under azeotropic feed composition disturbance—Study of the tetrahydrofuran/methanol system
- 1 September 2017
- Published by Elsevier in Computers & Chemical Engineering
- Vol. 104, 211-230
Robust synthesis of the pressure-swing distillation process under azeotropic feed composition disturbance—Study of the tetrahydrofuran/methanol system
- 4 April 2017
- Published by American Chemical Society (ACS) in Industrial & Engineering Chemistry Research
- Vol. 56 (15) , 4446-4462
Robust Design and Control of Extractive Distillation Processes under Feed Disturbances
- 4 April 2017
- Published by American Chemical Society (ACS) in Industrial & Engineering Chemistry Research
- Vol. 56 (15) , 4446-4462
Robust Design and Control of Extractive Distillation Processes under Feed Disturbances