Jose Luis Pitarch
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
Jose Luis Pitarch has a total of 63 co-authors. Here are the top co-authors by number of shared publications:
- Antonio Sala
Instituto de Automática e Informática Industrial Universitat Politècnica de València Valencia Spain
14 shared publications - Cesar de Prada
Universidad de Valladolid
9 shared publications - Carlos Ariño
D. de Ingeniería de Sistemas Industriales y Diseño, Universitat Jaume I, Avenida Vicent Sos Baynat, s/n., 12071 Castelló de la Plana, Spain
7 shared publications C. de Prada
6 shared publicationsCésar de Prada
Univ Valladolid, Dept Syst Engn & Automatic Control, C Dr Mergelina s-n, Valladolid 47011, Spain
6 shared publicationsCesar de Prada
5 shared publications
Latest Publications
- Open Access
- 1 July 2025
- Published by The CACHE Foundation, Inc. in Systems and Control Transactions
- Vol. 4, 1530-1535
Nonmyopic Bayesian process optimization with a finite budget
Optimization under uncertainty is inherent to many PSE applications ranging from process design to RTO. Reaching process true optima often involves learning from experimentation, but actual experiments involve a cost (economic, resources, time) and a budget limit usually exists. Finding the best trade-off on cumulative process performance and experimental cost over a finite budget is a Partially Observable Markov Decision Process (POMDP), known to be computationally intractable. This paper follows the nonmyopic Bayesian optimization (BO) approximation to POMDPs developed by the machine-learning community, that naturally enables the use of hybrid plant surrogate models formed by fundamental laws and Gaussian processes (GP). Although nonmyopic BO using GPs may look more tractable, evaluating multi-step decision trees to find the best first-stage candidate action to apply is still expensive with evolutionary or NLP optimizers. Hence, we propose modelling the value function of the first-stage decision also with a GP, whose data will correspond to virtual evaluations of second-stage decision trees build upon myopic rollouts. Thus, the nonmyopic initial decision can be efficiently optimized via BO and the virtually learned value function. Effectiveness of the approach is demonstrated in a wide benchmark with synthetically generated functions as well as to optimize small batch production with a chemical reactor.
- 1 July 2025
- Published by The CACHE Foundation, Inc. in Systems and Control Transactions
- Vol. 4, 1530-1535
Open AccessNonmyopic Bayesian process optimization with a finite budget
Optimization under uncertainty is inherent to many PSE applications ranging from process design to RTO. Reaching process true optima often involves learning from experimentation, but actual experiments involve a cost (economic, resources, time) and a budget limit usually exists. Finding the best trade-off on cumulative process performance and experimental cost over a finite budget is a Partially Observable Markov Decision Process (POMDP), known to be computationally intractable. This paper follows the nonmyopic Bayesian optimization (BO) approximation to POMDPs developed by the machine-learning community, that naturally enables the use of hybrid plant surrogate models formed by fundamental laws and Gaussian processes (GP). Although nonmyopic BO using GPs may look more tractable, evaluating multi-step decision trees to find the best first-stage candidate action to apply is still expensive with evolutionary or NLP optimizers. Hence, we propose modelling the value function of the first-stage decision also with a GP, whose data will correspond to virtual evaluations of second-stage decision trees build upon myopic rollouts. Thus, the nonmyopic initial decision can be efficiently optimized via BO and the virtually learned value function. Effectiveness of the approach is demonstrated in a wide benchmark with synthetically generated functions as well as to optimize small batch production with a chemical reactor.
- Open Access
- 17 July 2024
- Published by Elsevier in Computers & Industrial Engineering
- Vol. 194
Similarity-based decomposition algorithm for two-stage stochastic scheduling
- 17 July 2024
- Published by Elsevier in Computers & Industrial Engineering
- Vol. 194
Open AccessSimilarity-based decomposition algorithm for two-stage stochastic scheduling
- Open Access
- 9 July 2024
- Published by Universitat Politecnica de Valencia in Revista Iberoamericana de Automática e Informática Industrial
- Vol. 21 (4) , 328-338
Optimización Bayesiana no miope POMDP para procesos con restricciones de operación y presupuesto finito
Mejorar la toma de decisiones a partir de los resultados observados tras la experimentación es una tarea habitual en muchas aplicaciones, tanto a nivel de investigación en laboratorio como en procesos de producción industriales. Sin embargo, realizar experimentos suele acarrear un coste no despreciable, por lo que una excesiva exploración es perjudicial. La optimización bayesiana es una técnica muy utilizada en este contexto, decidiendo la siguiente experimentación en base a un modelo estadístico.No obstante, está técnica no tiene en cuenta explícitamente el coste real de realizar un experimento, ni si existe un presupuesto (o número de experimentos, tiempo, etc.) máximo. El problema de toma de decisiones bajo incertidumbre y presupuesto finito puede plantear como un Proceso de Decisión de Márkov Parcialmente observable (POMDP, por sus siglas en inglés). Este trabajo aborda el problema de optimización experimental sujeta a restricciones de operación con un enfoque POMDP, donde las posibles decisiones vienen proporcionadas por heurísticas de la optimización bayesiana, o de otra índole definida por el usuario. La estrategia consiste en construir un árbol de posibles escenarios partir del conocimiento (incierto) acerca del proceso/sistema aprendido a partir de experimentos previos. Dicho conocimiento se modela mediante procesos Gaussianos, que se actualizan con cada nueva observación. La evaluación sobre la mejor decisión a tomar se realiza mediante programación dinámica. El algoritmo desarrollado ha sido evaluado mediante comparación con otras opciones de la literatura en un banco de pruebas sintético, y para optimizar un proceso químico de producción por lotes.
- 9 July 2024
- Published by Universitat Politecnica de Valencia in Revista Iberoamericana de Automática e Informática Industrial
- Vol. 21 (4) , 328-338
Open AccessOptimización Bayesiana no miope POMDP para procesos con restricciones de operación y presupuesto finito
Mejorar la toma de decisiones a partir de los resultados observados tras la experimentación es una tarea habitual en muchas aplicaciones, tanto a nivel de investigación en laboratorio como en procesos de producción industriales. Sin embargo, realizar experimentos suele acarrear un coste no despreciable, por lo que una excesiva exploración es perjudicial. La optimización bayesiana es una técnica muy utilizada en este contexto, decidiendo la siguiente experimentación en base a un modelo estadístico.No obstante, está técnica no tiene en cuenta explícitamente el coste real de realizar un experimento, ni si existe un presupuesto (o número de experimentos, tiempo, etc.) máximo. El problema de toma de decisiones bajo incertidumbre y presupuesto finito puede plantear como un Proceso de Decisión de Márkov Parcialmente observable (POMDP, por sus siglas en inglés). Este trabajo aborda el problema de optimización experimental sujeta a restricciones de operación con un enfoque POMDP, donde las posibles decisiones vienen proporcionadas por heurísticas de la optimización bayesiana, o de otra índole definida por el usuario. La estrategia consiste en construir un árbol de posibles escenarios partir del conocimiento (incierto) acerca del proceso/sistema aprendido a partir de experimentos previos. Dicho conocimiento se modela mediante procesos Gaussianos, que se actualizan con cada nueva observación. La evaluación sobre la mejor decisión a tomar se realiza mediante programación dinámica. El algoritmo desarrollado ha sido evaluado mediante comparación con otras opciones de la literatura en un banco de pruebas sintético, y para optimizar un proceso químico de producción por lotes.
- Open Access
- 28 August 2023
- Published by Universidade da Coruna
- p. 447-452
Optimización experimental con presupuesto finito combinando heurísticas Bayesianas en un POMDP
Resumen Mejorar la toma de decisiones a partir de los resultados observados tras la experimentación es una tarea habitual en muchas aplicaciones, tanto a nivel de investigación en laboratorio como a escala industrial. Sin embargo, realizar experimentos suele acarrear un coste no despreciable, por lo que una excesiva exploración es perjudicial. La optimización Bayesiana es una técnica muy utilizada en este contexto, debido a su bajo coste computacional y a que proporciona un buen balance entre explotación y exploración. No obstante, está técnica no tiene en cuenta explícitamente el coste real de realizar un experimento, ni si existe un presupuesto (o número de experimentos, tiempo, etc.) máximo. El problema de toma de decisiones bajo incertidumbre y presupuesto finito es un proceso de decisión de Markov parcialmente observable (POMDP, por sus siglas en inglés). Este trabajo aborda el problema de optimización experimental combinando reconocidas heurísticas Bayesianas en un enfoque POMDP resuelto mediante programación dinámica, donde un árbol de escenarios se construye partir del conocimiento del proceso/sistema disponible (con incertidumbre) en cada etapa. Dicho conocimiento se modela mediante un proceso Gaussiano que se actualiza con cada nueva observación. El algoritmo desarrollado ha sido testeado con éxito para optimizar las consignas de un reactor de tanque agitado que debe producir una cierta cantidad de lotes.
- 28 August 2023
- Published by Universidade da Coruna
- p. 447-452
Open AccessOptimización experimental con presupuesto finito combinando heurísticas Bayesianas en un POMDP
Resumen Mejorar la toma de decisiones a partir de los resultados observados tras la experimentación es una tarea habitual en muchas aplicaciones, tanto a nivel de investigación en laboratorio como a escala industrial. Sin embargo, realizar experimentos suele acarrear un coste no despreciable, por lo que una excesiva exploración es perjudicial. La optimización Bayesiana es una técnica muy utilizada en este contexto, debido a su bajo coste computacional y a que proporciona un buen balance entre explotación y exploración. No obstante, está técnica no tiene en cuenta explícitamente el coste real de realizar un experimento, ni si existe un presupuesto (o número de experimentos, tiempo, etc.) máximo. El problema de toma de decisiones bajo incertidumbre y presupuesto finito es un proceso de decisión de Markov parcialmente observable (POMDP, por sus siglas en inglés). Este trabajo aborda el problema de optimización experimental combinando reconocidas heurísticas Bayesianas en un enfoque POMDP resuelto mediante programación dinámica, donde un árbol de escenarios se construye partir del conocimiento del proceso/sistema disponible (con incertidumbre) en cada etapa. Dicho conocimiento se modela mediante un proceso Gaussiano que se actualiza con cada nueva observación. El algoritmo desarrollado ha sido testeado con éxito para optimizar las consignas de un reactor de tanque agitado que debe producir una cierta cantidad de lotes.
- Open Access
- 2 June 2023
- Published by American Chemical Society (ACS) in Industrial & Engineering Chemistry Research
- Vol. 62 (23) , 9278-9289
Integrating Continuous and Batch Processes with Shared Resources in Closed-Loop Scheduling: A Case Study on Tuna Cannery
Scheduling tasks in production facilities are usually hybrid optimization problems of a large combinatorial nature. They involve solving, in near-real time, the integration of the operation of several batch units of continuous dynamics with the discrete manufacture of items in processing lines. Moreover, one has to deal with uncertainty (process delays, unexpected stops) and the management of shared resources (energy, water, etc.) including decisions made by plant operators: still, some tasks in the scheduling layers are done manually. Manufacturing Execution Systems (MESs) are intended to support plant personnel at this level. However, there is still much work to do in terms of performing automatic scheduling, computed in real time, that guides managers to achieve an optimal operation of such complex cyber-physical systems. This work proposes a closed-loop approach to handle the uncertainty arising when facing the online scheduling of supply lines and parallel batch units. These units often share some resources, so effects due to concurrent resource consumption on the system dynamics are explicitly considered in the presented formulation. The proposed decision support system is tested onsite in a tuna cannery, to handle short-term online scheduling of sterilization processes that deal with limited steam, carts, and operators as shared resources.
- 2 June 2023
- Published by American Chemical Society (ACS) in Industrial & Engineering Chemistry Research
- Vol. 62 (23) , 9278-9289
Open AccessIntegrating Continuous and Batch Processes with Shared Resources in Closed-Loop Scheduling: A Case Study on Tuna Cannery
Scheduling tasks in production facilities are usually hybrid optimization problems of a large combinatorial nature. They involve solving, in near-real time, the integration of the operation of several batch units of continuous dynamics with the discrete manufacture of items in processing lines. Moreover, one has to deal with uncertainty (process delays, unexpected stops) and the management of shared resources (energy, water, etc.) including decisions made by plant operators: still, some tasks in the scheduling layers are done manually. Manufacturing Execution Systems (MESs) are intended to support plant personnel at this level. However, there is still much work to do in terms of performing automatic scheduling, computed in real time, that guides managers to achieve an optimal operation of such complex cyber-physical systems. This work proposes a closed-loop approach to handle the uncertainty arising when facing the online scheduling of supply lines and parallel batch units. These units often share some resources, so effects due to concurrent resource consumption on the system dynamics are explicitly considered in the presented formulation. The proposed decision support system is tested onsite in a tuna cannery, to handle short-term online scheduling of sterilization processes that deal with limited steam, carts, and operators as shared resources.
- 5 August 2022
- Published by Elsevier in IFAC-PapersOnLine
- Vol. 55 (7) , 821-826
The Similarity Index to Decompose Two-Stage Stochastic Scheduling Problems
- 5 August 2022
- Published by Elsevier in IFAC-PapersOnLine
- Vol. 55 (7) , 821-826
The Similarity Index to Decompose Two-Stage Stochastic Scheduling Problems
- 1 January 2022
- Published by ARGESIM Arbeitsgemeinschaft Simulation News in SNE Simulation Notes Europe
- Vol. 32 (1) , 29-36
Machine Learning and the Digital Era from a Process Systems Engineering Perspective
- 1 January 2022
- Published by ARGESIM Arbeitsgemeinschaft Simulation News in SNE Simulation Notes Europe
- Vol. 32 (1) , 29-36
Machine Learning and the Digital Era from a Process Systems Engineering Perspective
- Open Access
- 1 September 2021
- Published by Elsevier in Journal of Food Engineering
- Vol. 304
Optimal operation of thermal processing of canned tuna under product variability
- 1 September 2021
- Published by Elsevier in Journal of Food Engineering
- Vol. 304
Open AccessOptimal operation of thermal processing of canned tuna under product variability
- Open Access
Integrated Process Re-Design with Operation in the Digital Era: Illustration through an Industrial Case Study
This work discusses what should be the desirable path and correct tools for the optimal re-design and operation of processes in the Industry 4.0 framework, as illustrated in a challenging case study corresponding to a complex network of evaporation plants in a viscose-fiber factory. The goal is to integrate optimal design, to improve the existing cooling systems, together with the optimal operation of the whole network, balancing the initial investment with the potentially achievable savings. A rigorous mathematical model for such optimization purpose has been built. The model explicitly considers different structural alternatives as a superstructure for the incorporation of new equipment into the network. The uncertainty associated to future operating conditions is also considered by using a two-stage stochastic formulation. Furthermore, the model is also the base from which a deterministic real-time optimization (RTO) builds upon to support the daily management of the future network operation. The RTO tool suggests the allocation of different products to evaporation plants, the distribution of the cooling water and the suitable number of heat pumps to switch on for optimal economic operation. Design and operation problems are formulated and solved via mixed-integer non-linear programming and the results have been tested with historical plant data.
Open AccessIntegrated Process Re-Design with Operation in the Digital Era: Illustration through an Industrial Case Study
This work discusses what should be the desirable path and correct tools for the optimal re-design and operation of processes in the Industry 4.0 framework, as illustrated in a challenging case study corresponding to a complex network of evaporation plants in a viscose-fiber factory. The goal is to integrate optimal design, to improve the existing cooling systems, together with the optimal operation of the whole network, balancing the initial investment with the potentially achievable savings. A rigorous mathematical model for such optimization purpose has been built. The model explicitly considers different structural alternatives as a superstructure for the incorporation of new equipment into the network. The uncertainty associated to future operating conditions is also considered by using a two-stage stochastic formulation. Furthermore, the model is also the base from which a deterministic real-time optimization (RTO) builds upon to support the daily management of the future network operation. The RTO tool suggests the allocation of different products to evaporation plants, the distribution of the cooling water and the suitable number of heat pumps to switch on for optimal economic operation. Design and operation problems are formulated and solved via mixed-integer non-linear programming and the results have been tested with historical plant data.