Detail publikace

Two-stage stochastic programming approach to a PDE-constrained steel production problem with the moving interface

Originální název

Two-stage stochastic programming approach to a PDE-constrained steel production problem with the moving interface

Anglický název

Two-stage stochastic programming approach to a PDE-constrained steel production problem with the moving interface

Jazyk

en

Originální abstrakt

The paper is concerned with a parallel implementation of the progressive hedging algorithm (PHA) which is applicable for the solution of stochastic optimization problems. We utilized the Message Passing Interface (MPI) and the General Algebraic Modelling System (GAMS) to concurrently solve the scenario-related subproblems in parallel manner. The standalone application combining the PHA, MPI, and GAMS was programmed in C++. The created software was successfully applied to a steel production problem which is considered by means of the two-stage stochastic PDE-constrained program with a random failure. The numerical heat transfer model for the steel production was derived with the use of the control volume method and the phase changes were taken into account with the use of the effective heat capacity. Numerical experiments demonstrate that parallel computing facility has enabled a significant reduction of computational time. The quality of the stochastic solution was evaluated and discussed. The developed system seems computationally effective and sufficiently robust which makes it applicable in other applications as well.

Anglický abstrakt

The paper is concerned with a parallel implementation of the progressive hedging algorithm (PHA) which is applicable for the solution of stochastic optimization problems. We utilized the Message Passing Interface (MPI) and the General Algebraic Modelling System (GAMS) to concurrently solve the scenario-related subproblems in parallel manner. The standalone application combining the PHA, MPI, and GAMS was programmed in C++. The created software was successfully applied to a steel production problem which is considered by means of the two-stage stochastic PDE-constrained program with a random failure. The numerical heat transfer model for the steel production was derived with the use of the control volume method and the phase changes were taken into account with the use of the effective heat capacity. Numerical experiments demonstrate that parallel computing facility has enabled a significant reduction of computational time. The quality of the stochastic solution was evaluated and discussed. The developed system seems computationally effective and sufficiently robust which makes it applicable in other applications as well.

BibTex


@article{BUT143382,
  author="Lubomír {Klimeš} and Pavel {Popela} and Tomáš {Mauder} and Josef {Štětina} and Pavel {Charvát}",
  title="Two-stage stochastic programming approach to a PDE-constrained steel production problem with the moving interface",
  annote="The paper is concerned with a parallel implementation of the progressive hedging algorithm (PHA) which is applicable for the solution of stochastic optimization problems. We utilized the Message Passing Interface (MPI) and the General Algebraic Modelling System (GAMS) to concurrently solve the scenario-related subproblems in parallel manner. The standalone application combining the PHA, MPI, and GAMS was programmed in C++. The created software was successfully applied to a steel production problem which is considered by means of the two-stage stochastic PDE-constrained program with a random failure. The numerical heat transfer model for the steel production was derived with the use of the control volume method and the phase changes were taken into account with the use of the effective heat capacity. Numerical experiments demonstrate that parallel computing facility has enabled a significant reduction of computational time. The quality of the stochastic solution was evaluated and discussed. The developed system seems computationally effective and sufficiently robust which makes it applicable in other applications as well.",
  address="Institute of Information Theory and Automation of The Czech Academy of Sciences",
  chapter="143382",
  doi="10.14736/kyb-2017-6-1047",
  howpublished="online",
  institution="Institute of Information Theory and Automation of The Czech Academy of Sciences",
  number="6",
  volume="53",
  year="2017",
  month="december",
  pages="1047--1070",
  publisher="Institute of Information Theory and Automation of The Czech Academy of Sciences",
  type="journal article in Web of Science"
}