Skip to content

The 'dwsimopt' is a Python library that automates DWSIM simulations for process optimization.

License

Notifications You must be signed in to change notification settings

lf-santos/dwsimopt

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

95 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

dwsimopt: DWSIM simulation optimization with Python!

PyPI Documentation Status License

The DWSIM Optimization (dwsimopt) is a Python library that automates DWSIM simulations for process optimization. The simulation dlls are embedded in the programming environment so that they can be accessed and modified by the optimization algorithms.

Mathematical background

Although very efficient to describe in detail complex systems that would otherwise have to be simplified or approximated, black-box process simulators lack the symbolic formulation of the process model equations and the analytical derivatives that are useful for optimization, for example. The use of simulation may also introduce noise to the calculations due to convergence and approximations of numerical methods, which can jeopardize the calculation of accurate approximate derivatives and, therefore, the use of gradient-based optimization methods directly [1]. Also, the lack of analytical formulations of the optimization problem prevents the derivation of rigorous upper and lower bounds of the functions that are used for deterministic global optimization [2]. In that sense, the optimization models that require simulations to calculate the objective function and/or constraints are often referred to as simulation optimization problem [3]. A simplified version of this class of problems can be described as finding an equation that solves globally the following constrained problem

equation

in which the objective function equation and constraints equation, being q the number of constraints, are somewhat expensive to calculate, slightly noisy, and black-box functions, i.e. there is no available mathematical expression for f or g, but for a given equation the values of f(x) and f(x) are calculated in a computer code simulation with some noise.

Requirements

  • Python <= 3.9 (python 3.8 recommended -- using python 3.9 requires installing dwsimopt from setup.py)
  • DWSIM v7+ (open-source chemical process simulation. Download here)
  • pythonnet == 2.5.2 (on Python 3.9 you'll need to download the pythonnet2.5.2 wheel and pip install path\to\pythonnet_wheel)
  • pywin32
  • numpy
  • scipy
  • scikit-opt

It is recommendable to start from a fresh environment and let the dwsimopt install the dependencies, see Installation section. DWSIM must be downloaded and installed manually.

Installation

Install the latest version of this repository to your machine

pip install dwsimopt

or

git clone https://github.com/lf-santos/dwsimopt.git
cd dwsimopt
python setup.py install

Make sure you have all the required packages and software. Navigate through the jupyter notebook examples. Use the OptimiSim class to embed your DMSWIM simulation into Python. Add degrees of freedom, objective function and constraints from your simulation optimization problem with the py2dwim python-dwsim data exchange interface. Solve the problem with a suitable optimization solver (surrogate-based optimization or global optimization meta-heuristics recommended) that you can find methods in the OptimiSim class (e.g. GA, PSO, DE)

Citing us

If you use dwsimopt, please cite the following paper: L. F. Santos, C. B. B. Costa, J. A. Caballero, M. A. S. S. Ravagnani, Framework for embedding black-box simulation into mathematical programming via kriging surrogate model applied to natural gas liquefaction process optimization, Applied Energy, 310, 118537 (2022).

@article{Santos2022,
title = {Framework for embedding black-box simulation into mathematical programming via kriging surrogate model applied to natural gas liquefaction process optimization},
author = {Lucas F. Santos and Caliane B.B. Costa and José A. Caballero and Mauro A.S.S. Ravagnani},
journal = {Applied Energy},
volume = {310},
pages = {118537},
year = {2022},
issn = {0306-2619},
doi = {https://doi.org/10.1016/j.apenergy.2022.118537},