Process Systems Enterprise Limited
email this page print this page
pdf overview

gPROMS for academics

Research into new numerical solutions techniques

Did you know?

gPROMS supports research in over 200
of the world's top universities

Researchers developing cutting-edge solution techniques benefit from gPROMS® in a number of different ways.

Typical application includes using gPROMS as a model server and a test bed for new developments – for example, the implementation of new global optimisation methods.

Who can benefit?

You can benefit from using gPROMS if you:

  • develop solution algorithms for complex, large-scale systems
  • need a transparent model-server architecture

    • "I interfaced gPROMS with C++ to automatically calculate solutions of optimisation problems to provide the necessary input for my multiparametric programming algorithm"

    — Diogo Narciso
    Imperial College London

  • want to validate your solver developments within an existing simulation/optimisation environment.

gPROMS provides a number of facilities to help you achieve these goals.

What does gPROMS offer?

gPROMS provides the following capabilities specifically aimed at researcher developing numerical solutions algorithms and techniques:

Model-server architecture. gPROMS can be used as a model-server to encapsulate the mathematical representation of complex physical systems in a transparent way, and at different levels of abstraction.

 

 

 

"gPROMS enabled us to investigate systematically how choices at the level of engineering/mathematical modelling and numerical solution algorithms affect the computational speed of model-based applications".

— Pablo Rolandi
Presented at ESCAPE 18

 

A framework for analysis of computational load of CAPE tools - Rolandi & Cano

A framework for analysis of computational load of CAPE tools (Rolandi, Cano), presented at ESCAPE-18

The gPROMS Server (gSERVER) allows you to run simulations, optimisations and parameter estimations during development and testing of higher-level solution algorithms may be built and implemented.

gSERVER also allows you to access a lower-level representation of the underlying systems of equations, the Equation Set Object (ESO). You can interface the ESO to your algorithm in order to separete solution aspects from model-formulation aspects, with the benefit of gPROMS's advanced modelling capabilities. This is related to CAPE-OPEN's Process Modelling Component (PMC) concept.

Numerical vs. structural information. Using gSERVER's capabilities, you can access both numerical and structural information on the system of equations.

Having additional information arising from structural considerations provides a competitive edge to cutting-edge hybrid (numerical/structural) solution techniques, including residual and analytical Jacobian expression. This is the case for full nonlinear models as well as minimal linear realisations of the equation set.

Built-in solvers. Provided that your solution algorithm supports CAPE-OPEN solver standards specifications, then it is possible to use gPROMS's built-in solvers as subsolver components, or to embed your algorithm as part of a higher-level solver component. This way you benefit from existing robust and efficient solution algorithm implementations, minimising rework and scope for error.
Modelling environment executive. If you implement your solution algorithms or solvers based on the CAPE-OPEN standards, you will be able to use gPROMS's ModelBuilder® (and the gPROMS family of product in general) as a modelling environment executive and a platform on which to validate (and possibly commercialise) your solution techniques.

Apart from modelling and flowsheeting capabilities, the platform provides full management and visualisation of results. See the 25 good reasons to use gPROMS ModelBuilder for more information.

Computational load analysis and benchmarking. gPROMS's state-of-the-art modelling engine and solution techniques make it suitable for benchmarking of computational cost and loads, as well as analysis of the causes of high computational times. gPROMS has been shown to be significantly faster than similar modelling technologies/languages.