Rapid and effective exploration of the process decision space
The main use of process simulation and modelling tools is to analyse "what-if" scenarios in order to improve process design and operation. Currently this is done manually as a point activity, using repeated simulation runs.
Global systems analysis (GSA) allows the comprehensive exploration of the behaviour of a system over domains of any user-selected subset of its input variables ('factors'), and output variables ('responses').
This provides a quick, easy and systematic way to explore the complex process design and operational decision space using high-fidelity models.
How does it work?
GSA can be applied very easily to any gPROMS model. The gPROMS ProcessBuilder screenshot below shows some results of GSA applied to the design of a styrene monomer plant.
In this example, the effect of uncertainty in the reaction kinetics and the tray efficiencies on the total annual profit of the plant was quantified.
To start, the user chooses:
- any subset of the model input variables (‘factors’). These may include external disturbances, control actions and model parameters such as reaction kinetic constants. The factors may vary over domains specified in terms of lower and upper bounds. They may be deterministic or probabilistic, in the latter case their variation being described in terms of univariate or multivariate probability distributions.
- the output variables of interest (‘responses’). These are typically key performance indicators (KPIs) that are of critical interest for plant design or operation.
Having selected factors and output variables of interest the model is then executed hundreds or even thousands of times, depending on the analysis. The results are then presented in a variety of ways, allowing easy understanding on the influence of factors on the responses.
What are the benefits?
Key benefits of GSA are the ability to perform sensitivity analysis and uncertainty quantification on complex process systems in a systematic and highly efficient manner.
This allows rapid and effective exploration of the process design and operational decision space, with rapid screening and ranking of alternatives.
More about GSA …
GSA can be applied to any gPROMS model very easily, irrespective of the model size or complexity. The GSA implementation:
- employs efficient techniques for sampling the input space based on low-discrepancy sequences
- is designed for deployment on distributed computing hardware
- has built-in resilience, e.g. in terms of being able to deal with failure of individual samples
- computes important aspects of responses such as measures of their probability distributions and their global sensitivity indices with respect to the input factors.
- incorporates extensive facilities for visualising and analysing the results
GSA is available as an option in gPROMS FormulatedProducts and gPROMS ProcessBuilder.