Development of a new column shortcut model and its application in process optimisation

2019 
Abstract Stage-to-stage column models are preferred for accurate process calculation in industry but include many non-linear equations making process optimisation difficult. Further, optimisation of feed and side draw locations or overall column stage numbers introduces integer type optimisation variables and transforms an NLP into an MINLP optimisation problem. We introduce the Adapted Edmister Model (AEM) as a new column shortcut method which is able to overcome the described challenges while satisfying industrial standards of simulation accuracy. The AEM is used in a two-step optimisation approach which consists of initial optimisation with shortcut models and subsequent optimisation with stage-to-stage models. Applying the AEM in two-step optimisation enables NLP optimisation of column stage numbers, feed locations, and side draw locations avoiding an MINLP problem formulation. In addition, using AEM results to initialise stage-to-stage column models improves convergence and gives rise to better optimal solutions. To demonstrate the performance of the AEM, it is applied in the optimisation of an air separation process where high calculation accuracy is crucial. Here, the two-step optimisation approach employing the AEM results in an over 2% better optimum compared to a classical optimisation approach which is high for this process.
    • Correction
    • Source
    • Cite
    • Save
    • Machine Reading By IdeaReader
    43
    References
    2
    Citations
    NaN
    KQI
    []