Funcction of Functional Modelling in (Bio-)Chemical Engineering1

The Function of Functional Modelling in (Bio-)Chemical Engineering

StenBay Jørgensen,a Niels Jensen, aMorten Lind,b

aDepartment of Chemical Engineering,Technical University of Denmak,DK-2800 Lyngby,Denmark

bØrsted_DTU, Technical University of Denmak, DK-2800 Lyngby, Denmark

cResearch Centre, Address, Zip-code City, Country

  1. Summary

Functional modelling links the modelling purpose with its physical/chemical implementations. Thereby this modelling methodology, which has been developed within computer science sems to provide a systematic basis for spolving many of the challenging problems related to multiple scales and multiple abstartion levels in process and product engineering.

Keywords: , multiscale modelling, modelling networks, complexity, Automated HAZOP, regulatory networks

  1. Extended Abstract

Functional modelling links the modelling purpose with its physical/chemical implementations. This linkage is achieved by systematically representing the modelling purposes and their derived sub-goals and subsequently connecting these to the means through which the (sub-)purposes are achieved. These connection may be made within a multilevel abstraction framework called functional modelling. By assigning cause and effect to the means it is also possible to reason in functional models. To facilitate model development a functional modelling framework may be implemented in a computer environment wherein also the reasoning can be carried out. Clearly functional modelling provides a natural basis for integrating design over many different scales which constitutes a fundamental problem in chemical engineering.

Functional models may be directly used in design, e.g. when a separation (or reaction) agent is desired for the execution of a specific separation (or reaction). Here functional modelling may be used to facilitate reverse engineering such that the process may be optimally designed assuming the availability of the proper ideal agent and in a subsequent step then the (nearly) ideal agent may be designed in a separate step. If design of a nearly ideal agent indeed is feasible then an optimal design has been achieved, if however a suboptimal design must be settled for, then however this may be rather easily verified as a neraly optimal solution can be provided as a starting point. Eden et al (2004).

Functional models clealrly may be used for many different purposes. A number of these have been illustrated recently. One example is to model the function of regulatory networks in microorganisms thereby illustrating how information at many different abstraction levels may be combined to achieve a desired functionality. Here functional modelling enables simultaneous representation of multiple functionalities, thereby facilitating the understanding of the different modes of control in micro-organisms, Gernaey et al (2006). Another example is to automate HAZOP analysis of an integrated chemical plant again based upon afunctional model of the plant combined with the causal roles which permit reasoning about consequences of deviations in process plants.. This methodology has been developed and implemented as a prototype in a JESS based system. The automated HAZOP methodology has been demonstrated in a case study performed on the Indirect Vapour Recompression Distillation Plant (IVaRDiP) at DTU. This case illustrates that it is possible in a flexible manner to reveal both local and quite remote cause and effect relations which normally are quite difficult to uncover for unexperineced personel. Thereby the autokmated HAZOP may provide a highly desirable assistance during Hazard identification, Jensen et al (2006).

Functional modelling of a process plant with its control system also directly reveals the importance of being able to model the functionality of the control system for designing the (optimal) plant operation. Thus a potential future usage of functional modelling may be to represent a process design and its potential control design thereby enabling reasoning on simultaneous process and control design including paying proper regard to the results of a HAZOP analysis as reflected in the requirements to the control system. Thus pointing to development of a suitable (optimal) operations managemnet system specification.

Combining mathematical models and their purposes, i.e. as expressed in Functional Models seem to be intuitively obvious, however this link is usually not explicitly available in traditional mathematical models. Instead mathematical models may be developed form a Functional Model once the means for achieving the desired functionality are specified and suitably classified.

Thus functional modelling indeed may be used in many different functions in (bio-)chemical processes. In fact engineering is all about developing systems to achieve a desired functionality, hence functional modelling can be used advantageously in many aspect of engineering design. With the dual notion of abstratction hierarchies and means end relationships then functional modelling provides a systematic framework and thereby a scientific basis for managing complexity. Therefore it can be highly advantageous for cheimcal) engineering to develop functional modelling as a separate discipline more explicitly.

References:

  1. M.R. Eden, S.B. Jørgensen, R. Gani, M. El-Halwagi, 2004, “A novel framework for simultaneous separation process and product design”, Chemical Engineering & Processing, 43, pp. 595-608
  2. Gernaey K.V., Lind M. and Jørgensen S.B., 2006 “Towards understanding the role and function of regulatory networks in microorganisms” in, Puigjaner L. and Heyen G. (Eds.) Computer Aided Process & Product Engineering. Wiley-VCH, Weinheim (Germany), vol. 1, Chapter 7, pp. 223-264
  3. Netta Liin Rossing, Morten Lind, Johannes Petersen, Sten Bay Jørgensen and Niels Jensen, 2006, “A Functional approach to HAZOP studies”, Chemical Engineering Transactions (AIDIC), Volume 9: Proceedings of 2nd International Conference on Safety & Environment in Process Industry, Editor S.S. Buratti, pp. 49-54