This section presents results from a survey of process
modellers or modelling managers in the process industry
based in various countries... Chemical plants, processes and products have lifecycles a sequence of phases they go through... Here are some steps in the process lifecycle. The graph shows the percentage of survey respondents who answered yes to undertaking modelling activities in that phase of the lifecycle. Modelling was defined as the activity of developing or using mathematical forms to describe the behaviour of a system and then using that model for simulation or other purposes. This is closely related to the previous graph... Why are modelling and simulation activities undertaken? Focus on the blue parts of the graph (strongly agree and agree responses). Predominantly it is for product and process design, optimisation and feasibility assessment, but also for operator support and fault diagnosis (e.g. investigations to find why something is not working), risk assessment and operator training. The table shows technical software only, so we are not including here word processing, presentation, database and similar applications. FORTRAN is a engineering computing language developed in the 1950s. It has undergone extensive revision. It is still used in places in industry (ie you might need to write or modify some FORTRAN code) and it is also used in Aspen Plus (for example, when typing in equations, you can use FORTRAN statements). Excel is ubiquitous remember that VBA (Visual Basic for Applications) underpins Excel you can write macros or more extensive VBA programs within Excel. Matlab has great numerical routines and graphics. It has many add-on toolboxes that extend its functionality e.g. optimisation, neural networks, signal processing, image analysis, statistics, mapping, Polymath is another general maths package, but was developed by some chemical engineers (http://www.polymath-software.com/).
Scad and JKSimMet are more common for minerals
simulation. Details extracted from the C&I paper: * A sample of industrial process modellers or modelling managers was asked which lifecycle activities (horizontal axis) their work group addressed through modelling and simulation work. They indicated either yes or no to each category. * The percentage figure reported (vertical axis) is the number of yes responses in that category divided by the total number of yes responses summed over all categories x 100%. * A wide range of software is used as shown. Microsoft Excel and flowsheeting packages are the most frequently used tools, each accounting for 17% of all responses. Earlier survey work by Edgar in 2003 reported that only 48% of chemical engineering organizations used flowsheeting packages, yet 9899% of recent chemical engineering graduates used a spreadsheet, most often Excel. Other tools that find high use in our study include Matlab (13%), direct coding ie programming (11%), CFD (8%), hybrid modelling and simulation software (7%), and optimisation programs (6%). * It is understandable that flowsheeting packages should find strong use for modelling and simulation of process systems (17% of responses), but, in fact, when the three most popular general mathematical/programming methods (Excel, Matlab and direct coding) are combined, they account for 41% of responses. This suggests that there are often tasks that require custom modelling, and they are not adequately handled by flowsheet simulators alone. For SMEs (small to medium sized enterprises), cost may also be a significant factor in the choice of modelling software. Excels spreadsheet format may be so popular for this custom modelling and data analysis role because of its low cost, wide distribution, ease of learning, ease of interface development and adequate graphics. Spreadsheets are particularly useful for quick, interactive analysis. However, some situations tend to be more suited to a programming environment (Matlab, direct coding), for example: complex conditional statements, variable numbers of entities, dynamic simulation and so on. Programming is also available in Excel through Visual Basic for Applications (VBA), possibly also contributing to Excels appeal. This will be something of a theme in this unit... How do you know your simulation is ok? Standard validation procedures are well-established company approaches to treating data and comparing it with model predictions to arrive at a decision regarding the validity of the model. Seventy percent of respondents said that steady state models are readily validated, but this reduced to about 35 percent for dynamic models. Another study (Foss et al., 1998) also found that efficient validation procedures are lacking for dynamic models and for very detailed non- standard steady state models. The way to modify inadequate models is clear for only around 40 percent of respondents. One survey respondent remarked that modelling and validation are often carried out by different people in his work group.
There is around 65 percent agreement and 20 percent
disagreement that it is easy to set up models for a stated purpose. In terms of the model solution phase, 45 percent agree that tools are available for resolving solution failures, yet only around 25 percent believe that existing diagnosis systems provide good advice. Approximately 35 percent agree that it is easy to recall previous modelling scenarios and case studies, but 40 percent or so disagree. These responses suggest that there needs to be work on improving the tools for diagnosing and fixing solution failures, and for archiving modelling scenarios. CAPE = Computer Aided Process Engineering which is virtually synonymous with PSE = Process Systems Engineering CAPE / PSE lies at the interface of chemical engineering, maths and computer science all good stuff! Scene from the BBC TV series Little Britain ... and what we would like to avoid in process engineering blindly accepting what the computer says. This picture used to be at: http://www.davidwalliams.com/LittleBritainTv2Ep1.htm not there any more! Under sizing of equipment cannot meet specifications; Gross oversizing of equipment costly, may not meet specifications, may have operability problems Inefficient operation e.g. higher than expected losses of a solvent, increased cleaning frequency, higher energy costs, ; Operability problems not just from oversizing equipment, but possibly also from choosing the wrong equipment shape or configuration Have difficulty meeting product specifications have trouble selling product, need to reprocess poor product, or accept lower price Misidentifying faults and consequently taking the wrong corrective action e.g. Kister examples in section 2 Need to operate at reduced throughputs A non-viable process it just doesnt work Safety incidents (property damage, near misses, injuries, deaths) as a result of serious design faults Environmental damage (pollution of the air, land and water; injuring or killing flora and fauna)
You can imagine a matrix with the simulation elements
in the columns, and the dimensions down the rows. Possibly there is something to discuss in each cell. For a short critical review, think about which cells are the most important (i.e. prioritise them) and just mention these. Lets look at some of those references on software-related problems in chemical engineering... Best and Kister are available from the librarys e-reserve system. The IChemE report can be downloaded (see references list on the next slide).