Stimulating SIMULATION

Aspen Technology acquires SetPoint and DMCC for $15.8m and $26m respectively. SimSci buys Biles & Associates. Aspen announces trade-in program for Biles' customers. AEA Technology buys Hyprotech in a £34m transaction. Aspen again acquires SAST, whose OTISS simulator is a central platform of SimSci's PROTISS flagship product. Simulation users, not surprisingly, want to know what on earth is going on.

What they are witnessing is a power struggle in an industry which is rapidly maturing. It's a fact. Simulation has become big business in the 1990s. Industry sources put the figure for engineering design and operational simulation software and related services at about $500m, with the design component accounting for about $160-170m of this.

The acquisitions are symptomatic of the jostling of large companies to acquire a dominant global share of this burgeoning market. Recent flotations of Simulation Sciences (SimSci) and Aspen and the takeover of Hyprotech by AEA mean that all of the `big three' simulation software providers, are now publicly owned, and they are using their new-found financial muscle to assemble broad-based solutions with simulation at the core.

Aspen Technology's stated aim is to provide `software and services for the analysis, design and automation of process manufacturing plants', and it has been pursuing this goal by concerted acquisition of companies with suitable technological base and market share. SimSci has also pursued the acquisition route, but to a lesser extent, while Hyprotech has concentrated on, as marketing manager Don Mahoney puts it, `wiping the slate clean' and developing a coherent single platform which is well-placed to take advantage of the continuing advances in current software technology.

Central to Aspen's DynaPLUS, SimSci's PROTISS and Hyprotech's HYSYS dynamic simulation products are comprehensive model and properties libraries, which package extensive engineering expertise and make it commercially available. These products all aim to make dynamic simulation available to the `non-specialist' and ease the transition from rigorous steady-state models to fully-configured dynamics.

Packaging problem

This `packaged' approach has spawned some debate in the industry concerning the desirability of non-specialists performing studies which touch on critical safety and operability areas, and the commercial benefits that this type of simulation can realistically deliver. Costas Pantelides of Imperial College and a director of the newly formed Process Systems Enterprise, which recently acquired rights to Imperial's gPROMS package, feels that `operating companies can only gain real competitive edge by embodying their own experience in simulation models.'

There is also the suspicion that, despite the best efforts of the vendors, rigorous dynamic simulation with its specialist engineering input and extensive data requirements will never be a tool for the casual user. Mahoney says that Hyprotech has anticipated this and its forthcoming HYSYS.Process and HYSYS.Plant packages will cater for different levels of engineering detail within the same general framework.

At the other end of the dynamic simulation spectrum are a number of modelling environments widely used by specialists within the operating companies and their consultants, including Aspen's SpeedUp, Matlab's Simulink, gPROMS, VisSim and ACSL. These take the opposite approach to the `packaged' solutions, providing toolkits of modelling and numerical solution techniques and analysis tools which allow specialist users to concentrate on describing the engineering detail. According to process control consultant Doug Scott of Doug Scott Associates, these are the tools which need to be used to pursue investigation into the far corners of the operating envelope, whence spring most of the benefits of dynamic simulation. `We have never had a job which did not require some novel modelling,' says Scott. `The problem with packaged solutions is that they incorporate past knowledge, which does not cater for novel situations.'

Costas Pantelides believes that increasingly detailed simulation is the key to the future. `Most simulation is essentially about operation,' he maintains. `But many processes have already been optimised to the point where there is only of the order of 5 per cent of the original scope left.' This requires models that can deliver accuracy of within 1 per cent, allowing simulation accurately to explore the margins of operation. He adds that `recent advances in understanding the physics of processes, coupled with advances in numerical solution techniques and software and hardware now make this type of simulation feasible.' gPROMS embodies this approach with comprehensive facilities for modelling distributed systems and imposing operating procedures on the models, allowing it to be used, for example, for dynamic optimisation of batch processes.

A third group of simulation providers are the turnkey simulator vendors, such as Honeywell Hi-Spec Solutions, ABB SimCon, Aspen/SAST and MDC Technology. These mostly provide large-scale, high-end applications within the large oil producing and refining companies, though chemicals, pulp and paper and power generation are also well covered. Their specialists use their own tools to provide customised solutions for training or on-line optimisation.

Business benefits

However, whichever approach they opt for, most users agree on what they want from simulation - business benefit - but there are a number of views on how to achieve this. Ron Murray, Operations Support Manager at BG in Reading, speaks for many when he says that a key business goal in operating companies is to `sweat their assets', using detailed simulation models to optimise plant economics and investigate problems in performance. Mike Knight, a Principal Consultant at BP Oil, believes that the `single model' concept - using the same basic model from conceptual design to operation - is central to maximising the benefit from simulation. John Wheadon, control systems group manager at Dow Corning's Barry plant, says that despite recent developments there is no single product that caters to all their simulation requirements as an operating site. He is not in favour of the `monolithic' approach, and believes that `small companies will continue to provide vital niche products... we need an open approach which lets all the components work together, while allowing each company to build on the strengths of their product'.

Many larger companies such as BP Oil and Bayer have formed strategic alliances with vendors to provide simulation technology into the 21st century. BP's Knight says that strong partnering brings key benefits to both companies. `The way forward is an alliance with management at the highest levels. What we want is somebody who can deliver the whole business benefit,' he says.

A problem for many potential users is that the major simulation products concentrate largely on the traditional simulation areas of refining, oil-and-gas and petrochemicals, and significant sectors of the process industries - such as food and drink, pharmaceuticals, fine chemicals and water processing - are not well served. Though these industries do not generally require high-end simulation, there is often a need for `feedback' modelling to provide directional dynamic process response to control system design and development, or for detailed modelling of specialised processes. Cape Software's VPLink fills a niche requirement here, allowing creation of simple or highly-customised simulation models, optionally connected to off-line DCS or PLC/SCADA systems for control software checking and operator training. Packages such as gPROMS, with its batch modelling capabilities and the potential for modelling highly-sequenced operations, are also particularly well-suited here.

Besides the high-profile products, there is a plethora of other simulation products available from specialist vendors or via the Internet. These range from design programs for gas sweetening systems to distillation tray simulators, and are often in-house solutions which have been made commercially available.

Given the number of different approaches, products and user requirements, there are concerted efforts to provide a set of standards to allow `interoperability' between software components provided by different vendors. This theme comes up time and again in conversations with users, who see it as the key to a coherent approach to simulation where `best-in-class' tools can be used wherever possible.

The two main standards are STEP AP231, catering for the needs of the wider computer-aided process engineering community, and Cape-Open, an EU-funded consortium of simulation vendors, major users, academic institutions and specialist consultancies. Concerns have been expressed that conflicting standards may emerge for reasons of short-term expediency rather than long-term benefit, and there is a desire to see Cape-Open operating within the STEP framework.

Interoperability will help pave the way for widespread use of `component-ware', such as InfoChem's Multiflash suite of physical property routines designed specifically with integration into simulators in mind; single-purpose packages such as Taftan Data's steam properties, available on the Internet; and, of course, the old standbys such as Excel, which can be tailored to provide process and data visualisation facilities.

What the dynamic simulation world will look like in the 21st century is by no means clear. What is certain, though, is that there will continue to be a shake-out at the top, and many smaller vendors will go to the wall. However, interoperability offers a ray of hope to the latter, and the continuing developments in the wider world of computer software, such as object-embedding and distributed processing, will help provide solutions that can only be sketchily defined at present.

Mark Matzopoulos is md of Dynamation, a specialist dynamic simulation consultancy. One of the authors of Imperial College's SpeedUp package in the 1980s, he is also the editor of the IChemE CAPE Subject Group newsletter.