Line-up for ON-LINE

The Americans knew it ten years ago and did something about it. And now in the UK, thanks to the DTI's Foresight initiative (see PE October), a number of organisations composed of industry and academia are guiding research in more fruitful directions than ever before.

The Centre for Process Analytics and Control Technology (CPACT) is one such collaboration. CPACT is run by the universities of Hull, Newcastle and Strathclyde, who last year won a £1.3million Foresight grant. BP Chemicals, BNFL, Glaxo Wellcome, ICI, SmithKline Beecham and Zeneca are now full members. For the first time, industry can influence the direction of new research and academia have a new source of funding. Frank Cotee of SmithKline Beecham also believes through close links with universities, CPACT will `enhance professional development of SB staff and provide access to trained and motivated graduates.'

The main problem for many small companies is finding out how to best implement the new on-line analysis technology. Small companies cannot afford their own research and are not clear where to start because of a lack of expertise. CPACT is running seven related projects to provide members with the latest on-line developments.

Danladi Mamman and Manori Weerasinghe of Strathclyde University are building a model process to provide a sounding board for calibration and new procedures. Known as project 1, the model process is being used to compare a range of measurement and control technologies in pilot scale reactors. The scaled process also provides a facility to test other CPACT projects.

Many of the calibration routines used in on-line analysis require very large data sets for multivariate analysis. Process analysts in industry are seeking calibration methods which are based on small data sets. Project 2, led by Steve Gurden, Kamal Setarehdan and Tony Walmsley, will provide process analysts with robust calibration methods, for a wide range of measurement techniques and processes, based upon a small data sets. This allows meaningful data to be produced from just a few batch runs.

Efficient industrial processes and reactors produce a batch product of desired quality, in minimum reaction time, with maximum operator and environmental safety. Research project 5, led by Guanglin Zheng from Newcastle University, is aiming to achieve reactor efficiency using multivariate analysis, dynamic neural networks and statistical boot-strapping methods which have been tested in spectral data calibrations from minimal data sets.

Running through all the projects is multivariate statistical process control (MSPC) which provides operators and plant managers with information to monitor the process, provide early warning and hence ensure consistent production.

Economic pressures and market perceptions of product quality demand that more slowly monitored `quality' measures from QC labs be predicted from more rapidly recorded process variables. In this way, the final quality of the product may be predicted as it is produced, resulting in more frequent quality forecasts. This is the focus of project 6 led by Sanjiv Bissessur of Newcastle University.

Yuchan Huang of Hull University is leading project 4 to develop a base of information about different techniques and procedures for on-line process analysis and control engineering. Huang is to produce a tutorial guide on the information a member company requires based on the results and experience of other CPACT.

{{CPACT Projects