Looking forward into the past

There is no doubt that savings can be made in those cost centres most favoured by accountants - buying and stockholding of raw materials, manning levels, manufacturing methods and distribution of the end product, for example. But significant savings can still be made by under-standing and optimising the process.

However, the very thing that can point the way forward - technology - has in some cases led to a lack of hands-on understanding of the plant's functionality. Although much of a plant may be automated, the end product quality is often measured by offline techniques that may lead to delays in implementing control strategies. This is expensive, as off-spec material can be produced. Even with the most sophisticated control systems this means optimum running conditions cannot be achieved.

In optimising a plant's performance a basic understanding of the processes involved and the direct influence of any parameters on plant performance is necessary. We can illustrate this with a case study of a multi-line operation to produce a fine powder with tight particle size distribution.

The recently introduced techniques of online particle size analysis, based on the Malvern Insitec system, and the advanced computing technique of data mining can be used to optimise such a plant's performance.

The Malvern Insitec is an optical, in situ particle measurement instrument providing real-time, continuous monitoring information that is recorded in a Time History Graph; while data mining is used to find hidden patterns and trends in large volumes of data. These are expressed as decision trees or data models - a visible model of what actually happens rather than the theory.

The process steps involved in our case study are: pre-blending, compounding, extrusion, pelletising, granulation, pulverisation, classification and sieving. Material is vacuum conveyed from bins to the filter receiver, and from there via a series of valves to the pulveriser feed hopper. A screw conveyor transfers the material to the pulveriser, which reduces the particle size using a fluidised-bed, opposed-jet mill with internal air classification. The pulverised product is then fed to a dedicated air classifier in which a fine fraction is removed. The coarse fraction is then passed through an air jet sieve to give the final product.

The optimisation objectives here were to: tighten particle size distribution; maximise yield; minimise specific energy use; and maximise throughput.

Operating data was collected at 30-second intervals and included all the key process parameters. Over 29 000 records representing in excess of one week of operation were acquired.

Using conventional study methods to gain an overview of operations identified two modes of operation with significantly different electricity demands in the data set. The data was analysed using data mining techniques to identify patterns that explain the variability in the key performance parameters.

The significant factors identified were the pulveriser bed pressure and the speed of pulveriser rotors, higher speeds reducing particle size.

Correlation between the process lines identified a variance on the reheat valve output on one of the process lines. Further investigation showed that the classifier wheel gap airflows were the most important factors. By tracing this through it was found that the volume of air through these gaps was too high, and explained why the product yield on this line was inferior. After resetting the gap airflows to the correct values, the decision trees have become more consistent and the performance of all the systems is far more comparable.

This combination of data mining and inline particle-size analysis techniques illustrates how new opportunities can be identified to improve process performance. Overall, in this case operating costs will be reduced by £125,000.

Iain Crosley is Technical Manager with Hosokawa Micron.