PAINT your wagon
15 Jan 2000
It was different in Michaelangelo's day. He only had to mix up enough egg tempura to paint the particular section of the Sistine he was working on at the time. Of course, he created a timeless masterpiece. ICI Paints, on the other hand, has more prosaic concerns; producing the right coatings to protect and beautify many ranges of cars. But making the right amount of paint, of the correct colour, and when it's needed, is more complicated now than in renaissance Rome.
ICI Paints has to be flexible, because its products are needed in wildly differing quantities. Because of this, it operates three types of plant: a conventional plant, which can produce its entire product range in large quantities; a 'pipeless' plant, which targets those products with relatively low demand, such as 'retouching' paints for large ranges of cars; and 'in-can' plants, which manufactures paint directly into the final container, for that person who needs just the right shade to hide the scratches on his 1958 Plymouth Fury. But although they make much the same products, their cost effectiveness varies considerably. In their attempts to find the best ways of running these plants, explains Bob Weiss of ICIPaints at Slough, the company decided to call in some high-tech help - the Process Concept Studio (PRESTO) at Imperial College.
PRESTO digitation
Part of Imperial's Centre for Process Systems Engineering (which recently launched its own company to market process simulation software see PENovember 97, page 7). PRESTO is specifically aimed at solving 'hot' process technology R&D problems in industry as they occur. For each project, it brings together a dedicated team comprising academics from CPSE, generally specialising in systems methods and techniques; and engineers seconded from the company whose problem the studio is trying to solve. One of the places on the team always goes to someone with a background in business.
This is basically a technology transfer project the aim is to find a solution to an industrial problem in an academic setting and utilise it back at the plant. But this is technology transfer on fast-forward. Instead of the ten-year timescale typical of process systems research, claims CPSEdirector Sandro Macchietto, a PRESTO project goes from identifying the problem at the site, through setting up the team and working on the problem, to implementing the solution, in six months. In some cases, they can do it in three.
In the case of ICI, the company needed to debottleneck one of its multipurpose plants in Stowmarket, Norfolk a unit which produced a range of 73 'basic' colours of automotive refinishing paints. The project was to find new ways of operating the plant to minimise the amount of unusable product the plant made (colours between those in the specified range), and to reduce the amount of product held in stock.
The first step was to produce a model of the process. The team decided to simulate the way the plant operates in February. This is always a very busy month, says Weiss, as the plant operates in high loading for the first three months of the year and February is, of course, the shortest: 'if it works in February, it'll work all year round,' he comments.
The process has five steps, Weiss explains: primary dispersion, where pigments are milled in high-speed dispersers (HSDs)to make a thick paste; secondary dispersion, where the paste goes through a bead mill to make it smoother; characterisation, a quality control step where a sample of the dispersion is removed and tested; mixing, which involves adding a solvent and other additives to produce a final product; testing, another quality control sampling step; and filling into a variety of cans.
The process steps were modelled using a variety of simulation tools. The key stage turned out to be the first, as the current scheduling arrangement led the HSDs being loaded unevenly some were running at full-tilt while others were only idling. The team found that this could be remedied by using a 'repetitive flexible scheduling' (RFS) production strategy, where pigments are introduced to the HSDs so that every colour in the range is made at least once per cycle. Changing the way that products were 'streamed' to even out the load on the HSDs reduced this cycle time from eight weeks to three, says Weiss: 'in effect, we found out that our HSDs had 50 per cent more capacity than we thought they had.'
Other suggestions included allocating dedicated teams to each product stream. As these teams gained experience of their colour 'family' and the pigments needed to produce them, they would reduce the time taken to make a perfect colour and minimise the number of quality control checks. Combining this with the introduction of the RFS strategy, and adding a new, small HSD, reduced the cycle time to a fortnight, says Weiss.