British Sugar spent £145,000 in one year developing an expert system to reduce energy consumption at its Ipswich sugar beet processing factory. The payback time was 13 months.
Sugar refining is a highly energy-intensive industry: about 40% of costs go on energy bills.
The project, started in late 1993 as part of the DTI’s Energy Efficiency Best Practice programme, resulted in energy savings of about 5-10% per year, an annual average of £108,000. It also cut maintenance costs by £24,000 per year.
But what these facts fail to reveal is the uncosted time and effort. First, people have to be convinced that smart systems are a good idea, which is not always easy. Top management has to be sure an expert system will deliver tangible benefits. Plant operators must be assured they will benefit from imparting their process knowledge to an intelligent system.
Projects can run into problems if the data being collected, say, from sensors, is not accurate. With all smart software, the quality and sometimes the quantity of data is crucial.
British Sugar’s project team was aware of all this. It was also fortunate in having sufficient systems and process expertise to develop the expert system in-house.
The expert system was built using Gensym’s G2 software. British Sugar’s Ipswich factory had used the same software in 1989 to help build an expert system to control a process machine at the site.
The energy-saving system applies process engineers’ knowledge to real-time, plant-wide data in order to make decisions. In some cases, decisions are implemented automatically by the system and, in others, advice is given to operators so that they can make the final decision.
One of the main functions of the system is to give operators an overall view of the beet refining process so they are aware of the knock-on effects of any local changes they make to the energy system. Sugar beet refining is a multi-stage evaporation process in which the various stages are interlinked.
British Sugar wanted to run its factory consistently more efficiently, without extensive manual input, to do more with less energy input. It has achieved this but warns others considering this technology to look beyond hardware and software costs; these are not the biggest hurdles.