NarrativeThe generation of electricity at a competitive price is vital in any developed economy. It is, though, a highly complex process with high upfront costs. When electricity is being produced, whether it be from a wind turbine or a nuclear reactor, it is essential that the installed equipment performs consistently properly.
Failure of the hardware – especially if it is unforeseen – can be disruptive and hugely costly. This is a problem which can only increase in the future as technologies such as offshore wind farms are rolled out into commercial production. Repairing a broken component many miles out to sea is almost certain to be a time consuming, tricky and potentially hazardous business, especially in the adverse weather conditions regularly seen around the UK. If it is also unscheduled, it can be doubly difficult.
Problems in power generation are always going to occur, but it is much better if the equipment can be closely monitored, degradation spotted and failure times predicted. Repair work can then be carried out as part of planned maintenance schedules.
Engineers at Strathclyde University in Glasgow are currently developing world-beating technologies which could have a major impact in this area. Its Institute for Energy and Environment – part of the Electronic and Electrical Engineering Department – is the largest university based research institute for electrical power engineering in Europe. The Institute, which has more than 200 staff and PhD students, is carrying out end-to-end research on technologies which will help to underpin engineering sectors such as renewable energy generation, transmission, distribution and Smart Grids. One of its most exciting and potentially lucrative projects involves the development of sophisticated condition monitoring software which uses highly developed artificial intelligence techniques to assess and predict potential failure.
Professor Stephen McArthur, the Institute’s Co-Director and Deputy Head of the Electronic and Electrical Engineering Department, explains that the project involves both diagnostics – evaluating symptoms to work out what the problem is – and prognostics, which involve calculating how long it will be before that problem becomes destructive. “We are developing artificial intelligence technology which embedded within and sits on top of existing hardware and software to provide decision support to maintenance engineers,” he explains.
This type of analysis is a highly complex process. “Every single piece of equipment provides a different kind of data – a wind turbine, for instance, isn’t the same as a power transformer. “A transformer might provide vibration data, oil quality data and data associated with the breakdown of the insulation. You have to look at multiple sources to see if there is something wrong.” One of most important parts of the development process is to incorporate knowledge about different behaviours and faults. “That means the system will be able to carry out automatic diagnosis,” says McArthur, “but the algorithms we use need to be robust enough to identify and validate the fault.”
The challenge for the users of condition monitoring systems is that large amounts of data are generated with only a few experts able to analyse it. Research is often focused on artificial intelligence based data analysis algorithms which either translate human knowledge and experience into software or automatically learn fault and defect patterns from the data. The systems currently being designed and demonstrated by Stephen McArthur and his colleagues at Strathclyde combine both of these in a single solution making it a unique decision support system.
“We realised we needed a single, dynamic software architecture which allows users to plug in different technologies and get a single, useful output and still deliver robust data analysis to the end user.”
The beauty of the solution is that by delivering highly sophisticated reporting and at the same time removing the task of manual analysis, it has the potential to deliver a speedy response, efficiencies and cost savings. “It gives end users as much information as they feel they need, and they can then drill further into the data if they want to.”
The potential benefits of this are compelling. They include reduced downtime; quicker identification of potential points of failure; reduced costs; and repeatable analysis which does not require expert human intervention. The last of these is likely to be particularly useful to large utilities which may, for instance, have a number of power stations around the country but only one or two data analysis experts. If these people happen to be at one location when another reports a potential problem, resolution may have to wait until they are physically able to be on site to investigate. This obviously involves an unwelcome wait. Currently within the electrical power generation, transmission and distribution industry online demonstrators have been implemented as prototypes.
“At the moment, it’s about proving the concept and the architectures,” says McArthur.
“We use multi agent systems – in other words, software contained in modules which are able to co-operate with each other to solve a problem. This is important as no one deployment is like another. It’s been designed so that users can plug in new advanced algorithms as they become available. The aim is for everything to dynamically integrate. Anyone should be able to plug their own software in without too much development work being needed.”
The commercial potential of this new approach is enormous. As well as the traditional power and renewables industries, it is also likely to have application in other sectors such oil and gas, petrochemicals, process control and the aviation industry. The Pathways to Impact funding has helped identify future pipelines of intellectual property and accelerate and prioritise market applications. Immediate opportunities have been identified in the nuclear market and the research systems can be rapidly progressed to convert into commercial offerings.
“This is a huge opportunity,” Stephen McArthur says. “Our first focus will be on the electricity generation sector, but the market could be worth hundreds of millions of pounds or more. In particular, we are going to be part of the renewables revolution, and the opportunities are immense.”
|Category of impact||Economic and commerce|