Slaying the downtime dragon

Posted on 8 Dec 2010 by The Manufacturer

A team from the University of Portsmouth has created software which might just banish the tyrannical beast that is downtime. Mark Young explores.

You ’ll often find in fairy tales there’s a fire breathing dragon persecuting the mild-mannered village folk and it is the only thing standing in the way of the self-sustained serenity which will form their happily ever after. Manufacturing, of course, is no fairy tale.

There isn’t just one tyrannical beast piling misery upon the masses, there are lots. The tax man is one, for sure; the unions, perhaps, are another; downtime, certainly, rivals them as a third.

In the fables, a hero emerges to slay the dragon and liberate the rejoicing townsfolk who celebrate in joyful song and dance, coaxing yellow rays of sunshine through the storm clouds. Recognising vanquishing HM Revenue & Customs, would only land him in jail and that the unions, as much as they might disrupt, do actually serve a very important role in society, our dragon slayer – Dr David Brown, Head of the Institute of Industrial Research (IIR) at the University of Portsmouth – chose downtime as the target for his lance.

If manufacturers want to stay away from the dragon’s fire of unscheduled downtime at the moment, the best solution anybody has come up with is counting hours so that parts can be replaced before their expected end of life. Clearly, this is not an efficient model. It doesn’t take into account subjective things like machine overloading, which the company may not be aware of, and parts could well be replaced too soon, as well as too late.

However, Dr Brown, backed by an 11-strong band of merry men comprising research associates and PhD students, has created a ‘virtual engineer’ – a piece of diagnostic software which informs companies when machines will need to be scheduled in for maintenance, instead of waiting for them to break down. It works by placing sensors on the machine which are prone to failure, such as motors and bearings, and picks up on changes in the algorithms of vibrations, temperature, humidity, frequency and sound. When there is a change, it means the part has deviated in some way and therefore poses a threat to the operation. Critically, the system detects change of less than one per cent – before it causes the machine to malfunction. The machine can then be booked in for maintenance at a time when the least disruption is caused, or when a back up can be drafted in.

“The sensitivity of the system is the most important part,” says Dr Brown. “The whole point is that we pick up changes early so that ample time is given for the machine to be scheduled in for maintenance. There are already fuses on machines which detect faults but, of course, they force it to stop running. That’s what we want to avoid.” Sometimes, machine maintenance providers might send an engineer out to the factory to put vibrations meters on machines and try to analyse performance this way. This is prone to discrepancy in readings, requires skilled personnel and only identifies the general area of a machine which is out of kilter. The Virtual Engineer can diagnose the specific part which is affected, be that the gears, shafts, belts, bearings or anything else within the motor. And real engineers will never be out of a job because they’ll still need to perform the maintenance. “It just makes the job a lot quicker and easier,” says Dr Brown.

Dr Brown credits his colleague, senior research associate Dr Farshad Fahimi, with the breakthrough for the Virtual Engineer with his work on image processing. Says Dr Fahimi: “It searches data like an intelligent and sensitive human, combined with the skills of a super engineer. It’s important for the operators to be alerted to a breakdown and to be able to estimate the break down time well in advance in order to check for availability of parts. This is one of the current trends in our research.” What is really impressive about the Virtual Engineer, says Dr Brown, is that, as an intelligent system, it ‘learns’ about the machine while it works.

The software records what algorithms are produced while the machine is running normally without faults so that it understands what deviations are part of necessary processes away from the primary function – automatic cleaning, for example. This prevents the system from ‘crying wolf’ and also means that companies can learn from the way one machine handles a combination of different processes to make future models more efficient.

The team are working on the project with food processing and filling equipment manufacturer Stork Food & Dairy Systems (SFDS) as a Knowledge Transfer Partnership. The importance of avoiding downtime for this company is clear. Their machines run for 24 hours a day, seven days a week, and what’s more they can cost up to £1.5m a piece so companies are unlikely to have a spare one sitting around doing nothing. The team have been testing the Virtual Engineer on Stork’s machines and Luke Axel-Berg, Stork general manager has hailed the Virtual Engineer as “an entirely new way of looking at maintenance”.

“An unplanned stoppage on a production line can be a total disaster,” he said. “It can spell chaos for a processing plant, especially a dairy plant where milk is literally arriving every single day. The cows don’t stop producing milk because a machine has broken. Instead the milk must be sent to an alternative location putting unexpected pressure on another plant.

“Our customers are already calling for a ‘zerofault’ level on their machines. Until now it’s been impossible to guarantee that level of customer service but this new diagnostic system looks set to change all that by taking away the risk.

“In the event of a major breakdown lasting several days we could even risk losing a customer – how do you put a price on that?” Dr Brown says the system is around 18 months from commercial roll out and between then and now his main priority is to make the user interface and reporting more accessible. “Currently the analysis provided is very technical but all that the user really needs to know is whether or not the machine is running OK and which specific part needs replacing,” he says.

There could be a wide range of applications for the system outside of processing machines as well, including in electric cars and in renewable energy equipment. “It’s a very arduous environment, a car, with all the rain water and all of the vibrations. We know of a small development company that are producing electric cars and we are approaching them. There is even more scope now that companies are looking at putting hub motors into wheels.

Wind turbines could be quite strong too. The ones out at sea are not easy to go out and analyse. We could place the Virtual Engineer on them and use radio devices or the mobile phone networks to communicate back a fault.” So let’s all hail the dragon slayer! There will be those who thought they’d never see the day, yet downtime’s reign of terror over manufacturers could soon be consigned to the stuff of legends. Now how about that tax man?