I've been to a couple of analyst events in the last two weeks focused on analytics: with SAS and IBM. One of the things I am interested in right now is the Internet of Things and one of the important application areas around this is for preventative maintenance. Yet neither of these companies—and I would expect them to be the market leaders—talks about preventative maintenance, instead referring to (predictive) asset optimisation. And, yes, I can see that that all of these words apply and they might imply something broader than preventative maintenance but why not use plain English?
The difficulty with preventative maintenance is rather like those applications that require ontologies—you need to have customised input because the parts or processes you will be monitoring will be different across industry sectors—and that slows down implementation. In due course, I expect the vendors will come out with templates or models for specific verticals such as chip manufacturing or oil wells or pipelines, which can then be customised by individual companies according to their own needs but for the moment it is very much a question of bespoke applications based upon a platform of pattern recognition and analytics. Nevertheless, I get the clear sense that this is an area that is ramping up and we may get to see these templates sooner than you might think.
In the context of big data, preventative maintenance is one area where I am a big fan. It's not that we couldn't do preventative maintenance before but it was very hard and very expensive—now it's a much more realistic proposition. Now it is more reasonable on a cost basis, thanks to Hadoop—and, I believe, it is set to become much easier as vendors develop more and more standard vertical applications. IBM, for example, is targeting the telecommunications, aerospace, automotive, chemicals and petroleum, and electronics sectors while SAS has implementations in pipeline (valve) monitoring and in chip manufacturing, amongst others.
Actually, the chip manufacturing example is interesting because it is not really preventative maintenance but is actually asset optimisation (I'm denying my own argument!): chip manufacturers typically produce a batch of products and then randomly test samples - if there are too many failures they throw the whole batch away. The application here is to determine early on that faulty chips are being produced so that the batch can be stopped at an early stage, thereby reducing wastage. So preventative maintenance is about identifying patterns that predict failures and, on top of that, we have what we in the IT industry might call "agile production" - test early and often - involving changed manufacturing/business processes in conjunction with the preventative maintenance to accomplish asset optimisation.
So, asset optimisation is a superset of preventative maintenance. I don't think that's immediately obvious and I think it would be more useful, and easier for users to understand, if both sets of terminology were used in relevant presentations rather than just relying on the rather more opaque "asset optimisation".