As I have been promising for a couple of months now, I am going to take a closer look at GE’s Predix offering and other similar uses of big data science to optimize equipment maintenance. I have been kicking the Predix can down the road for the past three issues so here is what I have gleaned from GE’s website and our own observations on how the Predix concept has evolved over the last couple of years.
We first came across Predix in 2014 in the context of predicting machine failure. Predix was the platform that supported GE’s ‘Predictivity’ offering. But how did ‘prediction’ become a ‘platform?’ My guess is that the marketing folks may have put the platform cart before the predictive horse but I digress.
Anyhow a visit to the Predix home page certainly clears up what the platform is, a mechanism for connecting distributed machines and sensors to the cloud. How is this done? Predix, at least in its current manifestation, builds of top of Pivotal’s Cloud Foundry. Which, Wikipedia informs me, is an open source, cloud computing platform-as-a-service, originally developed by VMware and now owned by Pivotal Software, which is a joint venture between EMC, VMware and GE itself.
Users of Predix therefore get two things. On the one hand they get the data access and information needed to support engineering and operations. On the other hand, they get some hand holding from GE’s cast of thousands of developers as they venture into the brave new world of the cloud. A quick spin through Predix’s software components suggests that such help may be appreciated. And it is always a good idea to offer something for both the engineers and for IT to get their teeth into.
As an aside, and as we observed before, the cloud, along with containerized services, notionally make it possible to run your software across multiple service providers, all supplying ‘micro services.’ Turning today’s monolithic apps into decoupled micro services is another area where considerable hand holding may be required. The future of Paradigm’s reservoir-driven production optimization may be worth watching in this context. On the other hand, the soup-to-nuts nature of the platform may be scary and oils are unlikely to countenance handing over their whole IT stack to GE. But if it does work and provide access to PI, to SAP, to this and that and if, as we understand, GE has 800 developers working on Predix in oil and gas the option may be attractive.
I just got back from the EU SAP in oil and gas conference (more on that next month) where I learned that SAP is also offering data science-based analytics. ‘Yes,’ I hear you say, but SAP is about finance. ‘No,’ says SAP which has just offered its high end Vora analytics to the (real) scientists operating the Square kilometer (radio telescope) array. The endgame for SAP is similar to GE’s. In fact there is considerable overlap in the MMO* space between the two companies’ offerings. SAP also offers a platform with its Asset intelligence network and repository.
In a more modest and focused example, National Oilwell Varco is offering remote BOP monitoring and predictive analytics to users of its subsea blowout preventers. Behind NOV’s RigSentry BOP monitoring service is ‘Max’ NOV’s very own industrial data platform (see page 12).
So now, after a short spin through three service providers offerings, we have three ‘platforms.’ Those enamored of the cloud and ‘micro services’ will argue that this is not a problem, that all of the above will be consumable as micros services and that all your IT/data integration worries will be over. That supposes that the different platforms will allow for such seamless interoperability. Experience suggests that this may not be quite how things will work out. Cynics will remember a similar false dawn back in the days of business objects.
Another problem looming for platform users is data ownership. The issue of data ownership and data transfer across borders is often raised in the context of foreign countries with awkward regulations. But the issue of who owns data, who gets to keep it, and who pays for added value services is even more pressing with the kind of platform offerings above.
Claude Molly-Mitton, president of SAP’s French user group writing in Le Monde, addresses the issue of ‘indirect access’ to data. This has it that your data, once it has passed through SAP’s software, increases in value. Thus SAP expects you, the data owner, to pay (again) for such indirect use. This is not totally unreasonable as in some circumstances, indirect use can be a way of sneaking in extra licenses/seats to the software.
In the context of a supplier of equipment (compressors, BOPs) the notion of data ownership and re-use is also interesting. Even a large oil company only sees a subset of the data that comes in from a single vendor’s compressors. The vendor is in a good position to collate data across all of its kit, worldwide. On the other hand the operator has a significant amount of information about operating conditions that can inform decisions as to why a particular problem is recurring and what the root causes are. So the vendor can improve its knowledge from the operator and vice-versa.
While the big data idea catches the attention, and while the intricacies of programming access to data in the cloud are fascinating, the big issue here is less the size of your data but its ownership. The cloud magnifies the problem of indirect access to the nth power!
* maintenance management and operations.
@neilmcn© Oil IT Journal - all rights reserved.