Oracle - raising the game for oil and gas data
Thursday, May 24, 2012
the oil and gas industry's IT systems could be better at working with real time data, managing new data types and providing analytics, says Jay Hollingsworth, director of oil and gas with Oracle
The oil and gas industry's IT systems are not able to handle many new data types, don't connect real time and static data well, and don't allow very good real time analytics of drilling data, said Jay Hollingsworth, director of oil and gas with Oracle, speaking at the Digital Energy Journal March 13th conference in Aberdeen on developments with subsurface data.
'When you talk about industry challenges with IT people, they are all surprised that we have some of these problems,' he said.
Mr Hollingsworth joined Oracle in autumn 2011, having previously worked at ExxonMobil, Landmark and Schlumberger,
IT systems cannot yet manage many of the data types the industry is now using, he said.
'People are taking all this microseismic data with shale plays. What's the industry standard for holding microseismic data? There isn't one.'
'What's the industry standard for Controlled Source electromagnetic (CSEM) data, distributed temperature surveys? There isn't one.'
'There's a lot of new data types people are really excited about using, but the industry has got ahead of our ability to handle that information.'
'The [subsurface] databases people are most familiar with are a year or two behind [being able to hold it].'
There are similar challenges in the production arena, with lots of new data types, such as continuous temperature readings, but no standards for how the data should be managed, he said.
Static and real time
A major challenge with subsurface data is incorporating static models together with real time data.
This challenge goes in two directions - with drillers wanting to use the subsurface data models together with their streams of drilling data, and subsurface modellers wanting to incorporate real time field data into their models.
When it comes to bringing real time data into earth models, the current subsurface interpretation tools can take an input of real time data, but 'That's opening a WITSML channel into your earth model, not really managing that real time data together with your earth model data,' he said.
'The geoscientists want to bring the full suite of real time information back into the earth model. That's something the industry isn't doing very well.'
Another problem with subsurface data is that much of the software is provided by companies which also provide oilfield services.
'Whether it's true or not, there's a perception that if I rely one [subsurface software provider], I'm going to have to use their directional drilling services or wireline logging services,' he said.
'There is a desire to have a kind of neutral party.'
Perhaps the biggest area where IT systems are letting the industry down is in how real time drilling data is managed and analysed.
Some drilling companies send real time data from the drilling rig to a shore office, where someone watches it.
But these real time data centres, as they are currently set up, often provide such little value that oil and gas companies are starting to question whether the cost of them is worthwhile, he said.
It would be more useful to have the data displays in real time operations centres actually available on the rig. Then the driller can see what is happening for himself, rather than trust what someone is telling him on the phone.
'How many drilling rigs have you been on where, in the dog house in the drilling rig, they see the same data display that someone is able to see at the operations centre back in the office? Not many,' he said.
'The drillers crave having the same information that the college kid in the office has. For whatever reason, we as an industry aren't doing this.'
'That driller is not interested in what [someone working onshore] has to say about what is happening on his rig.'
Another problem is that oil service companies charge extra if the operator wants to keep the real time drilling data.
'You pay a service company to drill your well, you pay a service company to steer your well to where it needs to be, you pay a service company to tell you where the well is in space.
'You can pay the service company to do logging while drilling and tell you if you're still in the reservoir zone you think you are. You can pay them to send the data, you can pay them again to put a person in front of a wall full of monitors.'
'When that data leaves the screen it dribbles off the floor. Operators don't usually get a copy of their real time data that they've paid someone to drill and monitor.'
'If you want a copy of your data, you get to buy that back from the service company.'
Mr Hollingsworth said he only realised how many people at operators felt about service companies, when he left Schlumberger to join Oracle in Autumn 2011. 'It was surprising to me, I didn't realise how people felt about it,' he said.
Real time drilling data would also benefit from more analytics.
'It takes a bit of experience to watch that on the screen and infer that something is happening,' he said.
Why don't companies develop technology to interpret drilling data and make it easier to work with? 'The fact that we haven't applied technology to that is a problem,' he said.
Banks and supermarkets have worked out how to do analytics on continuous streams of data, and oil and gas companies can do the same thing, he said.
'It seems pretty simple but for whatever reason we haven't really done this as an industry.'
On the rigs, there have been some advances in drilling data systems. Some companies are installing process control systems which make constant adjustments to keep the weight on bit constant, to optimise rate of penetration. 'You no longer have a driller with his hand on the brake,' he said. 'Why do we need that anyway?'
The financial case for investing in better IT systems for drilling is quite easy to make.
'Anything that you can do with data to make it possible to do a drilling operation with fewer workers on the rig is a good thing,' he said.
'Anything we can do to help reduce the risk of loss of reputation, on behalf of the companies which are conducting the drilling operation, in those very public environments, that's a big driver in the industry right now.'
This is a particular issue with shale gas operations in the US, where drillers are finding themselves drilling very close to homes, and have people watching continuously with cameras waiting for someone to make a mistake.
With unconventionals, there is a massive focus on reducing non productive time, with some wells being drilled in under 8 days. 'We talk about well construction not drilling,' he said. 'They talk about well manufacturing.'
There have been people selling data analytics tools saying 'just point our analytics at the seismic data, and you can fire your geophysicists, our analytical tools are so great that we will find all the oil and gas for you,' he said.
You should be wary of such solutions, because the data analytics tool doesn't know what to look for unless there is an oil and gas person to explain, he said.
As an example, if you are pulling drill pipe out of a hole, the hook load should reduce by a certain amount with every pipe length which is taken out.
'Most people's analytical software are capable of making an alarm, tell me if i'm about to collapse the derrick - I've exceeded 60 per cent of the collapse load of the derrick - everybody's system can handle that,' he said.
'But most people's systems can't handle is - if the curve kicks up slightly, which shows i'm pulling up through a part of the hole that's collapsing.'
'It requires someone with E&P knowledge to say, that's problem worth looking for. A business analytics solution isn't going to find it by itself.'
'You can tell it what pattern to look for - now you're applying the E&P knowledge to the tool.'
Big data is a fairly new technology term, which means a specific type of data, where the value is found from the analytics, not individual components of data.
In this way, a list of bank transactions, where every piece is valuable, would not be considered 'big data', but a database of online purchases used to work out what someone might like to purchase next, would be.
An example of big data would be a stream of temperature data from a well. Each individual reading is not worth so much, but by analysing how the temperature changes over a long period, you could get useful information, such as which factors cause the temperatures to change.
Oil and gas companies often reduce the temperature data to a reading every hour to make the data volumes easier to manage, but in doing so lose a lot of the value.
'A surveillance engineer could make use of high rate data but he typically doesn't have that available,' he said.
Big data typically has lots of varieties, including structured data (databases), unstructured data (documents), and something in between, for example when a report has fields in it which contain free text, like 'Toolpusher shot a cow and had to buy the rancher a set of tyres to make up for it.'