You are Home   »   News   »   View Article

High performance data management

Friday, November 11, 2011

One of the main reasons why the 'digital oilfield' hasn't worked out is that companies are simply not able to absorb and work with enormous amounts of data. Oracle might be able to help, says Jay Hollingsworth

Ultimate business performance in the oil and gas industry comes down to the 'size of data set that you can operate on, so you can make faster decisions, or decisions with more data taken into account,' said Jay Hollingsworth, director of the Oil and Gas Industry Business Unit at Oracle.

He was speaking at the 5th Annual Oil & Gas Day at Oracle's OpenWorld in San Francisco in October.

The conventional approach to E&P management, which focuses on having data labelled and stored properly, which 'we've talked about for a really long time,' has the problem that it 'doesn't scale,' he said.

In other words, the volumes of data we are talking about require a completely different approach to data management, with automated tools to move data to the right place and work out what is happening.

There have been plenty efforts to make digital oilfield work over the years, to enable people to have the best available data to make the best decisions, but 'somehow it hasn't all worked out,' he said.

'There seems to be a breakdown in every chain in this. People can't even get the answers to simple questions, such as, 'has anything changed since yesterday.''

'So we're going to propose high performance data management.'

Mr Hollingsworth is a director of the PPDM association and technical representative at Energistics. Before joining Oracle, Mr Hollingsworth was consulting data architect at Landmark Graphics and principal data architect at Schlumberger Information Solutions.


Oracle has the tools to 'move that data from where it comes from to where it needs to go,' he says.

What is special about Oracle is that, as well as a world leader in database software, it has acquired a number of hardware and software companies over the past decade, and is putting their expertise together to make a hardware and software system for storing and working with data which runs extremely quickly.

If the objective is to process data faster, or process more data in the same amount of time, you need a system which can work with data as fast as possible.

Oracle has launched the Exadata, Exalytics and Exalogic systems, which are hardware designed for running databases and analytical tools. The database can run directly in the computer's memory, rather than have the computer memory calling for data from a separate data store. 'So it runs extremely fast,' he said.

Exadata 'is as fast as it can possibly be,' he said. 'The hardware is built to run that one thing extremely well.'

Normally, companies have IT systems with hardware and software built by many different manufacturers, which is equivalent to car with parts designed by many different companies and bolted together.

The Oracle solution, with everything all designed to run together, is equivalent to what we normally expect from a car, 'where everything was made by one company designed to work as an integrated whole.'

Mr Hollingsworth says he thinks it is inevitable that the oil and gas industry will end up using lots of Oracle products, because it is the 'only company that can provide both the hardware and software' to make high performance data management work.

'Oracle is the only company that has the hardware and software stack to solve this problem,' he said. 'The long standing problems we have in the oil industry are solvable.'

Why we need more data

Mr Hollingsworth explained some of the reasons the oil and gas industry needs more data.

Business objectives can change between wanting to maximize production or maximize financial returns, and any software business management system must be able to switch between the different objectives.

Most new drilling projects are in more difficult parts of the world, or closer to residential areas and so under more scrutiny.

High oil prices mean much more demand to reduce non productive time.

Operators are working more with offset well data, and want to provide that to drillers.

There is pressure to be able to monitor drilling and production wells in real time, particularly after the Deepwater Horizon disaster.

We have fewer engineers, and they have less experience, and need to make more complicated decisions on more wells.

Database challenges

The first problem with data is figuring out how to store it.

Data volumes have increased so much. Mr Hollingsworth fondly recalls his time at university, when he heard about a fellow student (called Michael Dell) who was running a business in his dorm room installing 5 megabyte hard drives in personal computers.

Now people talk about petabytes of data. All of this has to be managed and stored so you and your colleagues can find the right data later.

Systems have been built to store oil and gas data, but they are not developing as fast as the number of different data types is changing.

For example, if you have interpreted the same seismic data in 6 different ways, there's no obvious way to manage the different versions.

Subsurface databases might not be designed to hold gravity data and electromagnetic data.

'If you buy a commercial [database] product and we come up with a new drilling technique, that database isn't going to know how to hold that data. Those kinds of products don't deal with it because it's newer or not well understood by the data business.'

There is also plenty of data which is only partly structured, such as forms which contain free text.

'That's like the drilling morning report, where it is structured in the sense that it comes in a file and it has an organisation, but inside there are big comments field, what did you do over the last hour? 'I circulated 23 minutes, then we increased the mud weight to this.''

'I call that semi-structured data because it has a structure, but it has a column for comment or remark.'

'Managing that unstructured text is a different problem to managing your text document and your web pages.'

Real time data flows

More difficult than working out how to store your data, is working out what to do with the continuous flows of data from the field, the 'real time data'.

In the past, the 'analyse / execute' loop for drilling problems might involve someone monitoring the hook load on the drill bit from a shore office, and if there was suddenly a change (indicating a possible problem downhole), which the driller might not have noticed, someone in the office would phone the driller and tell him.

'Assuming that the relationships are such that the driller won't mind taking advice from a college guy in Dallas,' he said.

But now the data is moving much faster. 'We used to measure production monthly, then daily, then hourly, now we can measure production in real time every second or every tenth of a second,' he said.

'At each step, in order to do technical analysis, I have to integrate different data types. I have to pull data streams, and I need additional data types,' he said.

'You need infrastructure that can enable you to capture, analyse and execute data at different time scales, depending on if you're doing reservoir characterization, drilling or production.'

A driller might want to take immediate action based on the data. 'If he can see a screen that says you're about to enter a thief zone he's got to act on that data now.'

A reservoir engineer might make decisions over a longer period of time. 'If you're drilling along and you accidently start to leave the zone, you'll be able to figure that out over the course of the day and adjust accordingly to put that bit back.'

Then there are long term decisions that can be made on drilling data, if you have stored it. One company recently published a SPE paper about how it took many years of drilling data to look at performance of individual drilling crews, and found that some crews were much better than others at certain jobs. 'You can only do that if you have access to the full real time data over multiple wells.'

It would also be helpful if the cost of working with real time data would reduce. Many companies have real time operations centres, but find them so expensive to run, they only use them on the most high risk wells, he said.

Oracle solutions

To automate the real time data management, Oracle acquired an 'amazingly cool piece of technology' called Real Time Decisions, which can help you work out which are the variables you need to be keeping most attention on, he said.

'It watches the decisions that you made and updates itself based on the choices that you made,' he said. 'It is capable of learning through the life of the enterprise.'

'You decide, in a certain situation, you're going to pull up and circulate for a while. It notices that if you do that often enough, it will be saying, the last 4 times you did this you pulled up and circulated for a while, maybe you should do that now.'

Oracle is also working with different types of database structures which can compare and manage data in different ways, with systems known as 'NoSQL' or 'not only SQL'.

So you can look at all the data which relates to the same depth, or the same point in time, no matter how the data is actually stored.

Oracle Site Hub is a system which organisations can use to get all of the data related to a specific geographic location together out of other systems.

Oracle Data Integrator is a tool which is designed to help move data from one place to another periodically.

Oracle GoldenGate is a tool to manage real time data integrations and keep systems continuously integrated.

The Oracle data mining tools can run directly on the database, rather than pulling data out of the database to examine it. This means it can run much faster.

Oracle is developing a version of R programming language, an open source statistical analysis system, which can run directly inside the database. This language is currently being taught in university statistics courses, he says.

Oracle Complex Event Processing is a system for building applications to filter, correlate and process 'events' in real time. An 'event' can be defined any way you want, for example if someone pushes a button, or a value exceeds a certain limit, that can be configured as an 'event'.

The system can then put together a dashboard, or alert system, to tell people what they need to know. This tool can integrate with other systems, such as personnel systems or business activity monitoring.

Note: during the Oracle OpenWorld event, there were presentations from Oxy, Transocean, Santos, BP and Southwestern about their use of Oracle technology. Reports from some of these sessions will be published shortly by Digital Energy Journal.

Associated Companies
» Oracle Corporation


To attend our free events, receive our newsletter, and receive the free colour Digital Energy Journal.


Latest Edition Mar-Apr 2024
Apr 2024

Download latest and back issues


Learn more about supporting Digital Energy Journal