You are Home   »   News   »   View Article

Improving drilling - Intelligent Energy

Thursday, September 13, 2012

The SPE Intelligent Energy conference in the Netherlands included a range of technical papers on how to improve drilling performance with better data and better use of data - reports of papers from employees of ExxonMobil, SINTEF, Saudi Aramco, Chevron

Speaking at the 2012 Intelligent Energy conference in Utrecht, Matthew T Prim of ExxonMobil said that his company has managed to increase rate of penetration by 80 per cent in some cases, with the help of a drilling performance management system.

The paper was SPE 150208 'The Critical Role of Digital Data in a Physics-Based Drilling Performance Workflow.'

If you ask drillers how to drill faster, they always say 'more weight on bit', he said. But you eventually get problems, such as bit balling (cuttings sticking to the bit); bottomhole balling (when the drilling fluids can't carry the cuttings away from the drillbit fast enough), or vibration.

With better data, you should be able to work out how hard you can drill without running into problems.

There are many new data sources you can work with, including photographs from downhole, photos of cuttings and cavings, measuring torque and drag on the drill bit, gathering log data, monitoring cuttings.

All of these digital data streams are evolving, he said.

For vibration it is possible to work out exactly how much the energy losses are.

Pressure while drilling (PWD) data gives an indication of cuttings in the fluid.

To make it work, 'we need a deep understanding of the source, quality, filtering of the data,' he said.

To monitor if the drill bit is jerking ('stick-slip'), it is important to take surface torque at high frequency. Conventionally, surface torque is measured every 20 seconds, but the jerks can be only 3 seconds long and impossible to detect this way. So Exxon mandates that all of its rigs collect torque data every second, he said.

There are 'almost definitely' stick slip cycles without a shorter period than 1 second, he said.

Mr Prim made a differentiation between 'complicated systems,' which have more parts than the brain can process, and 'complex systems' which are not understood well at all.

'We can't automate what we don't understand,' he said.

In order to use digital data in a performance management workflow you have to reduce the complex to the merely complicated, he said.

'There are so many things going on - to ask a small group of individuals to focus on all the limiters is impossible,' he said. 'But we do believe that specific modules are suitable for automation.'

SINTEF - Better drilling data

Roar Nybø, PhD research scientist, SINTEF, said that there isn't much sign of improvement in quality of drilling data. "Unlike computing power or data storage capacity, data quality does not follow Moore's law, it is in the same place it has been for several years," he said.

His paper was SPE-150306, "The Overlooked Drilling Hazard: Decision Making from Bad Data," written jointly by SINTEF and Shell. The paper is a summary of experience with real-time drilling simulators in an operational setting, gained over five years in the Center for Integrated Operations in the Petroleum Industry.

Many people in the industry are interested in real time drilling simulators, a computer model of the drilling which is continually updated using live data, because you can use the simulator to get a better idea of what is going on.
But these developments are being held up by poor data quality, he said.

A common problem is when a new piece of background data is provided just before drilling starts, such as information about the BHA configuration and there isn't time to verify or correct the latest information. "We sometimes had to stick to the old data, because the low data quality meant the new data could be even more misleading", he said.

Then once the drilling starts, there are often problems with sensors and middleware, he said.

Operators also rarely record real time data about their drilling. "In our experience the operator gets access to the real time data during the operation, but often leave permanent storage to the service company," he said.

On one project, SINTEF (working on behalf of an operator) tried to track down the person at the service company who had the real time data to ask if they could have it.

"He was shocked, no-one had ever asked him for data. Then he said 'This data will need quality control before I hand it over, who's going to pay for that?'. This issue had simply not come up before.

"If you have real time data you can feed it into a model. If the result doesn't fit, there's either something wrong with the model, something wrong with the data or something wrong with the well. This means the model output is a versatile resource, which can be used for both automatic quality control and reduction of false alarms in alarm systems. In addition to decision support, which is of course its main purpose.

"But just like any other equipment, models need calibration" he said. "What you find is that ordinary drilling operations don't give you an opportunity to calibrate your model. Then you can't even tell if you have bad data or not, because you have nothing reliable to compare against."

It is frustrating that people commonly see data quality as an IT issue. "But bad data quality is not a computer problem it is a drilling hazard and should be treated accordingly", he said.

"Nor is it just a technology issue; it is also a people and process issue. We can draw an analogy with the computer industry. While computers are getting faster at an exponential rate, you're not seeing the number of software bugs falling exponentially. Avoiding and fixing bugs take human effort"

"Similarly, faster computers and broadband connections to the platforms are not going to solve the data quality issues, we have to do that ourselves."

Among Mr Nybø's recommendations is to start using the data, even if the quality is not good enough initially. "Establishing the human connection between those who produce and those who consume data is just as important as connecting the right cables. People need to know that the data they provide is used, by whom and for what purpose. In a busy operation, nobody would prioritize the quality of data that nobody asked for." The human connection is also vital for providing real-time feedback on data quality, so that it can be fixed there and then.

That's not to say technical improvements should be ignored. Wired drill pipe, which can carry a lot more sensor data from downhole and sensors along the string, provide a lot more redundancy in the data, letting you both calibrate your model more accurately and spot if a sensor is malfunctioning, he said.


Majid Al-Shehry, system analyst at Saudi Aramco, and member of the WITSML Executive Team, talked about the developments with WITSML.

The paper was SPE-150278, 'WITSML: Laying the Foundation for Increasing Efficiency of Intelligent Wellsite Communications,' written by staff from Digital Oilfield Solutions, Baker Hughes, Schlumberger, Halliburton, Saudi Aramco, Petrolink and Energistics.

With WITSML 1.4.1, Energistics wanted to simplify the process, he said.

So for example instead of having separate objects for different logs, doe MWD/LWD, mud logs, wireline logs, and pressure pump logs, it just as one 'log' object.

Altogether the number of data objects has been reduced by 60 to 70 per cent.

Saudi Aramco started implementing WITSML 1.3.1 in 2008 with the help of Petrolink. It started moving to version 1.4.1 in 2011, testing the new data objects.

Saudi Aramco has an online system for gathering drilling data, called 'Saudi Aramco Drilling Knowledgebase' or SADK, which connects to the drilling site.

The rig sites provide real time data in a variety of formats, with some still generating WITS data only.

The company has centres to monitor geosteering and look at lithology in real time.

The new WITSML version is 'more robust than previous versions and has faster performance,' he said. 'It offers more strong functional support.'

The new WITSML version can also be compressed using GZIP, which should make it faster to move data, he said.

WITSML 1.4.1 should also make data easier to manage, which should lead to easier quality control, commented one audience member.

Integrating WITSML and PRODML

William McKenzie, IT Strategist at Chevron, talked about the work Energistics is doing to integrate its standards together.

The paper was SPE-150057, 'Integrating WITSML, PRODML & RESQML for Cross-domain Workflows,' written jointly with Total, Paradigm, IFP, Computer Modelling Group, Energistics and Schlumberger staff.

WITSML covers data exchange for drilling; PRODML covers production data and RESQML covers reservoir data.

PRODML is seen as the 'middle child' in the 3 main Energistics standards. There are still people in the market asking for explanations about why they should be using it, he said.

To develop the standards, Energistics has a range of 'special interest groups' or SIGS, covering discipline areas such as drilling and completions, production, reservoir management, asset and data management.

Some people sit in several SIGs, which helps integrate the work across different disciplines, he said.

Also, each 'ML' has its own steering committee, executive team and functional work group.

'We've been successful with 'MLs' because we've broken the system down, that has enabled us to come up with far more useful standards,' he said.

'We all want the same thing, to share data and for data to move seamlessly across boundaries.'

Involving the SIGS is important in any new initiative. There were efforts to develop an all encompassing 'ML' called 'EnergyML' but this initiative is 'gone away' because 'it was presented in the wrong way,' he said.
'If you don't make people part of the process you're going to fail.'

Energistics is moving towards a structure where there the executive team of the Energistics has representation from each of the special interest groups.

Companies often use the standards together, for example when Chevron is doing gas lift it uses both WITSML and PRODML, he said.

To make this easier, there is some commonality of data objects between the standards, for example each 'ML' has the same data object for wells.

The RESQML exchange standards for structural models can use drilling data from WITSML, covering the wells and wellbore, he said.

Associated Companies
» Intelligent Energy Event
» ExxonMobil
comments powered by Disqus


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


Latest Edition Jan-Feb 2024
Jan 2024

Download latest and back issues


Learn more about supporting Digital Energy Journal