Speaking at the 2013 OSIsoft user conference in San Francisco last month, Chevron’s Jim Crompton announced his imminent retirement, after 37 years with the company. He is apparently in good company as 50% of oil and gas knowledge workers are to retire in the next five years. Their replacements are more likely to have 8 years experience than 37 and ‘we have left them the hard problems.’
Crompton turned to the hype around big data observing that this has been commonplace in the industry for decades. Oil and gas is good at handling big data sets although we are always looking for ways to improve. Oil and gas is a ‘very dangerous industry’ and we need to avoid mistakes by using data in a timely maner. On the deepwater Horizon, data indicative of the nature of the problem was available 45 minutes before the explosion, but operators were ‘focused on other things.’
While OSIsoft’s PI System is often associated with the downstream, it is seeing use in the upstream to provide a holistic view of real time asset data. Here data volumes can be overwhelming. Fiber sensors can produce 2TB/hour, billion cell reservoir simulators have been deployed and seismic generates petabytes of data. A modern production facility has more sensors than a refinery although it shares many components with the downstream.
A new offshore Angolan FPSO has some 80,000 sensors of which 1,500 are down hole. The upstream ‘factory’ is discovered not built—this is not a ‘retail type’ issue. There is a strong but healthy tension between standardization and system complexity. As the sensor count rises, there is an emerging problem with the information pipeline. Most users are only interested in their own output. Data is disappearing into shadow systems (or Excel!) through a lack of standards and a common vocabulary.
But there are significant successes with management by exception and in the use of virtual reality in training—used to develop standard operating procedures years before a platform is on-site. Alarm management is also an area of focus where systems can direct operators straight to a root cause. But ‘as we get better the problem is getting harder.’
Crompton believes that silos remain a problem as there is a tendency to educate and reward by function. Drillers don’t talk to petroleum engineers. Chevron is addressing such issues with a new focus on modeling and analytics, the surveillance, analysis and optimization program.
Here, a preliminary maturity analysis has shown the company to be ‘all over the place.’ With worldwide recovery stuck at around 33% of oil in place, the potential for improvement is vast. ‘We are not at the start but we have not yet finished the race.’
Ken Startz presented on two of Marathon’s PI System projects, MaraDrill (more of which next month) and on production operations support in the Piceance basin, Colorado. Marathon spotted an opportunity for an improved real time infrastructure with the arrival of a new engineering hire from a company ‘with better data visualization than [Invensys’] Wonderware.’ The new functionality was developed with PI Coresight in under a day. The solution has produced a 720% return on investment through speedier start up for operators and a reduced need for contract staff. The system also provides one year of look back production data. More from the OSIsoft User Conference.
© Oil IT Journal - all rights reserved.