At Shell Upstream Americas, Tom Moroney and his technology team are doing what they can to optimize use of the company's deepwater assets. Forward-looking big-data analytics is an integral part of the plan, as described in a recent post. But technology only gets the team so far.
An equally sizable challenge is instituting cultural change, said Moroney, who is deepwater technology deployment manager for Shell Upstream Americas, during a phone interview. For the last two years, the technology team has been using SAS Predictive Asset Maintenance to see if it can get ahead of events, identifying patterns and heading off problems before they occur. But having data showing that something is going to go wrong is far different than having the confidence to take action.
Engineers and operational managers are still a bit reluctant to trust Moroney and his models, he said. "They'll say things like, 'OK, so you're telling me there's a chance that in 10 or 20 hours I'm going to have an event. But what's the probability of that really happening? Is that a 100 percent certainty... or a 90 percent certainty?' "
An ocean of data
In hindsight (and if cost were no issue), Tom Moroney said he would have loved to have a lot more instrumentation on the company's deepwater facilities. But that's not to suggest there isn't plenty of it -- and an ocean of data to use in its predictive asset management:
- 20,000: the number of instruments taking a variety of well and platform measurements, like a pressure temperature or flow weights
- 6,000: the number of calculations performed for each measurement, with the measurements taken continuously throughout the day
- 300,000: the total number of daily calculations
- 430 million: the number of data points collected daily
The caution is understandable, given the cost, and potentially, risk of intervening, Moroney told us:
They don't want to intervene and perhaps cause something to happen that they didn't mean to -- these are highly, highly complex sets of physics. And the intervention is likely to mean spending money on something, like increasing the amount of foamer sent downhole, but not knowing if it's absolutely needed.
Some managers would rather wait for an event to take place and then deal with it. At least then, they reason, they know for certain that something really does need attention.
Despite such reticence, Moroney said he isn't discouraged that his sophisticated predictive asset management hasn't been embraced whole-heartedly yet. It's all part of the change management journey and making people comfortable with the technology. Patience is required. He said:
We're at the stage now where we're making some decisions and doing some interventions, but there's still more learning and more acceptance that has to take place because, again, we are using statistics and analytics that are extremely advanced in their capabilities in terms of the insights they can create and deliver to the business. People need to get comfortable that we're providing them credible insights that warrant action be taken.
Moroney and his team had to exhibit the same patience when it introduced exception-based surveillance to the business. Initially engineers and operating managers didn't trust the technology to tell them where in "the haystack the needle was." The journey took a long time -- roughly six years -- but now, exception-based surveillance is well embedded in the organization, he said.
That was Phase One, which took the organization from a distributed, person-dependent view of analytics to one that values the automation of analytics, Moroney said. He explained:
Now they appreciate the ability to use technology to systematize the filtering of information to make sense of it in real-time and pick up on issues and conditions and deviations from our defined operating windows -- in other words, our target performance levels. And they want to use the technology to answer those questions for us, or at least put us in the position to have high-quality data and put people in front of the issues rather than having the people have to go find the issues.
Moroney said he doesn't see another half a dozen years passing by before the organization embraces predictive asset management. "The cycles are collapsing and condensing, so while it took us six years to build out our event-detection capability I can see this next phase occurring in two to three years."
In that timeframe, Shell Upstream Americas will then have a second-tier approach to analytics, he added. "We'll not only be able to detect events but also actually to predict performance and predict reliability, which will help us gain much deeper insights about performance levels."
The promise and the potential of this next level of optimization is powerful stuff, Moroney said. The lesson he keeps in mind is good for anybody involved in an boosting a company's reliance on analytics. "You've got to understand that there's change management. You have to bring people and the organization along and make them comfortable with a complete, competitive analytics approach."
Do you apply change management principles to your analytics deployments? Share below.