Data is not the problem. Good decisions are the problem.
When I was at SPE’s annual Annual Technical Conference and Exhibition in Anaheim, California last fall, I posted an article on my Actenum blog where I wrote that the oil & gas industry has realized that they’re drowning in data. I am currently in Amsterdam, participating in SPE’s Intelligent Energy conference and several presentations here in Amsterdam are addressing the same issue.
The issue is that over the past few years, new technologies have become available that have led to an exponential increase in the volume of data available to operations and management. Upstream oil and gas operations are a case in point, where there now are constant data feeds from sensors embedded in the fields and the drilling equipment, sent to the surface and transmitted via satellite to operations centers.
“When you collect data, you’ve simply added cost. You need to add decisions to add value”.
– Roger Hite, Business Fundamentals Group [1]
However, adding decisions—and, if I may add, good decisions—about operational situations remains challenging.
- The first reason is that knowing, at any time point, the location of every truck, the local stock of a certain brand of dental floss, or the pressure value at every valve, does not mean that you have a clear picture of the situation. Information overload is a real issue. Last August I attended the 25th DARPA Systems and Technology Symposium where a representative from the U.S. Armed Forces told me that they now have evidence showing that the quality of critical decision making sometimes has worsened with more data-rich support systems.
- The second reason is that a clear picture does not mean that you will know how to act. In many situations, it may be very difficult to make a good decision even if you have a perfect picture of the situation. Usually, we have many constraints, competing objectives, and ad hoc knowledge, and often the problems themselves get exponentially more difficult to solve when the number of components increases. Leaving this challenge to the user alone is often not appropriate. There are mathematical reasons for this that I shall not bore you with, but consider this: When playing chess you have a perfect picture, and it is still a very difficult game to play.
Frequently, organizations try to overcome these barrier to effective decision making by using various analytics tools, including data mining, statistical pattern recognition and decision support systems that usually are focused on finding ways of collecting, transmitting, mining, and visualizing data. However, these approaches provide only a part of the solution. They support the two lowest levels in the hierarchical process flow shown below [2].
However, the last critical piece, decision-making, is left to the user. According to to Computas AS, an Oslo-based services and integration company, the current technology focus in the market is at the bottom of this value chain. That is interesting, since the business value is increasing as you go up the chain:
This is starting to change. Some companies are investigating technologies for providing decision proposal systems. For instance, Oslo based Computas ASA is using Bayesian reasoning for managing the complexity of decision coordination, while Actenum (where I am working) is using new search and optimization technologies to handle decision complexities and provide decision proposal in operational environments.
Data is not the problem. Good decisions are the problem.
More information
[1] “Barriers to Implementation of Real-Time Optimization Technology”, presentation given by Roger Hite, Business Fundamentals Group, at the SPE ATCE conference in 2007.
[2] “CODIO—Collaborative Decisionmaking in Integrated Operations”, Roar A. Fjellheim, Computas; Reidar B. Bratvold, University of Stavanger; and Mike C. Herbert, ConocoPhillips Norge, Proceedings of the Intelligent Energy Conference 2008, SPE 111876, Society of Petroleum Engineers.