Log Data Management: Do You Know What's Lurking in Your Log Database?
One of the key data sources for reservoir work is digital log curves that are stored in various databases. These data are used for all aspects of reservoir work, but in our experience the curve data may not be verified for accuracy. Digital log databases have been in existence now for over 25 years, and in many cases they have been moved from one computer to another, and modified and added to repeatedly. Everyone uses the data assuming it's good. Log curves often come with a recent property acquisition and nobody knows much about them. A few problems occasionally are repaired when somebody happens to see them, but who ensures that these fixes make it back to the central data store? And what about all the time lost when people have to stop and check the data before they can get their jobs done?
Whatever the history (although I'm sure there are exceptions), what we have seen all too often are datasets with problems severe enough that the results of our petrophysical analysis would be negatively affected. So the first step in our field study methodology is always data verification. Here's a partial list of the kinds of problems we've seen:
Some log data storage systems make it much too easy to create these kinds of problems, and very difficult to repair. When we talk to people in the industry about this topic, many have their own horror story of (at best) a near miss or (at worst) an expensive mistake that was caused by decisions based on bad data. What's your story? Let us know and we'll post it on this page (identifying information removed) to help others know what to look for. We've started the discussion with log problems we've seen, but I know we haven't seen it all!
We presented a talk on the subject of log database problems at the San Joaquin Well Logging Society meeting in Bakersfield a few years ago. The slide set from this talk is available here.