Posted 09.12.2017 18:58 by Magnus Tvedt
Let's have look at some new ways to analyse offset wells. We'll talk about the goals of offset analyses, and give you a bit of insight to how we recast this process.
Through both writing the code for the data science and creating the work process in our software, we have gained unique insight, and that is what we are sharing with you in this article. If you find this interesting, you will find links to other articles which you will enjoy, further down.
First off, the goal of an offset analysis is to learn what are significant parameters for a well or a field.
They are typically carried out at different points of time in well planning:
It is a multi-skilled operation to do an offset well analysis, where the operational geologist often plays a central role. Drilling and well engineers support on the hole making, and reservoir and production engineers support on the hydrocarbon side of the problem.
Digital offset well analysis are data driven, and excellent for statistical models and data science. It's about screening through a multitude of variables, and coming up with anomalies and unique features based on some search parameters.
The traditional offset well has been limited to very few parameters, such as geology depths and presence of certain challenging formations, in a very few reference wells. The digital big brother takes in mode data, and can view the data from any angle.
When you combine drilling experience with data science and well prepared data, you can learn more in a few seconds than you would do in a lifetime before. You can imagine what this means for the bottom line of your company.
We allow the user to screen out 5 of the most common mistakes in well planning within the first three hours:
In 2017 this would be an up to six month project involving 4-6 specialists.
The engineering experience is what differentiate a good offset well analysis from a statistical or data driven analysis. When it comes down to wells, made up of fluids, steel, rigs and equipment, you can go the extra mile in extracting value from datasets. Our experience is that the data we have received from oil companies are often fragmented and contains errors which will mislead the users.
When we educate our engineers on data science and force our data scientists to write equations for pipe handling on the rig, we can get value out of any data set.