Automated Large Data Processing: A Storyboarding Process to Quickly Extract Knowledge from Large Drilling Datasets

摘要

Substantial volumes of data are collected during modern drilling operations. However, the business value of such data is limited unless it can be analyzed quickly to derive practical knowledge for application on subsequent wells. The sheer quantity and messiness of data can overwhelm oilfield personnel, making it difficult for them to extract value. An automated process is necessary to extract knowledge quickly and efficiently from large datasets. Our team identified a preliminary set of 12 questions with answers that provide immediate knowledge to help improve the drilling of subsequent wells. Each of these ten questions is best answered through a storyboarding process. The process involves the automatic creation of a series of one-page visuals with just the right amount of information on each page to validate the answers to the questions. Standardizing the structure of the data (well-site data, survey data, geology data, well plans, etc.) enables software to rapidly create these visuals and is an important step in the process. This work describes how the storyboarding process was applied to a dataset of more than 100 gigabytes (GB) from 16 shale wells drilled in North America. Examples of questions that could be quickly answered using the process are: ‘What was the best drilled well on the pad?’ and ‘Did a particular bottom hole assembly (BHA) improve drilling in a particular section of the well?’ Scripts were written in Matlab and Python to automatically process the raw data and generate more than 20 different types of one-page visuals that are well suited to present the answers to such questions. The illustrated information includes insights into BHA performance, wellbore tortuosity and quality, vibrations, weight on bit transfer, and other drilling dynamics. Identifying the relevant KPIs to satisfactorily answer the questions and present exactly the right information from the vast amounts of data was a challenge. This paper documents and describes the concept of storyboarding that uses visuals to answer comprehensive questions. This concept is not yet widely applied in the drilling industry today, but is expected to be quickly adopted by stakeholders interested in drilling performance improvement and cost saving opportunities.