Drilling Rigs & AutomationNews

Anadarko fast-tracks development, implementation of real-time data analytics system to optimize drilling process

By Kelli Ainsworth Robinson, Associate Editor

In order to provide rig site personnel with real-time data, Anadarko fast-tracked the development of an in-house data analytics system. The company was able to develop the system and get it deployed within three months, Dr Dingzhou Cao, Data Scientist at Anadarko Petroleum Corp, said at the 2018 IADC/SPE Drilling Conference in Fort Worth, Texas.
In order to provide rig site personnel with real-time data, Anadarko fast-tracked the development of an in-house data analytics system. The company was able to develop the system and get it deployed within three months, Dr Dingzhou Cao, Data Scientist at Andarko Petroleum Corp, said at the 2018 IADC/SPE Drilling Conference in Fort Worth, Texas.

Across the E&P industry, companies are looking to real-time data and advanced analytics to optimize the drilling process. “Real time data systems … have proven to be a vital tool for the drilling operation,” Dr Dingzhou Cao, Data Scientist at Andarko Petroleum Corp, said in a presentation at the 2018 IADC/SPE Drilling Conference in Fort Worth, Texas, on 7 March. In early 2017, Andarko fast-tracked the development of an in-house real-time data analytics system to optimize the drilling process on their wells. “There is a business need from our operations team; they don’t have this kind of real-time system,” Dr Cao said. “They want to get more information and make decisions in real time.” The project team was able to develop such a system and bring it online within just three months. And within 11 months of the project’s inception, the team had four modules running on the system, with an additional module in testing and two under development.

While there are existing real-time data analytics systems on the market, Anadarko opted to develop its own system in house. The operator wanted to have a system that would capable of providing all the functionality it was seeking on a single platform. Additionally, Dr Cao added, the company wanted the ability to develop additional data-driven and physics-based modules as needed. “We wanted a real-time data system with a plug-and-play analytics system so the data scientists and subject matter experts could build the model, just plug in and run it,” he said. “The system is like a playground for the data scientists.”

The system is made up of three layers: the data acquisition/real-time streaming layer, the analytics layer and the graphic user interface (GUI) layer. Typically, a real-time streaming processing pipeline consists of source data, data collection, a messaging system, real-time processing, memory storage and caching, and a live user interface. In the open source market, there are various tools and multiple vendors for each of these functions, but configuring these individual tools can add complexity to the project of creating a real-time data system.

Therefore, Dr Cao said, Anadarko opted to use Tibco Software’s StreamBase complex-event processing platform to build the data acquisition and streaming layer of their real-time data system. StreamBase provides a single platform for data acquisition and processing, reducing the time and complexity associated with getting the real-time data system up and running. It has a WITSML connector, which queries real-time data from the WITSML server and feeds it into the system’s analytic layer. The system also obtains static data inputs – such as information about the bottomhole assembly and drillstring components – from Anadarko databases and from user input.

Modules that have been developed for the analytics layer of the system include: drilling analytics recognition (DAR), sliding drilling guidance system (SDGS), real-time torque and drag, real-time hydraulics, and wellbore trajectory correction.

The DAR module uses one-second data to identify the drilling activity that’s taking place – including reaming, drilling, tripping in/out and sliding. This categorization is same across all drilling modules. The SDGS module automates directional calculations and notifies the directional driller about the motor yield, bit position projection, suggested next slide length and a summary of the toolface in past slides. The torque and drag module plots estimated torque and drag against actual numbers. Deviations from the expected torque and drag indicate a potential issue. Using a machine learning algorithm, the RDGS module leverages data from offset wells in a pad or field to develop a best composite well the driller can follow on a new well. The real-time hydraulic module combines up-to-date fluid rheology measurements of the mud system with real-time rate of penetration and pressure measurements to calculate equivalent circulating density. Finally, the wellbore trajectory correction module instructs the driller on how to get back on track if the actual well path deviates from the planned well path.

The final layer to the system is the user interface. Tibco Software’s Live Datamart platform stores and caches the calculations from the models and streams live to a web browser.

Future work on the platform, Dr Cao said, includes reducing the latency associated with using WITSML data. “With WITSML, we get one second data with a delay of 10 to 15 seconds,” he said. “For the second generation of our system, we may switch to another connection.” Further, he added, future iterations of the system will be designed to make better use of historic data.

For more information about this project, please see IADC/SPE 189595 “Rapid Development of Real-Time Drilling Analytics System.”

Related Articles

Leave a Reply

Your email address will not be published. Required fields are marked *

Back to top button