As the drilling industry improves its efforts to capture drilling operation activities in real time, it has generated a significant amount of data that drilling engineers cannot process on their own. With this in mind, companies are now seeking to utilize real-time data through new databases and artificial intelligence (AI) algorithms that can help them process data points and make better decisions, an expert said.
In a webinar, “Drilling Real Time Prediction Environment in Saudi Aramco,” hosted by the SPE Drilling Uncertainty Prediction Technical Section, Salem Gharbi discussed the challenges Saudi Aramco faced while designing and building the infrastructure for its “Drilling Ahead of Bit” initiative. Gharbi is a petroleum engineering systems analyst with the company.
The initiative was established to develop an advanced engineering system that monitors, analyzes, and interprets real-time drilling data that may be useful in mitigating potential drilling issues such as borehole instability, stuck pipe, or a low rate of penetration. The architecture for each model centers on a processor that contains a quality control/quality assurance (QA/QC) function and components that help convert the data into a model-compatible format. The processor then sends the reformatted data to the relevant model.
The system contains eight AI prediction and correlation modules that test drilling data and generate predictive and analytic results. Gharbi said that determining the type of AI to use with each module was difficult.
“You need to customize your own model to help or design your own artificial intelligence,” Gharbi said. “In the drilling domain, most AI models are time-consuming. They require high computational power, but they also need some time. You need to have the result in real time. Even the top-shelf AI models need to be customized and optimized to fulfill drilling operational needs.”
Gharbi said when Saudi Aramco started developing its infrastructure, it had to work with several real-time drilling data service providers. Each provider operated in a different environment that required its own setup, database, configuration, and visualization technology. This variance led to a lack of a drilling exchange standard and a lack of a central visualization solution that, with more than 200 wells to monitor, made it difficult to process data points in a timely fashion.
“In the drilling domain, the big data complexity is not only dealing with the volume, variety, and diversity, but also, you need to deal with all of that in real time. The drilling operations team does not have the luxury of waiting 3 hours until an application finishes processing the data,” Gharbi said.
Saudi Aramco used wellsite information transfer standard markup language (WITSML) to transmit its drilling data. Gharbi said the adoption of WITSML ensured that the data processors followed the same standard, and by having data in one format, the company could develop a universal real-time drilling data viewer for its engineers to access.
Gharbi said integrating WITSML with its various vendor systems was a manageable process. The only difficulty came in integrating the hardware for the processors, since some of it was coded to a different standard.
“Each vendor had their own standards that they’re used to following, but the good thing is that WITSML standards are very clear, so if you have a good information technology department, your vendor should be able to follow it,” he said.
With the system architecture developed, Saudi Aramco will focus on developing models that will identify specific drilling problems and simulate possible solutions. The company is developing a simulator for bottomhole assembly and string designs, well hydraulics, and torque and drag.
Saudi Aramco Initiative Simplifies Real-Time Data Processing
Stephen Whitfield, Staff Writer
14 August 2016