Case Study- Big Data and The Internet of Things – Performance Improvement

Increase asset utilization, reducing unplanned down-times with Predictive Analytics

We are witnessing an era of unmatched business innovation; breakthrough technologies have matured, supporting these 2 trends:

• Hyper-connectivity

Every person, process, asset, and machine will be connected in ways that will disrupt the established rules around oil and gas operations. Automation will evolve, connecting both structured and unstructured data to enable collaboration and create true situational awareness. Connectivity will drive the digital energy network through collaboration

• Super computing

Supercomputing power, which until recently had been reserved for seismic data processing and 4D visualizations, is now available to carry out predictive analysis on data from across all aspects of an oil and gas operation. Decisions can therefore be made ahead of the curve based on a holistic view of potential business outcomes

Assets - the Internet of Things and Predictive Maintenance

We are overwhelmed by information, not because there is too much, but because we don’t know how to tame it. Information lies stagnant in rapidly expanding pools as our ability to collect and warehouse it increases, but our ability to make sense of and communicate it remains inert, largely without notice

Stephen Few, Now You See It

For example, only a small percentage of data from a typical oil rig with approximately 30,000 sensors is currently used to make decisions. A great deal of additional value remains to be captured, by using more data, as well as deploying more sophisticated IoT applications, such as performance data for predictive maintenance.

Performing maintenance with a focus on timeliness, acting exactly when needed instead of at regular intervals, and predicting and preventing failures before they happen, based on learning from historical data. Predictive maintenance—just-in-time maintenance—will massively transform how organizations and consumers manage equipment and collaborate with employees and suppliers. Predictive maintenance also informs more traditional preventative maintenance patterns, optimizing routine maintenance activities.

With sensors and connectivity, it is possible to monitor production equipment in real time, which enables new operating models that can be far more cost-effective, improving both capacity utilization and asset productivity by avoiding breakdowns. Essentially, IoT in combination with advanced analytics techniques help to predict the failure of mission-critical equipment or assets.

PEMEX depend on the safe and reliable operation of their physical assets. The cost of unplanned downtime can be significant, and traditional approaches to ensuring high levels of equipment utilization have reached their limit. Predicting equipment failure does not always require data from the equipment itself. Effectively inferring equipment failure often involves analyzing large volumes of unstructured exogenous data, such as production data, ambient temperature, and data from peripheral equipment.

These advanced analytics will expand beyond measuring and describing the past to predicting what is likely to happen, optimizing what to do and how to operate based on an increasingly varied set of data sources and types.

PEMEX is starting a transformational journey using these digitally enabled advances to reimagine roles and ways of working that exploit new digital technology such as machine learning

Case Objective: To boost heavy and extra-heavy crude output in the Gulf of Mexico’s Campeche Sound area by implementing a predictive maintenance system for the Electro-Centrifugal Pumping system, known in Spanish by the acronym BEC

• Elimination of redundancies and execution lag times.
• Workforce engagement that includes both employees and contractors
• Supplier collaboration; procurement direct, indirect and services
• Reducing unplanned down-time

Basis and Conditions

• Available data is collected in high-fidelity time-series from a diversity of sources - sensors and technical components and stored in a Pi System every minute
• The PI System centralizes both time-series and event-based data in real time

Problem:

 Pump Failure-Stops the BEC System; estimated down-time:

• Procure a new pump (72 hrs.)
• Logistic to set up the required equipment to replace the pump (96-144 hrs.)
• Replace the pump (48-72 hrs.)
• Well Downtime (144-216 hrs.)

Hypothesis:

Anticipate the system failure with at least one week notice to be able to prepare all the required logistic to minimize the downtime.

Figure 1: Actual Situation

Each square represents 24 hours; the logistic starts right after the system failure and the entire operation can take up to 8 days

Figure 2: Hypothesis

Predictive analytics sends and early alert for a probable failure in a specific timeframe, the data science team evaluate the alert and program the required logistic in anticipation, spare parts, equipment and required labor are at a specific location at the moment of the event, minimizing the downtime impact to 2 days.

Exploring other uses of IoT and Advanced Analytics

Stick-Slip Prevention - predict possible stick-slip events during the drilling process allow the drilling crew to make just-in-time changes to the operation to minimize the stick-slip event, thereby increasing the average rate of penetration.

Remote Asset Performance - highly scalable, cost-efficient method for storing and remotely accessing real-time information will help extend equipment lifecycles and optimize productivity

Anomaly Detection - detecting outliers in the data. The data consists of 'normal' applications and 'risky' applications. Risky transactions are assumed to be anomalous and dissimilar from the training data.

Energy Forecasting - Forecast consumption, reduce outages, and monitor assets to improve efficiencies

References: Microsoft, SAP, Gartner, Cisco and PEMEX Innovation Team