Industrial IoT Helps Prevent Costly Unplanned Downtime at BHP Billiton

BHP Billiton, a leading global resources company, operates a broad range of industrial assets. These include offshore oil and gas production platforms at which unscheduled downtime can be extremely costly. With the help of L&T Infotech, BHP Billiton developed a predictive maintenance application for the gas compression equipment. This has helped avoid unscheduled downtime and the associated lost production and revenue.

Unscheduled Downtime Results in Lost Revenue

After extracting crude from the earth, the first processing step involves separating the constituent parts - gas, oil, and water. The gas streams from the separation process go through a compression process after which the gas will either be sold or re-injected back into the reservoir. The compression process is critical to crude oil production. Without gas compression, the associated gas in the crude oil would be sent to the flare. However, environmental permits limits the amount gas that can be flared. If a compressor should go down for an extended period, crude oil production would no longer comply with the permit and have to be halted until the compressor could be repaired or replaced. In one example alone, compressor downtime caused the company an unplanned shutdown, resulting in revenue losses of about $4 million per day.

Slow Degradation

The associated gas from crude oil contains long-chain hydrocarbons that often tend to accumulate slowly inside the compressor. While operators routinely make minor tweaks in the controls to overcome the gradually increasing resistance to flow, the accumulated effect often goes unnoticed. Eventually, the power needed to maintain flow through the compressor increases beyond available capacity and its output starts to decline. With limited staff on the platform, there is no resident rotating equipment engineer to monitor the condition of the compression equipment.

While each oil platform is equipped with a spare compressor, it takes up to five days to swap-out and replace one. If this replacement causes an unplanned shutdown, the missed production can result in $20 million in lost revenue.

Assuring High Uptime

In an effort to avoid these kinds of losses, BHP Billiton and L&T Infotech developed a solution using an Industrial Internet of Things (IIoT) approach. Analytics are applied to data gathered from sensors mounted on the remote offshore equipment to generate appropriate alerts before something bad occurs. This often enables needed repairs/replacements to be performed during scheduled maintenance outages, so no production is lost.

Engineered Algorithm

Modeling the behavior of a compressor required research and engineering to understand the operating parameters and performance characteristics. An on-line and real-time aerodynamic performance modeling capability was developed. The two main components that determine compressor performance are head (foot-pound force per pound mass) and efficiency (aerodynamic). The calculations include some geometry (impeller diameters), and have non-dimensional parameters that make it easier to apply to variable speed machines. This research, documented in the paper “Centrifugal Compressor Performance Modeling and Monitoring,” by Todd A. Barham, BHP Billiton and published by Pipeline Simulation Interest Group in May 2015, provided the basis for determining an algorithm and calculations to predict compressor failures.

IIoT and Big Data

Fortunately, the calculations use existing sensor data, so the company did not have to make any mechanical or electrical modifications to the compressors. The sensor data are transmitted from the platforms to an onshore OSIsoft PI software application. This serves as a repository for the Big Data collected from the multiple compressors installed on a number of different offshore platforms. Data collected across a range of operating conditions and performance were used to create a performance baseline for each compressor. The program monitors each machine against its baseline.

Software Development

BHP Billiton engaged L&T Infotech, a global IT service and solution provider with extensive experience with the OSIsoft PI system, to develop and implement the software program. The predictive analytics algorithm was coded and implemented on the OSIsoft PI suite of applications. This involved several applications - some running in PI and others external. Using the algorithms and associated calculations, the program generates alerts based on thresholds set by the engineer. These alerts appear in a dashboard in OSIsoft PI ProcessBook.

Condition Monitoring and Predictive Maintenance

When the program identifies a problem, the rotating equipment engineers located onshore see alerts in the dashboard and assess the situation. Monitoring includes the mechanical response of the machine as well as the aerodynamic performance. Should the engineers see a change in aerodynamic performance, they engage the process engineers to help determine if a process requirement changed. While a chemical injection can clear some fouling on the internals of the compressor, not all fouling mechanisms can be managed by chemical injection. In these instances the machine may have to be taken down for cleaning.

The alerts include enough diagnostic information to make them credible to operations, maintenance, and management, so that appropriate action is taken. The algorithms provide enough advance warning of a pending issue so that the compressor can typically be replaced during a planned outage.

Avoiding Unplanned Downtime

According to BHP Billiton, the preventive maintenance program improves operations, revenue, and cash flow. In the past, when the compressor “hit a wall” at maximum horsepower, people would have to scramble to do the repair, often requiring an unplanned shutdown and the associated lost revenue. Now, engineers are able to monitor the compressor’s performance to anticipate when economics justify replacement with a refurbished compressor.

In addition to avoiding unplanned downtime, which in the past has cost as much as $20 million per incident, the solution helps save energy, since maintenance can be performed before power input has to be ramped up excessively to overcome routine turbine clogging.  With this kind of return, the project’s cost can be quickly recovered.

Conclusion

In the past, big databases, analytics, custom coding, and software maintenance were often difficult and expensive to implement or perform; resulting in resistance to proactive reliability programs. The rapid adoption of IIoT has created economies of scale for smart sensors, connectivity, and robust software platforms. This change warrants a re-examination of predictive maintenance programs. Industrial companies should examine unplanned downtime events and the related costs and evaluate the potential to eliminate unplanned downtime with the new technologies for IIoT, analytics, and preventive maintenance. It could also be beneficial to engage with a global service provider such as L&T Infotech that can offer deep domain expertise developing this type of solution.