Petronas cuts costs with AI-infused AVEVA solution in the cloud

Image by Job Savelsberg

Malaysia’s Petronas is expanding the deployment of AVEVA Predictive Analytics at an additional 10 plants with a total of 150 equipment trains, following a successful first run.

The pilot project in their corporate cloud running on Microsoft Azure involved four upstream platforms and at two downstream plants. 

Producing 2.4 million barrels of oil equivalent per day, the state-run firm recognized the importance of plant stability to achieve its sustainability goals. Their engineering division was keen to optimise the performance of their equipment to improve plant reliability and reduce downtime, and turned to AVEVA’s asset performance management (APM) solutions.

Petronas intends to continue rolling-out the APM solution in the cloud to all its assets to enjoy similar results across the business. The energy firm also uses cloud-based AVEVA Unified Supply Chain to optimise its entire supply and distribution network, cutting crude evaluation time and reducing margins.

Trisystem Engineering (TSE), a systems integrator, was hired to deliver the project. In each site, the solution was up and running within two months as AVEVA Predictive Analytics comes with out-of-the-box purpose-built AI that has been customised for each industry, meaning that no coding is required.

According to AVEVA, the pilot implementation accurately predicted failures in advance that enabled Petronas’ team to fix issues ahead of actual failures. 

In 2020, the first year of deployment with 200 models deployed, the solution accurately identified 51 major early warnings, creating value of 73.1 million ringgit, equivalent to savings of US$17.4 million, and 14 times ROI.

Out of the 51 warnings, 12 were identified as high-impact warnings. Resolving these ahead of actual failure has reduced unscheduled downtime.

Many of the catches helped to reduce critical rotating equipment failure and downtime and have led to improved reliability through proactive asset monitoring and maintenance. For example, an instrumentation fault was identified leading to a catch in a liquid separator that saved Petronas $222,000 in potential asset failure and wasted material. 

A potential motor failure was also caught when AVEVA Predictive Analytics identified increases in the motors lube oil temperature, the winding temperature and the hot air temperature, saving $82,000 in equipment replacement. 

In another situation, a mechanical fault was identified allowing maintenance engineers to pinpoint a water supply temperature out of specification along with an increase in bearing temperature. Catching this issue before it cascaded into a major equipment failure saved Petronas $48,000.