A global leader in Solar Thermal Power generation leverages BLUE BOX™ to enable risk-based maintenance of boiler feedwater and heat transfer fluid pumps. Day and night cycling requires a high degree of operational flexibility. Continuously changing operating conditions increase the operational risks to these critical pumps for the plant.

Sulzer’s pump-specific, Artificial Intelligence (AI)-based analytics platform leverages proprietary machine learning models. It combines pump OEM physical equations and state-of-the-art mathematical AI methods. This combination increases the accuracy of identifying abnormal operational behavior in pumps and highlights pre-failure conditions.

Shortly after commissioning and training of BLUE BOX in a Solar Power plant in South Africa, various anomalies were detected in a specific pump train for a period of two days. A physics-based anomaly detection condition was triggered when the torque of the pump was significantly higher than expected, compared to healthy conditions. The customer-dedicated Sulzer expert reached out to the customer and the issue could be isolated, while at the same time, further anomalies could be detected. A bearing showed significantly higher temperatures compared to normal values, however, still below the customer’s set threshold value. It could be demonstrated how modern AI machine learning gains precious time to assess and take timely action to prevent unplanned machine downtime.

Meanwhile, BLUE BOX analyzes more than 40 pump trains for this customer at four power plants with no unplanned outage of pumps since deployment of the solution. Sulzer is contributing to a reliable and sustainable electricity supply with a total installed capacity of about 1’000 MW, powering more than 300’000 homes.

Identifying pumps operating at pre-failure conditions

• Improve uptime and reliability of supply
• Increase operational flexibility
• Reduce life-cycle costs

Reliable sustainable electricity supply for more than 300,000 homes.