The Challenge

VROC was contacted by a Tier 1 FPSO asset to assist with tracking asset degradation on Water Injection Pumps. After being impressed by the VROC platform, we were quickly asked to commence AI modelling on the Produced Water Pumps to monitor the system, predict trips and track performance degradation. 

Early Warning

VROC commenced detailed AI modelling on three Produced Water Pumps.  The AI models look at live and historical data analysing what normal and abnormal operation looks like, tracking the slightest deviation.  We found some significant anomalies unfolding in the live data for Pump A, as can be seen in the image on the right. Healthy operation is normally green, Pump A can be seen changing from yellow to red. 

The Catch

A Produced Water Pump failure was detected by the DNA map as shown on the coloured bar chart on the left. Investigating the data further the models of the seal pressure and temperature clearly show the reduction of seal pressure and temperature from normal, revealing a mechanical seal leakage.  The Produced Water Pump failure was subsequently confirmed by the client.

The Results

VROC successfully identified in advance the first failure to occur on this Produced Water Pump.  The Produced Water Pump was subsequently taken offline for maintenance on the leaking Mechanical Seal.  VROC was quickly engaged to commence modelling on the remaining functioning Produce Water Pumps to ensure that any potential failures would be predicted in advance, preventing failures which would disrupt production.

Our client was extremely happy with the results of the modelling, which was able to detect the degradation more than 5 days before a failure. This has given them great confidence in the VROC platform to help predict and avoid any future performance issues.


VROC's Lead Data Scientist explains this case study in further detail - watch below