Simularity Case Study - Oil Well Predictive Maintenance

Version 1

    .The Problem

    On December 9, 2015, at 4:56pm local time, a Progressive Cavity Pump (PCP) driven oil well in Oman known as “AV0902” suffered a break in the sucker rod string approximately 4700 feet below the surface.The unexpected shutdown required a completion workover service and replacement parts amounting to approximately $75,000. Industry average turn around for artificial lift repair is a week which also amounted to another 2,100 barrels of lost production. (At $40 per barrel = $84,000).

     

    Simularity Solution

    Earlier in the year, AV0902 had been outfitted with an innovative permanent down-hole gauge system which had 12 different sensor measurements including: intake and discharge temperature and pressures, downhole speed, rotor position, twist, downhole vibration, and more.

     

    The customer wanted to know:

    ● What caused the failure?

    ● Could it have been predicted and avoided?

    ● Were there other things happening deep in the well which still need to be investigated?

     

    Simularity provided the answer. Based in the Silicon Valley area of California, Simularity has developed innovative software that can analyze large volumes of time-series data in real time at the edges of the network. By capturing real time data from multiple sources, the artificial intelligence software can ‘learn’ what’s normal and predict incidents before they happen, including “time to failure” estimates and explanations for its conclusions. For the AV0902 incident, Simularity visualized the data for each sensor (see in attached White Paper) and learned the complex

    correlations between the sensor variables.

     

    After a normal period, two significant alerts appear 45 and 33 days in advance of the failure. To know more, read attached White Paper and call us at 408.505.6456