IIoT based Predictive Maintenance 4.0 (Part 3)
How did we develop a comprehensive platform for PdM?
The amount of data collected by the digital assets farm is humongous. To acquire, store, analyze and visualize such data using traditional software may not be efficient and scalable. Accurate prediction of failure events well advanced in time requires more powerful and robust ML algorithms. There are various stages in building an end-to-end solution for PdM. This system is an end-to-end PdM platform that analyzes IIoT sensors using advanced AI/ML algorithms and big data technologies to improve asset availability, extend assets’ lifetime and reduce cost.
System Overview
IIoT sensor data is gathered into the scalable and efficient centralized data lake. Further from the data lake analysis pipeline carries out various data manipulation stages, namely data integration, data clean-up, data processing, data annotation etcetera. Then analytical experiments are performed on over-processed data. Hence, experimental reports and failure alerts are visualized in a user-friendly, interactive dashboard.
What are the financial benefits of IIoT PdM?
The goal of IIoT PdM is to provide alerts of asset degradation and evolving failure early enough to prevent unscheduled downtime. If operators are alerted to prevent failures, they can reduce the burden of inventories of equipment. Reactive Maintenance is not only expensive but also time-consuming. An overall goal could be lowering operations and maintenance (O&M) costs and higher yield rates.
A specialized oil and gas pump may fail 4-5 times in its typical life cycle of 3 years. The downtime for each such instance is 6-7 days which means, a single pump is out of production for approximately thirty days in its life span. Thus, it costs OGIs huge production outage, labor costs, repair or replacement pump cost and other overhead costs.
Let’s look at a pump example, these pumps are specialized and if any damage occurs consequently the pump may have to be replaced. The replacement cost of one pump is up to 8K $ plus gas production loss 6K $ (200$ per day * 30 days), hence gross O&M cost is at least 17K $.
Assuming a 30% reduction in the O&M costs, this would translate into an annual cost saving of 3700$ for a pump.