” ? can we make these TRENDS SPEAK rather than SLEEP ? “
Process Data Overload – Measurements and Records are inherent to automated process plants. The automation DCS not only displays “NOW” status of plant machinery on CCR terminals but also continuously logs DATA into plant historian. In a typical medium scale process plant DCS on an average backs up 50 mB of data every day. Various data types, This data back up is a dead inventory unless harnessed to create value.
¤ Possible KNOWLEDGE in process DATA you always WANTED to EXTRACT..
* Why was my production down yesterday. What went abnormal?
* Which set of operating parameters give best output?
* What is the effect of variation in input material on energy consumption?
* How well are my basic controllers working?
* Which are the strongly influencing variables?
* Is there a scope for improving productivity/energy efficiency?
* Which CCR operator has best productivity?
* When is the right time to take the equipment for maintenance?
* Can the abnormalities be predicted before they lead to major damage?
¤ BIG Data Analytics – is there something for my plant? .
* How to decipher patterns from huge time-series data to establish important events, causalities, process abnormalities?
* How detect, diagnose and correlate processes fluctuations occurring over different time scales and at different locations?
* How to build predictive models of multi-variate effects of process variables on important KPIs?
* Is it possible to predict non-measurable or difficult/costly to measure quality/safety/environment parameters from available data?