We are in a world of exponentially growing data sets. ‘Big data’ is a key buzzword and a real trend reflecting the rapid growth of digitalization within the oil
and gas industry. Big data by itself is just numbers. Predictive analytics is what makes it so powerful i.e. the ability to model the world, predict events and
make data-driven decisions with accuracy. The increasingly widespread interest in using ‘big data’ sets to make informed decisions via predictive analytics
leaves us with two important open questions. To what extent does big data help in answering questions to problems that actually matter? Does bigger data
mean better predictive performance?
In this talk, using a case study, we will numerically demonstrate that ‘little data’ goes a long way. Using predictive analytics we produce a truly optimized
inspection program for a hydrocarbon piping system susceptible to internal corrosion. The results are based on inspection data from one of the biggest oil
& gas companies in the world. We show that most of the inspection points can be inspected at longer intervals, reducing inspection scopes by ~ 40%. This
implies that using less data we can achieve better focused results.