From traditional Oil and Gas to Sustainable Energy OR From a traditional RDBMS to MongoDB

With the transition to sustainable energy, the data tier supporting a surface-network becomes more and more important. We will show how MongoDB plays a role in that transition, by ignoring recommended patterns, using some radical approaches instead for a high-performing Sustainable Energy Solution.

In long running projects and with products that have a long deployed life-time, you are often confronted with technology that is catching up on you. You start out with the best technology available at that moment, come up with nifty solutions to circumvent the technological boundaries of that technology and in some cases accept that you cannot do any better. This proposal will discuss how a 15 year old project has transformed from the then best RDBMS solution to a MongoDB NoSQL Solution:

In the Oil and Gas and Energy industry in general, uncertainty plays a major role: Engineers make assumptions on what the subsurface looks like, how physical properties are distributed over a reservoir, etc. In a typical exploration workflow, engineers therefore work with Experimental Designs that create many realizations that are a slight to a major variation of base assumptions that have been made.

One of the unique features in this project is the ability to design and optimize on a digital twin of the production components connected to the oil and gas producing reservoirs. With the transition to new Clean, Sustainable and Renewable energy this proves to be a game-changer: while the source may change, the optization challenges on the energy distributing network remain largely the same.

This differentiating feature also proved to be one of the biggest challenges from a data-management perspective: types of equipment, different vendors, different performance specs and different properties that need to be recorded and calculated on. The ability for engineers to ‘try’ different setups. On op of that an ED system that needs to create many realizations of different configurations and throw away results that do not meet performance requirements.

The best solution the team could come up with was an EAV schema, but with the production of more than a billion of measurements per day per asset, this has become unmanageable.

In this talk we will demonstrate how we could bring MongoDB into the equation to solve this EAV problem. We will cover time-series, object models, recommended practices from MongoDB that work, but best of all: some unconventional approaches to tackle situations where those recommendations do not work. We will also show how, by following a hybrid approach, this could be achieved without any down-time and a TCO improvement of over 45% for the data-tier alone.