Last night, distinguished Society of Petroleum Engineers (SPE) lecturer, Muhammad Khakwani from Saudi Aramco, gave a presentation titled “Essential Pre-Requisites for Maximizing success from Big Data”.
Muhammad’s comprehensive and inspiring presentation included four cases where Saudi Aramco used Big Data to save money within their operations. Embarking on these projects is not a straightforward path. Several subjects and considerations are required, for example:
- Obtaining “buy-in” from teams and department managers,
- Understanding what data you have now and will have in the future,
- What technologies you are going to use to cope with that data..etc.
In my opinion, to describe Big Data, one needs to imagine high-frequency stock trading. Vast volumes are traded with speed across networks across several trading hubs, across different time zones. Patterns are analysed using complex mathematical algorithms. This leads to insights and strategic decisions/moves that can be implemented for the benefit of the trader/trading organisation.
This type of Big Data success is what vertically integrated companies like Saudi Aramco are trying to replicate for their business operations. This can lead to significant cost savings across the supply chain.
I’ve gathered some points-of-view based on Muhammad’s presentation. This is not a step-by-step recipe but an interchangeable checklist:
- Define a project with business goals in focus
- Start with a pilot project
- Implement Design Thinking to find the right project
- Define realistic goals early across the disciplines involved
- Manage expectations (what Big Data projects could deliver vs. what is desired from Big Data projects)
- Think about and agree across the project team how you measure success
- Educate/empower your internal teams with shallow knowledge of Big Data, IT Infrastructure and Data Science techniques
- Have a Subject Matter Expert (SME) co-steer the project with the Big Data Team Lead
- Don’t be afraid to fail
Working With Big Data Projects
Defining a project that focuses on business goals can sound obvious. But progress and/or intended results could easily drift out of focus, the deeper one delves into the subject. Additionally, you may need to anticipate that you or your team may get it wrong, just like other big organisations have done so. Learning from that failure can significantly impact how ongoing or future projects can be successful. In fact, failing in a pilot project can be a good thing.
Big Data can be a powerful tool to help make business decisions and save your department millions of dollars per operation or per year. Specialised techniques could be used by an Oil Major to review large volumes of compliance documents rapidly (compared to humans). In order for the company to operate in a specific country, certain regulatory requirements would need to have been met. Before submitting their paperwork, an anomaly in the compliance application could be detected using Big Data Analytics which would end up saving the company millions of dollars in non-compliance fines.
Trust issues are a part of life when dealing with results/data tables/plots coming from simulations/new techniques etc. Predictions and patterns can be trained based on historical data and work on paper 99% of the time with some tuning of the algorithms. Yet, operational experience may contradict what Big Data results show. This is where the expertise of an SME is helpful. A mutual understanding must flourish between the SME and the Big Data Engineer/Data Analyst/Data Scientist/Machine Learning Engineer to make sensible decisions based on a blend of practical experience and data results.
Startups can offer a great deal of agility and knowhow when looking at projects of this type. As Muhammad suggested in his talk, one should “try to do the project internally at first”, if little or no knowledge exists internally. It might take time, it might go nowhere fast but at least you will have a grip on what the external parties are talking about. I think that if this approach is taken, extra emphasis should be placed upon points 1-5 above.
Lastly, topical issues such as data inconsistencies, data security assurance and governance to handle data projects are still a barrier to fully realising cost benefits associated with Big Data. These barriers could hopefully be lowered in the coming years due to more and more projects being launched, lessons learned and building trust and controlling expectations between assigned teams and sub-contractors.
Thanks to the SPE France section for organising this event and TechnipFMC for hosting. Follow SPE France for news updates and notices of upcoming events.
This article was originally posted here.