Matt Quinn, TIBCO Software, USA, explores the growing and increasingly important role of data scientists in the oil and gas industry.
A successfully working oil or gas field is the product of many millions of components and systems, all of which have to work - and work correctly - in order for production to be achieved at all, let alone maintained at the expected levels. There are so many things that can go wrong, and of course they often do. And to date, the most common way of establishing that a well has a problem is by finding that the problem has already occurred.
If it was possible to fix the problem before it happened, not only could companies prevent large production losses and save money, but could also prevent having to contain or clear up spillages or gas escapes.
Now imagine that it was also possible to identify any components and systems that might be close to needing maintenance. Could those jobs also be carried out while the most significant problem is being rectified? At a time when many budgets are fixed and there is little in the way of contingency funding, conducting all other imminent or necessary maintenance at the same time as a major problem is resolved is a sensible and cost-effective use of resources.
Data scientists are a new breed within business and industry. It is their role to find, identify and understand such important factors.
Their job is at a new front line in business - making sense of the vast and ever-growing mountains of raw data that production systems produce so that vital information can be identified and interpreted in ways that save businesses money, and/or open up new business opportunities.
The key skill of the data scientist is knowing what questions to ask, extracting the right information from the mountain of data that will frame those questions, and knowing how to interpret the answers so that meaningful results can be found quickly and easily.
There are two broad areas where data scientists can make a significant contribution to the efficient and effective running of oil and gas facilities. One is in the day-to-day monitoring and management of every functional aspect of a well and the systems in it. Get this bit right and not only will the production levels stay high, but the costs caused by problems leaping out unexpectedly will be minimised, and often removed entirely.
The second is related, and uses much of the same information but for more long-term prediction purposes. This is risk analysis, where the goal is to spot major problems long before they become live issues. Ideally, as the pool of data grows and the experience of what the data means broadens, it can become possible for data scientists to design out problems in the first place.
Surveillance and optimisation
The day-to-day monitoring of all the systems operating at a well is hardly new. Indeed, many of the systems already produce vast amounts of data that are currently going to waste. Yet that data can be valuable. It can be used to provide rich surveillance capabilities over the production status of not only individual wells, but also the entire oil or gas field. It can also be used to optimise the production coming from both individual wells and the field, as well as optimising the maintenance of assets.
The ability to provide surveillance of the production levels across a number of wells, when mapped against data on the performance and operations of individual wells, can give a wider understanding of which ones are not producing according to plan, why they are not producing, and what needs to be done to rectify the situation.
This allows companies to optimise production, especially across an oilfield, which maximises efficiency when operating with fixed budgets. If there are a number of wells that need maintaining, there is a need to decide what is the optimal set of assets to have in production at any one time. Production across a whole field can then be optimised on a well-by-well basis.
Optimising maintenance is, of course, a crucial capability. There is an obvious difference between scheduling maintenance of equipment as per the manufacturers’ schedules, and scheduling it when the monitoring on it shows it actually requires urgent attention.
There is a need to figure out when to do the maintenance to give the lowest cost and the least interruption to production. In practice, undertaking maintenance to a manufacturer’s recommended schedule can often lead to greater cost because the work done is unnecessary. By the same token, there is a need to avoid creating too much non-productive time, where there is an unplanned downtime event and the well has to stop producing.
The skill of the data scientist is in getting the balance between those two right.
There have been several catastrophic incidents - such as the Piper Alpha in the North Sea and Deepwater Horizon in the Gulf of Mexico - that have been put down to a variety of potentially predictable issues where the skills of the data scientist could have been utilised to build models that would have alerted the right people at the right time to highlight the issues - and their potential severity - in advance. Predicting where poor design, or manufacturing, or if maintenance practices indeed need to be changed, based on past history and early testing results can save businesses significant sums and significant brand damage.
In practice, of course, while such big events make the headlines, there are many more, smaller, events that occur on an almost daily basis. These days, these can be - and need to be - analysed on a systematic basis, particularly when they are often occurring in tricky, offshore locations. There is a need to understand which wells may have underlying issues that are not always apparent.
This is particularly the case with new exploration and production techniques, such as hydraulic fracturing. This is an expensive, and today still contentious, approach that requires careful monitoring and analysis of the processes involved. For example, it is important to use the appropriate amount of fluid. This is just one of a number of different factors that have to work together to achieve a cost-effective result.
There are also, of course, a wide range of health and safety regulations surrounding the development of hydraulic fracturing. For example, there is its potential for polluting local water courses. Data scientists can play an important role in ensuring all these regulations are met and predicting when a breach is imminent.
What goes to make a data scientist will vary according to type of industry they are in. For example, in a more direct marketing-driven business, such as retail, they will need good grounding in the realms of numbers and psychology - if these three products are put close together, what should be the percentage increase in their sales volumes?
When it comes to working in the oil and gas industries, however, they will still require a solid foundation in science and technologies. They will, for example, need to be able to spot that even a slight change of trend in the numbers from certain monitors means that a specific system is in need of maintenance or is starting to cause a well to under-produce.
Adapted by David Bizley
Read the article online at: https://www.oilfieldtechnology.com/exploration/12112013/a_new_breed/