Analytics Black Box

Reluctance to adopt such techniques due to the opaque nature of such models. How does one explain exactly why a neural network produces a particular result? The industry is observing other industries being ‘transformed’ by advanced analytics, but generally and due its enginnering DNA does not trust these outsiders to build bespoke solutions in Oil & Gas. But we see applications of Neural Networks, Support Vector Machines, Genetic Algorithms being published frequently from the major IOCs, NOCs and Independents regularly, and there has been a noticeable uptick in the hiring of data scientists across the board. This tells us that the industry believes there is value to be generated, but it must be an extension from existing physical and engineering models, not blind statistical models. We refer to this group has ‘hybrid models’

Presenter: Petrospan example of domain experts incorporating advanced analytics (neural networks) with physics based reservoir simulations to bridge commercial and technical analysis. There have been numerous attempts to do similar work from groups such as Kaggle and OAG analytics, however these techniques were generally “bottom up” approaches to take any and all data, mix them together and identify predictive parameters that defied subject matter experts understanding of physics. Petrospan, takes a top down approach, introducing dimensionless Reservoir Productivity Factors (RPFs) and Completion Efficiency Factors (CEFs) to deconvolute the impact of reservoir rock and completion parameters to evaluate well performance and reserves using only publicly available data.

Presenter: CEP: Is an example of domain experts that through years of hypothesis testing on the field have come across multiple opportunities including Auto-driller dysfunction. This commonly used feather on drilling rigs has proven to results in oscillation, equipment damages and NPT. The use of complex event processing is a well-known technique that combined with domain expertise has resulted in a nimble way to convert the power of experts in ROTC directly to the drilling console.

Partnership

Advanced Analytics companies in O&G are forced to create niche solutions that are dependent on a variety of other infrastructure, services and integration issues. These small companies have a hard time creating sustainable revenues if they can not promise an end to end solution to the customers. Analytical techniques themselves are fairly commoditized, as many of the most prolific algorithms used across different industries can be accessed straight from GitHub. The problem is that no single company has the domain knowledge, data management, physical infrastructure/hardware, and analytical expertise to deliver a homegrown end-to-end solution at this point. Over the last few years a very complicated web of partnerships has emerged between the stakeholders entering this space. We do not claim to know the exact recipe for success, but we believe any analytics company must have a partnership approach that gives it the flexibility to deliver products to a highly variable customer set (i.e. sophisticated IOCs with major greenfields to smaller independents with less resources).

Preseter: Flutura Is an example of a company that has built an end to end analytics and remote monitoring platform through strategic alliances with vendors upstream in the technology stack (i.e. network and physical infrastructure, and instrumentation vendors). The company uses its domain expertise in advanced analytics and decision sciences and outsources the rest which enables it to be more nimble and focused on its core competency.

Product vs. Service

Big data and advanced analytics companies still struggle to productize their technology.The model that has worked in the O&G space thus far, has been a service model with the development of custom products that slowly evolve into product centric organizations. There is no ‘one size fits all’ analytical technique, so you see two options:

• Consultancies who can build bespoke solutions after extensive scoping engagements
• IT firms who provide comprehensive, state of the art, tool boxes, for customers to build their own models.

When facts change, these models must be rebuilt or modified, and the service cycle continues and the tab gets expensive. Specific products are often times too niche to build an entire company around them. Companies like Schlumberger and GE have attempted to solve this with application marketplace, but customers view them as being too biased and are far too familiar with vendor lock-in issues.

Presneter: Ruths AI Is an example of a data science consultancy that created a sustainable service company to enable the build out of bespoke solutions for the O&G industry that would otherwise not be large enough to create a going concern on a standalone basis. Akin to the evolution of the automobile industry, they have taken a modular design approach to enhance the scale and interoperability of their applications. Large incumbents like GE and SLB have attempted similar services with limited success, largely in part because of their bias as OEMs and Service Companies.

Prestner: OVS OVS has productized data integration with a library of fit for purpose workflows tailored to the O&G industry. This work has historically been dominated by boutique data management consultancies (i.e. Noah Consulting). Similar to Ruths AI for analytics, we like OVS because they have productized end-to-end specialized workflows, in lieu of building a continuous service model with a high price tag.

Talent

There is a lack of Data Science and AI talent within the Oil & Gas industry, we can safely assume that the cultural differences between the two communities, and the incentive structure for these individuals to become rank and file members of large Oil & Gas organizations has played a major role here. The analytical talent that has been in the industry has typically been stationed in R&D shops within the service companies, working on ‘moonshots’ that take decades to monetize. These individuals have been groomed for the last decade plus, capturing the requisite domain knowledge. Portions of this talent pool are beginning to break out on their own, likely accelerated by the downturn in the market today. It is our opinion that there is the consumer space is going through crowding out which is why there is an uptick in analytical solutions coming to O&G from outside.

Presetner: I2k Is an example of scientific experts in fields outside of O&G who have spent decades of their lifes work IN O&G. We see this group as an example of a wave of talent that is becoming available in the current price downturn. The company’s founder is acknowledged widely in the field of Artificial Intelligence. To the typical engineer, their solutions seem very “IT” heavy, but this is the reality we are in today if we want the conveniences of the consumer technologies we have come to rely on in our personal lives to become equally ubiquitous in the workplace.