Follow these best practices to navigate common challenges.
by Michael Kanellos
August 21, 2019

Oil and gas is one of the driving forces in industrial internet of things (IIoT). But it is also challenging to separate the reality from the hype. Below are five questions and a set of best practices to help navigate the challenges.

It is often necessary to install artificial lift systems in early production depending on the producing formation and expected reduction of the natural reservoir pressure. In unconventional production, including shale and tight formations, artificial lift is needed right at production start to compensate for rapid decrease of the flowing pressure. The well bore architecture is also complicated with deeper wells, deviated and long horizontal legs and multi-well entries.

1. How is IIoT different than IoT?

To people in the information technology (IT) industry or in IT departments, “IoT” and “IIoT” often mean collecting and analyzing machine data to achieve a business goal such as reduced energy consumption. To people in operations, that is business as usual. To them, IoT means putting in a secondary network on top of the so-called IoT network that is already reducing energy.

Yes, this sounds like a word game, but it is an important distinction because it will help avoid confusion and redundant or overlapping efforts. Users need to bring IT and operational technology (OT) onto the same page. The first step is a common vocabulary.

2. Where are potential data silos?

Simply put, it costs less and takes less time to link new or stranded assets through wireless technologies and IoT gateways than to retrofit existing supervisory control and data acquisition (SCADA) or distributed control systems (DCS). Connecting untethered devices with IoT gateways and stick-on sensors can cost a few thousand dollars. Upgrading a SCADA system to do the same might cost $250,000 or more.

These stick-on sensors will feed data into something other than SCADA, and that can result in data silos, which then means headaches and an incomplete picture of operations. Whatever the architecture in use, merge data sources.

3. What is the business case?

IoT typically can deliver three macro benefits: cutting costs, increasing revenue or reducing regulatory risk. Or, as analyst James Brehm likes to call it, save money, make money and stay out of jail. Those are all worthy goals, and the best bet is to knock them out in order.

Start with predictive maintenance and reducing uptime. Then look at ways IoT can increase the productivity of existing capital. Natural gas processing company DCP Midstream has created an interesting program that gives plant operators visibility into current production, potential production and market pricing. The idea is to show employees how they can squeeze more out of the plant. Once those are established, many move onto things like emissions monitoring. If users try to implement them all at once, they are just increasing the chance of failure.

4. Who is accessing the data?

With IIoT, not only are users collecting more data, they are also creating a system that will encourage more people to use the data. That requires thinking about the user experience. Having mobile access to contextualized operational data coming from all data sources is an absolute necessity. Operational data will be accessed by operators and maintenance technicians in the field responding to emergencies as well as executives at the airport who want to compare current production with pricing forecasts.

If people cannot start quickly using this data in their daily routines, the project will likely fail. Providing easy-to-use, contextualized data in a mobile, self-serve environment is fundamental.

5. Who is performing analytics?

When someone says “analytics,” people often envision data scientists leveraging algorithms to sift through mountains of data in the cloud. And users will see gigantic problems solved with the IoT data in the cloud.

For example, MOL, a large refinery in Hungary, is conducting analytics on its machine data to determine how and when it can use less expensive, higher sulfur “opportunity” crudes in one of its refineries without introducing risk to its operating parameters.

Still, many analytics problems are performed directly by people looking at a few finite data streams, which could be viewed as human analytics. For example, a sudden drop in oil pressure on a truck working in the tar sands of Canada can be a leading indicator that a major malfunction may be in the offing.

Companies should think about taking a “layered” approach to analytics that will take into consideration edge versus cloud capabilities and costs, whether human or algorithms might work better and the feedback loop between these systems to ensure insights are operationalized.

Users cannot anticipate every use case they will have in the future, but if they take a broader view of what “analytics” means, they can create an analytical framework that can take care of the unknowns.