Intelligent edge computing at the wellhead improves performance
The digital wellhead promises efficiency and productivity increases for hundreds of thousands of wells across the U.S.
Take any flight from Texas to California and you will fly over West Texas and New Mexico. A glance out the window will reveal a land pockmarked with a seemingly boundless patchwork of square sandy areas. What the eyes won’t discern from that height is that each one of those squares contains an oil wellhead. Welcome to the Permian Basin that supplies almost a third of the U.S. domestic crude oil.
The oil fields in the basin contain tens of thousands of wells, and the number is growing every year. Over the past five years, more than 5,000 wells have been added to the inventory. Aside from the wells themselves there is little else in the region as far as infrastructure goes. Each of these wells is unmanned, and often, located in areas difficult to access. Despite these challenges, the wells need to be inspected for gas leaks and any structural damage, which can be a costly and time-consuming operation.
Aside from the need to avoid falling foul of compliance or health and safety regulations, there is the matter of data. In modern oil production, data is vital for planning and assessment, and wellheads are a significant and abundant source of data that at present, is mostly untapped. Although some of the newer wellheads may have smart digital sensors with built-in wireless communications, the vast majority are legacy installations with analog gauges. Even the new smart sensors require a technician to be near the site to download data to a handheld device.
What is a digital wellhead?
Operators are trying to digitalize these fields and wells so that they can detect all the important production parameters from a remote location. The answer is a digital wellhead that provides integrated functionality at the edge. This will allow real-time and predictive analytics of wellhead integrity, well performance and environmental risk.
The idea of the digital wellhead is to give it a brain, or more accurately, an edge computing device. The plan is to bring everything onto a level playing field. Whether the well instrumentation is analog or digital, we would like all the wellhead information to be available on the same platform even though it comes from different devices and vendors.
That is the ultimate vision, but the most important thing is to achieve a digital wellhead regardless of whether it has remote connectivity, because there must be multipurpose computer intelligence on site. This will have a positive impact on three areas. The first is automation, the second is precision, and the third is prediction. These are the three big advantages that were not able to be achieved before we added edge intelligence.
Delivering wellhead automation
A typical example of efficacious use of automation is gas leak detection. This is a significant concern at wellheads, and even for abandoned wells where the operator is still liable for any leak from the structures. The challenge around gas detection without automation is that it always requires people to visit the site. These inspection teams will attend to the wellhead site with infrared sensors or cameras, move around the area and try and detect any possible gas leaks. These wellheads are remote, so there is travel involved, the inspections are infrequent. When they do visit the wellheads there is the added risk that they may be exposed to health and safety risks.
The idea behind having automation and intelligence at the wellhead is to be able to put gas sensors at the wellhead that can constantly monitor for any gas in the area. With an automated drive and the sensor located on a rotating mount, it can turn in various directions controlled by the edge computer to remotely monitor conditions. This reduces the need for people to visit the site.”
There are a range of philosophies when it comes to digitalizing the wellhead, either through a device strategy or an edge strategy. One option would be to replace the dumb, analog sensor with a smart sensor that has embedded computing and communications capabilities. This can be expensive and often only delivers a point–to–point solution with the individual device providing its information in isolation. Aside from that lack of integration, this method can also be challenging to scale.
Edge-based digitalization is about being open to the low-level sensors. The sensor can be dumb, but it is given intelligence by the edge computing device. You are adding intelligence to sensors inside one computing device that can include gas detection, pressure or flow monitoring and even structural monitoring.
Precision and prediction
Precision at the digital wellhead is another benefit of edge computing. There are two ways you can detect any issues such as leaks or corrosion. One is to use a threshold-based anomaly such as pressure changes. With edge computing, detection can be more precise. On the market now are relatively low-cost gas detection sensors that can be attached to seams or seals between the flanges. These can detect even small leaks; however, it requires data processing at a high level to filter out the noise of the baseline fluctuation. The software needs to be able to look at trends and continuously compare the anomalies before it can confirm a leak. It is not a single point detection, and when you have edge computing resources there, that makes it much more feasible to do at the wellhead.
Another area is prediction, which encompasses two separate things. The first is incidents, whether there are a leak or pressure issues in the well. The second is risk factors. There being a risk factor does not mean there will be an incident. For example, there may be some structural changes at the flanges that are not bad enough to create an incident yet, or there may be heavy corrosion around the wellhead structure that could lead to a future leak. The prediction is derived by using the risk factor monitored to be able to derive the probability of a wellhead incident so the maintenance and inspection visits can be targeted, and condition based.”
So why would prediction require edge processing? One of the newer ways of monitoring corrosion or erosion of the wellhead is by using advanced, AI-based image analytics. It will continuously monitor the patches of color changing on the pipes in a well structure to be able to detect the color and pattern changing as the corrosion and erosion have been advancing. Then it will need to be able to integrate the risk factors of humidity and temperature in that well.
All those risk factors need to be monitored, and, combined with the prediction, the algorithm determines which well could be more exposed to incidents or quality issues. All that sort of prediction requires edge computing and is not something that can be done by a single sensor or device. Risk assessment with predictive analytics creates digital wellheads.
There are two types of resources that cost money at the wellhead. The first is people, including the wellhead maintenance crews that visit the well sites. Then there is the cost of any regulatory infractions. By using intelligence at the wellhead these visits can be dramatically reduced, not down to zero, but they would become condition–based as opposed to schedule–based. Every well is different and can be exposed to various risk factors, so optimizing the schedule for individual well visits reduces the cost of inspections.
There is also an additional, secondary effect of the digital wellhead, and that is the increased data capturing and integrated processing. “The reality is that there is more data generated at a wellhead than there is data processed,” Ren explains. Wellhead data is very valuable in monitoring and diagnosing the health of the well, the productivity of the well and quality of the well. You have to aggregate and analyze multiple wellhead data sources together in context, to be able to make better conclusions about the reservoir.”
Most of the digitalization efforts at wellheads are currently centered on creating digital data without being able to integrate this information to undertake analytics. By putting computing power at the wellhead, you are not only putting it at the edge but aggregating multiple edge devices into a center that can create more valuable intelligence on the reservoir and its long-term health and productivity.
Jane Ren is the CEO and co-founder of Atomiton. Atomiton, founded in 2013 and headquartered in Santa Clara, CA., is an enterprise IoT software company.