Predictive Analytics for Oil and Gas
Predicting Problems Before They Happen
When I first heard of predictive analytics, machine learning and cognitive security, I was skeptical.
I am an engineer at heart. That means I thought condition-based maintenance was the only way to effectively look at predictive maintenance. You start with the physical asset, you deploy sensors to monitor critical components, and you analyze the data.
The thing is, this approach is expensive and time-consuming. Sensors need to be deployed and installed on existing equipment. Software to collect, store and process the data needs to be integrated. O&M teams need to be trained on the technology, and the software needs to be constantly updated.
The complex algorithms and predictive analytics of machine learning can make decisions almost automatically. Meanwhile, data analytics provide a systematic way to make sense of the large volumes of data being collected across the entire oil and gas value chain.
This can help predict when an asset will fail, or ensure that resources and personnel are in place where needed.
Whereas predictive analytics will ensure O&M efficiency and optimization, cognitive security is designed to monitor and protect the IT network. This makes a cognitive suite like that of SparkCognition a true cyber-physical solution, protecting both the IT infrastructure as well as the OT network.
With the unparalleled optimization and increased safety that predictive analytics and machine learning provide, asset monitoring and analytics have never been more important for the industry.
When a top drive starts, stops, or fails, there are repercussions on the generators in the engine room. When the shale shaker’s state is changing, the mud pumps are affected.
Often, the entire data set on the rig is full of cross talk between the different assets. Oil and gas companies are increasingly having problems learning from data to understand the different operational states and failure modes of critical assets.
The sophistication of cyber attacks in the IoT environment is making it more and more difficult for traditional security systems and teams to differentiate between natural machine failure and equipment sabotage.
It is clear that the industry status quo of condition-based monitoring and traditional security solutions is outdated, and is quickly becoming more expensive and less effective.
A Sense of Reality
Let’s take the case of a drilling rig and assume there are performance issues in the drilling process. The O&G sector has become susceptible to cyber attacks and the assumption that performance is a mechanical issue is no longer a given.
We must now ask whether we have a problem because the rig has been compromised by a cyber threat — is the top drive under the control of hackers? Do we have a maintenance issue and the mud pump is starting to fail? Or do we have a true drilling dysfunction that needs to be mitigated with a change of the drilling set points?
With predictive analytics and cognitive security, you can protect the rig from cyber attacks and be sure the equipment is operating safely and securely, per specifications.
By leveraging patented cognitive technology and utilizing data already available, SparkCognition, a leading provider of AI-powered cognitive analytics, is able to predict failure and provide advanced warning before a critical asset incident occurs, so operators can plan for corrective actions.
By providing answers to a large percentage of the questions surrounding the remaining life of each particular asset or component, cognitive technologies are equipping O&M and security teams with the means to optimize human resources, equipment inventories, and budgets.
A Wealth of Data
Today, there are vast amounts of data which are unused or incompletely used. With patented technology, we remove the clutter that exists in this data and extract the hidden value. This can only be achieved with cognitive analytics and security.
Using machine learning, we are deploying implementations where our system is as good as a handful of outstanding experts in the field at predicting upcoming failure. From there on, the more you use the system, the better it gets.
The future is augmented intelligence, and the future has arrived.
The London Stock Exchange is using our platform to detect fraudulent activity in finance. Flowserve, the largest manufacturer of pumps for the oil and gas industry, is using our software for predictive maintenance.
The second largest U.S. utility is using our technology to improve the efficiency and reliability of their most expensive spinning assets.
Generally speaking, predictive analytics is a fast-growing industry. The oil and gas sector, however, has been timid to adopt the approach. The biggest barrier to the use of predictive maintenance in drilling is related to the limitations of traditional approaches.
Under the constraints of machine learning, a model typically requires 15 failures as part of the training process of the model that will be used when the technology is deployed.
In the oil and gas industry, we typically don’t have this number of failures available, as we tend to perform a lot of preventive maintenance (at a very high cost) to ensure equipment does not fail.
Also, when a failure does occur, we tend to replace a lot of components. We do this sometimes without understanding which specific components failed first and why.
With an automated model building approach provided by our patented algorithms, we find trends in data with a very limited amount of failures. In fact, we have built reliable models in situations where we had no failures available to train the model.
Taking the World By Storm
For as long as I have been in the oil and gas industry, I have heard the term “artificial intelligence.”
Twenty years ago, the most advanced artificial intelligence was able to state, “six months ago, you could maybe, possibly have avoided failure with some sort of preventative maintenance.” Ten years ago, AI was telling us, “six months ago, you should have taken a specific preventative action.”
Today, with the advances in AI-powered software, and sensor hardware, we are now able to look at very large amounts of data and give real-time responses on the best future course of action.
AI is taking the industry by storm.
With predictive analytics and cognitive security, you can be sure that your equipment is operating safely and securely. You could also receive recommendations on action to take to avoid future equipment failure or remediate a security breach.
The measurable value of predictive analytics comes from a significant reduction in the operating cost, while the biggest savings are in the increased safety brought to both personnel and equipment.
Today, the use of predictive analytics is improving the operations of the O&G industry’s early adopters. Using AI for drilling optimization and integration to other corporate applications and corporate knowledge is starting to be explored.
Natural Language Processing can query maintenance manuals to instantly obtain a maintenance procedure for a specific asset under question.
The promise of AI is already starting to be realized in the O&G industry. Early adopters are taking advantage of their position to get a head start on the competition.
The industry has always leveraged technology to adapt to change, and the early adopters have always benefited the most.
As the oil and gas industry continues to be more and more competitive, companies cannot afford to be left behind.
However, if companies can understand the opportunities inherent in adopting cognitive technologies, their future surely looks bright.