BSquare’s senior solution architect, Matt Smith, spoke as a keynote at TMC 18 on IoT adaption especially as it relates to vehicles.
“Experience has shown that moving along the path to full IoT can be broken down into five stages,” says Smith. “Each stage provides increasing levels of benefits. Progression through all five stages culminates in a holistic, intelligent, automated IoT system that delivers the broadest range of positive outcomes for multiple business goals.”
Adopting the right data for the right adaptation
Don’t adopt IoT for IoT’s sake, Smith says. Instead, focus on specific business use cases and outcomes, using IoT as the basis for an overall strategy to achieve those goals. It’s important to look at IoT not just as a journey but as a progression as organizational needs and adoption changes.
The Big 6 problems to solve are asset utilization, predictive failure, adaptive diagnostics, IoT device management, asset optimization, and condition-based maintenance. IoT offers ways to structure data in a code-like construct. Show computers (or AI) in your data what goes wrong and how to fix.
“My North Star is predicting when,” says Smith.
The more data the better
From device connectivity to data monitoring to data analytics to automation. The solution sweet spot is ultimately automation.
“If this, then that,” says Smith. “When you’re at automation, you have learned how to take advantage of the data from all the computational possibilities, and you know how to effectively harvest the data.”
Especially for fleet operators — how do we know what’s going on without going through reams of data? And how are the repairs done?
“My North Star is predicting when.”
“Asking the right questions. That’s a fine place to start,” says Smith. “Start with what you’ve got, and what’s available. It doesn’t even need to be transmitted.” Real-time monitoring is a “step in the right direction.”
But in the long view the goal is to be able to address complex failures through smarter, adaptive diagnostics. Condition-based maintenance becomes possible. For richer, smarter diagnostics more data and input is better.
Less data equals less accuracy. This might mean, for example, that instead of being able to predict equipment failure with 80-90 percent accuracy, predictions can only be made with 40-50 percent accuracy. This makes systems less valuable to the point where it even drops ROI to unviable.
What do you do with all the data? How do you do something with it so that all your data isn’t just sitting around in one place? First steps are adding real-time things like on-board monitoring, event direction, and value processing.
From “If, then” to “If a, b and c and NOT d (and time window <e minutes) and (asset model=x) then f,” says Smith.
Such programming is more complex to achieve, but will ultimately yield better results through greater accuracy.
Predictive failure a case study
Many oil and gas operators employ “run-to-fail” strategies with regard to production equipment. This is because, historically, the cost of periodically testing and maintaining equipment at widely distributed sites exceeded the cost of simply replacing the equipment when it finally failed.
But with IoT, this is no longer the case. The ability to automatically predict equipment failure with sufficient advance warning to allow smooth remediation without entailing unplanned downtime changes that cost equation considerably.
Run-to-fail may have, in the past, been the most cost effective strategy but it still resulted in unplanned production downtime. With IoT, not only is the cost of monitoring and predicting equipment failure substantially reduced, unplanned downtime can be eliminated.
“From ‘If, then’ to ‘If a, b and c and NOT d (and time window < e minutes) and (asset model=x) then f.’”
Adaptive Diagnostics
Even with the ability to accurately predict equipment malfunctions, failures will still occur, albeit less frequently. Another critical aspect of IoT is the ability to assist in the diagnostic process and bring equipment back online more quickly, thereby further reducing production downtime.
In a typical adaptive diagnostics scenario, equipment remediation steps have been predetermined before repair technicians are even dispatched. Predetermined diagnostic steps are based on real-time and historical operating information from large populations of similar equipment and can pinpoint likely repair steps with greater accuracy than traditional troubleshooting. On the maintenance side, repair technicians can be outfitted with the right equipment.
Of course all this takes time. BSquare discusses the maturity of lloT (Industrial Internet of Things) in a recent white paper, and the industry is certainly still maturing.
Something else for an organization to strive for is called “condition-based maintenance” and “asset optimization” that play into the data analytics of IoT solutions. Upstream oil and gas production operations are complex and expensive yet vital to the national economy. Improving the efficiency of these operations has long been of paramount importance but information technology. With oil and gas operations digitalized, it is now possible to substantially improve production efficiency while reducing costs.
Best Practices
Overall, your best practice is to start with an operational goal so you can have a variety of inputs. Decide how and to what extent your data might be shared. Know that component data content varies greatly. Diagnostic data supported between providers varies widely. Know what the underlying data is meant to support.
Finally, ask yourself: what am I contributing to the industry?
Stay up-to-date with the latest commentary and insights on FreightTech and the impact to the markets by subscribing.
