Closing the Analysis-to-Action Latency Gap
When an inspector tags a pump as ‘Out-of-Order,’ your revenue doesn’t just slow down, it stops. The analysis-to-action latency gap compounds your losses and creates problems that weren’t there to begin with. Although you may start by solving this one persistent problem, the ultimate goal is to implement prescriptive AI oversight to your Oil & Gas business.
A Real-World Example
Consider an example. Assume that an ISD system detects that positive pressure in the underground tank headspace has exceeded 1.5 inches water column. The system logs a ‘First Period’ overpressure event. The station manager sees the warning light but assumes it is just winter volatility (assuming it’s winter). The manager waits 24 hours to see if the problem clears itself. By the time they call an ISP, the system is in a second consecutive period of overpressure.
The software triggers an automated shutdown of all fuel dispensers. The analysis-to-action latency, in this example, could be as much as 32 hours. The cost to the retailer is not only an emergency service call, it’s the lost revenue from having an out of order tag.
Yes, Ghosts Are Real
Yes, during winter months, many sites can be subject to a high “ghost alarm” rate. That’s to be expected. But the cost of unnecessary truck rolls in this situation can lead station managers to take a “wait-and-see” approach. That is a problem with traditional ISD systems. They generate data but don’t indicate root causes of problems.
Decreasing Latency with Predictive and Prescriptive Analytics
Using predictive and prescriptive analytics, you can decrease the analysis-to-action latency by a significant amount. Using operational AI, the system can now identify a pressure spike, as described, early on and correlate that spike with RVP data.
Instead of waiting for a station manager to notice and then take action, the operational AI system can determine when an alarm is not a “ghost alarm” but instead is a failing P/V vent valve. This predictive capability is based on the current and historical data that the operational AI has collected and analyzed, about how your systems work.
Prescriptive Analytics
Using prescriptive analytics, the system understands the probability of certain issues, based on past history, but also knows how best to solve the problem most efficiently. It automatically generates a repair order and sends it electronically to an ISP, or even to a specific local technician who is close by and equipped to fix the problem.
The operational AI cuts the time between problem identification and a repair action to a fraction of what a traditional ISD system requires. The situation never reaches a second period of overpressure (in the example).
Identifying Micro-Leaks
Micro-leaks are extremely difficult to locate and assess. Buried tanks and connections are hard to reach. But, performing an assessment of your delivery systems can identify potential problem areas. For the ISP, this is critical information that helps them meet their SLAs and save money and time. Optimizing technician time is one of the best ways to increase slim margins.
Deploying Continuous Pressure Decay Analytics
Completely trusting a 3-year inspection to catch micro-leaks is a leap of faith. Your customers’ underground systems could potentially develop a micro-leak (defined as ($< 0.1$ gph), which means less than one-tenth of a gallon per hour or about the amount in a standard soda can every hour) the day after your inspection test. Static pressure testing is like using a telescope to study single-celled animals. You need to shift the investigation process to Continuous Pressure Decay Analytics.
Typical regulatory tests are snapshots that only verify that the system is contained at one specific moment under artificial pressure. In contrast, predictive and prescriptive AI sensors continuously monitor pressure decay. That’s right, 24/7. The sensor data is collected and analyzed by the operational AI to figure out if the vapor loss is from a leak or is just some sort of environmental noise.
That is why these types of tests are best run in the middle of the night, when there is no fuel in the lines and no noisy operations occurring. In any case, the analysis is much more granular than you get from static pressure testing.
Deploying a High-Frequency Data Layer
Standard ISD systems typically log data over long intervals. To detect a micro-leak, the IoT gateway needs to pull data from the RS-485 serial bus or the Current Loop at a much higher sampling rate and for a continuous period of time:
- Headspace Monitoring: The system monitors the underground storage tank (UST) ullage pressure in increments of 0.01 inches Water Column (WC). This is enough to help identify a trending pressure rise, and to dispatch a technician to fix a P/V valve before the system hits the regulatory failure threshold of 1.5 inches WC. Continuous monitoring means that an alert and action will be taken even if it happens in the middle of the night.
- The 2 AM – 4 AM Isolation Window: The AI focuses on late-night periods when there are no active fueling transactions underway. Since most fueling stations are closed at night, or customer traffic is very low, environmental noise is low. This allows the sensors and AI to determine potential leaks with a higher level of confidence.
Differentiating “Vapor Shrinkage” from “Mechanical Breach”
Of course, fuel naturally expands and contracts with temperature changes. This is one reason why it is so difficult to differentiate between normal vapor shrinkage and an actual mechanical breach in a pipe or connector. Operational AI solutions perform some sophisticated analyses to note that differentiation.
- Thermal Compensation: The AI correlates pressure data with temperature sensors in the fuel and the soil. If the pressure drops but the fuel temperature is also dropping, it may signal normal shrinkage.
- The Decay Signature: When a sensor detects a micro-leak at an elbow joint or a loose gasket, the data displays a distinct mathematical decay curve. Unlike thermal shrinkage, which eventually stabilizes, a micro-leak displays a steady, linear decline in pressure that never levels off even when temperatures are constant. This means that pressure is dropping regardless of what the temperature is or how it changes.
Signature Analysis & Fingerprinting
Standard tests just tell you “it’s leaking.” Continuous AI analytics give you a hardware fingerprint to tell you where something is most likely leaking.
- Micro-Leak Profiling: The AI can use its understanding of historical data and current data by analyzing how the pressure behaves and has behaved in the past. If the AI determines that the decay signature isn’t a normal case of evaporation, then it can distinguish between a leak in the liquid-side piping vs. a leak in the vapor-side headspace by comparing those two decay signatures.
- Geometric Risk Weighting: Weak points in these fuel systems are generally well-known throughout the industry. The AI system applies higher scrutiny to these known weak points. As noted earlier, up to 19% of piping failures happen at elbow joints. These are typically the spots where pressure and turbulence rise as fuel hits a 90-degree angle.
While modern installations have moved to more corrosion-resistant materials, like fiberglass reinforced plastic and high-density polyethylene, many legacy installations exist throughout the country. These often contain steel piping or a combination of steel and newer materials. And while researchers are making progress with non-jointed piping, fuel delivery systems still require the use of elbow joints and connected piping. Even with better technology, joints are the weak links in a system.
The Transition to Prescriptive Intervention
Once the AI identifies a micro-leak, it shifts into prescriptive guidance mode:
- The “Pre-Fail” Alert: Instead of waiting for a catastrophic failure that triggers a regulatory shutdown, the system prescribes a specific maintenance task: “Check gasket integrity on Tank #2 fill cap” or “Verify torque on Dispenser #4 elbow joint.” When using the traditional ISD systems, these tasks might turn out to be ghost alerts. However, in those less-intelligent systems, you wouldn’t get the level of detail that shows up in the task alerts noted above. Instead, you’d see something like “Service manager noted a persistent drop in tank pressure over the last 12 hours,” or something equally vague.
- Revenue Protection: By catching the leak at the 0.05 gph or 0.1 gph stage, the ISP can fix the issue during a scheduled monthly inspection, avoiding the “Out-of-Order” tag that stops all revenue. Decreasing the gap between alert and fix also means that pumps are available for use and suffer lower losses due to out-of-order issues.
Making RRIM Work for You
If you’re not already using the PHMSA Risk Ranking Index Model (RRIM), from now on referred to as The RRIM Framework or RRIM, there’s no time to lose. The framework can be simplified for use in the field. Several important factors are given a score, based on inspection tests or visual inspection. The five primary factors a technician must grade include:
- Material Integrity – Bare pipe or ineffective coating, for example, increase the threat score.
- Enforcement History – Previous violations or notifications act as leading indicators of future risk.
- Commodity Threat – The volatility and hazard level of fluid being moved (either on a transport vehicle or through a pipe, for example).
- Time Since Last Inspection (TSLI) – The longer an asset goes without a deep dive audit, the higher the risk.
- Location/Consequence – How close an asset is to “high consequence areas” (HCAs) or sensitive environmental zones.
Historical documentation of these inspections and RRIM evaluations can serve as powerful input to an operational AI system. This data can help prioritize preventive maintenance tasks and can ultimately lead to a more advanced predictive maintenance automation. Use RRIM to grade as many elements as possible at each site. The operational AI can build a dynamic and up-to-the-minute view of all systems, allowing it to make highly educated decisions about maintenance and compliance issues.
Unifying Disparate Systems
Bridgera helps you create a unified system, from software to hardware. Each of your systems, like SCADA, ERP, or ATG, perform specific tasks, but when you are able to truly define the relationships between the data these tools produce, you can use all of that data much more effectively. Most of these tools are recording the state of a system or component at a given point in time.
The real value in data is when you can see how it changes over time and how the performance of one system impacts other downstream systems. Bridgera is experienced at helping companies understand and automate their data collection and evaluation processes to provide superior decision-making capabilities. Unified Operational AI for Oil & Gas takes things a step further by building in a response framework that can take action on many routine issues that otherwise require multiple service calls.
How Enterprise Organizations Optimize Field Logistics
Automating the more routine and predictable aspects of fuel storage and distribution frees up your valuable labor resources, otherwise known as your technicians, to focus on more important or urgent tasks. By increasing the amount of “wrench time” and decreasing technician “windshield time,” you expand your company’s revenue opportunities. The company can more quickly scale by automating routines and can focus service on more valuable types of projects.
Frequently Asked Questions (FAQ)
1. What is “analysis-to-action latency” and why does it matter to fuel retailers?
Analysis-to-action latency is the time gap between when a problem is detected and when a corrective action is actually taken. In traditional ISD systems, that gap can stretch to 32 hours or more. Every hour in that gap represents lost revenue, potential regulatory exposure, and compounding damage that may not have existed had the system acted sooner.
2. How does operational AI distinguish a real alarm from a “ghost alarm”?
Ghost alarms are a persistent and costly problem with traditional ISD systems because they can trigger an Out-of-Order tag. Older systems generate data without identifying root causes. Operational AI can differentiate between a routine winter pressure spike and a genuinely failing P/V vent valve by looking at known failure patterns. That distinction eliminates unnecessary truck rolls while ensuring that real issues trigger immediate action.
3. How does AI detect micro-leaks that standard inspection methods miss?
Standard regulatory tests are snapshots that verify containment at one specific moment under artificial conditions. A micro-leak can develop the day after a 3-year inspection and go completely undetected until it becomes a much larger problem. Continuous Pressure Decay Analytics addresses this by monitoring underground tank pressure 24/7. Critically, AI can also distinguish between normal thermal vapor shrinkage and a true mechanical breach.
4. Can AI actually tell us where a leak is occurring, not just that one exists?
Yes, and this is one of the most significant advantages over traditional testing. Standard inspections can tell you a system is leaking but operational AI can tell you where it’s most likely leaking and what the probable cause is. Through micro-leak profiling, the AI compares liquid-side and vapor-side decay signatures to isolate the location. It also applies geometric risk weighting, focusing extra scrutiny on known high-risk points like elbow joints.
5. How does RRIM fit into an AI-driven maintenance strategy?
The PHMSA Risk Ranking Index Model (RRIM) provides a structured scoring framework across five key risk factors: material integrity, enforcement history, commodity threat, time since last inspection, and proximity to high-consequence areas. When this historical inspection and scoring data is fed into an operational AI system, it becomes a powerful input for dynamic risk prioritization.
About Bridgera
Operational Intelligence. Production-Ready AI.
Bridgera partners with operations-heavy enterprises to move AI beyond pilots and into real production systems. Through AI consulting, specialized talent, and scalable platforms like Interscope AI™, Bridgera embeds intelligence directly into the operational workflows that power the business.
