So Many Risks and Faults at Stake

Author: Jon Herlocker


Some take the reliability of critical equipment and systems for granted, but those tasked with keeping them operational are keenly aware of the threats. Because there are so many potential risks and faults to manage, advanced condition monitoring and analytics are needed to hold them at bay. The Tignis solution can detect a minimum of eight different types of errors or unfavorable conditions, enabling a comprehensive path to reliability and optimization for connected industrial systems.

While most plants do not have the necessary data to immediately start utilizing supervised ML, many plants have collected at least a few months of historical sensor data. With this data, here are examples of classes of faults that Tignis can detect using anomaly detection and engineering principles:

  • Mis-sized or out-of-spec equipment: Tignis can detect incorrect or improperly sized equipment. For instance, a pump that had been shipped from the manufacturer with the wrong size impeller inside could have produced devastating outcomes had it not been discovered by the software. Another example is valves not sized for the expected differential pressure, which can make them hard to control.
  • Out-of-control / unstable controls: Almost every plant that Tignis monitors has an automated control loop that attempts to maintain certain setpoints. Control loops must be tuned to ensure that they are stable and will always converge. Just as a car constantly overcorrecting will swerve back and forth, posing a dire hazard, issues like this can occur in control systems. They are technically “out of control” not only because they may never reach the setpoint in some cases, but the oscillations can propagate through the system and cause havoc in unpredictable ways. Tignis has detected and resolved issues where systems had multiple control loops that interacted with each other in unpredictable ways.
  • Sensor failures: Stuck sensors, bias sensors, drifting sensors, mis-calibrated sensors—Tignis has discovered these and more. Failed or failing sensors can invalidate decisions made based on their readings. In systems with automated control loops, a bad sensor means an incorrect control decision. Depending on how bad the sensor error is, outcomes could be catastrophic.
  • Mechanical wear and obstruction: When the right sensors are available, Tignis can pay close attention to the amount of work a system is doing and compare it to the power utilization of the associated assets. When changes or trends in asset or system efficiency are identified, it often can be traced back to problems such as a bearing that is starting to wear, a filter that is becoming clogged, fouling on the inside of pipes, or stuck and leaky valves or dampers. Identifying these issues when they are just emerging provides time to avoid critical conditions and failure.
  • Hidden component failures: Tignis brings hidden conditions to light. For example, large and complex monitored systems often have redundant components. In the case of multiple fans ventilating a cleanroom, when one fan fails, another can provide the full service, but that “backup” fan is now the “critical” system as it does not have its own backup. If Operations is unaware of the primary fan’s failure and need for repair or replacement, it puts the cleanroom at risk. Another hazard is intermittent or momentary component failures, which can precede a complete failure. These, too, may be missed by Operations if not automatically detected.
  • Automation programming errors: PLCs and similar control hardware must be programmed. The quality of automation programming can be highly variable, particularly with respect to handling exceptional conditions. When a system enters a state that was not predicted by the programmer, unexpected results may occur and lead to unanticipated and potentially catastrophic situations. Tignis’ automation can detect some or many of these errors by monitoring the resulting physical properties.
  • Incorrect schematic data or incorrectly labeled sensors: P&IDs are not always updated correctly, and sensor tags are frequently wrong or undecipherable. These issues raise the potential for detrimental events, as both humans and computers may be making decisions based on faulty knowledge of how the system is built or currently being operated. Tignis can help to detect such errors so they can be proactively corrected.
  • Violations of best practices or regulatory compliance: Most plants have well-documented design standards for equipment, and some plants are subject to governmental regulations. The standards and requirements can be encoded in Tignis’ proprietary query language and continuously monitored to prevent violations.

Staying on top of all the possible threats is easier with an intelligent, physics-driven condition monitoring and analytics solution. There is no better way to keep your critical assets running reliably and help your company consistently meet its performance objectives.