Sensors and Knowledge Required to Reduce Threats to Aging Industrial Infrastructure
Knowledge and aging industrial infrastructure require organizations to think about the knowledge embedded in their infrastructure, and the knowledge required to embrace digital transformation.
For most factories, plants, and industrial facilities no artificial intelligence monitors the status of far-flung equipment, and what little data flows from the amalgam of aging equipment does so to individual devices. Some of the primary and secondary equipment in those facilities predate the Internet, let alone the Internet of Things. Technicians with years of knowledge in their heads keep things running through inspection and preventative maintenance repeated in monotonous, pre-programmed routines. Many technicians will retire before the equipment they maintain.
Aging industrial infrastructure puts many organizations in peril should they lose too much knowledge from their retiring employees. All of them likely suffer from suboptimal infrastructure performance caused by infrequent inspection and over-stretched duty cycles.
Aging industrial infrastructure puts many organizations in peril should they loose too much knowledge from their retiring employees. All of them likely suffer from suboptimal infrastructure performance caused by infrequent inspection and over-stretched duty cycles.
The solution to this problem is twofold. refit older equipment with sensors and capture knowledge from maintenance and repair employees. These two investments together will provide for an improved span of control and prepare for more automated remote monitoring of equipment.
The Case for Sensor Refits
Learning During Knowledge Capture
Because expert technicians know what they know based on their current environment, changes in that environment can both validate and challenge that knowledge. As organizations increase their use of sensors on older equipment and replace older equipment with connected devices, the new data may provide additional insights that do not fit into the existing mental model for technicians.
The periodic capture of data is very different than experiencing a continuous stream of data in realtime. More data, more context, more potential for insight.
Organizations need to explicitly include learning time in their knowledge capture investments. They cannot assume that all of the knowledge employed by their technicians will hold true once captured. Differing points of view will be brought to light and will need to be reconciled. Existing assertions and beliefs will also require modification as circumstances change, including deploying new equipment, the building of new facilities, and the availability of more detailed data.
While current technicians certainly have the most qualified perspectives on existing operations, and perhaps the best potential to rapidly infer implications in new facilities, they can only provide consistent insights if their organizations recognize that shifting circumstances require learning and that the technicians are given time to assimilate emergent knowledge into their internal and external models.[/framed_box]
The lack of data and holistic operation visibility results in older facilities facing several threats with varying degrees of severity. They face increased energy use because inefficiently maintained equipment draws more power. They face a variety of pollution issues from increased emissions to leaky fluids. Heat pollution creates not only more stress on equipment, it also causes other equipment, like chillers, to work harder.
In primary systems, failure also results in shutdowns with an immediate impact on productivity and revenue. In secondary systems, productivity, fiscal, and remediation may cost less in individual instances, but even sporadic failure results in an ongoing drag of efficiency. In rare cases, secondary failures may cascade into more systemic outages.
Older equipment in both primary and secondary systems should be considered for retrofits with sensors that can measure electrical, thermal, or vibration information, along with environmental conditions like humidity and ambient temperature. The sooner this is done, the more likely organizations will be able to correlate existing knowledge with the new data while experienced technicians are still around to help them make sense of what they are seeing.
Knowledge and Aging industrial infrastructure: Capture Knowledge Now
Knowledge capture (see What Knowledge Should We Capture below) can assure continuity while sensors will increase the frequency of data capture.
Organizations need to start with an inventory of what knowledge they need to capture, decide on the best technology in which to store and retrieve that knowledge, and then do the work to collect, correlate and codify knowledge in a way that makes it easily accessible through both search and structure.
A simple repository of existing documents made searchable will not prove sufficient. Without creating a structure (by equipment type, environment, problem mode, etc.) it will be difficult to assure inclusive search that returns all relevant information. Structure will also be required when building the model and rules for any future automated monitoring system.
Declarative knowledge or facts. These facts include what a piece of equipment is, where it was purchased, its bill of material, its maintenance history, and any special names, or nomenclature used in reference to a piece of equipment or a process. Metadata, or data that describes other data, is also a form of declarative knowledge.
Procedural knowledge captures task-level, sequential procedures. In industrial applications, procedural knowledge includes how to test a piece of equipment, and how to repair it—and it also includes frequency and other timing considerations.
Semantic knowledge relates things together at the very deepest level of understanding. For industrial applications, this may include information about how new product prototype runs affect equipment in a different way than long production runs. It may include correlation information about different sensors, as well as how different types of equipment related to one another. Semantic knowledge also includes legal and safety regulation contexts related to workflows.
Episodic knowledge records events. Case studies represent examples of episodic knowledge. A case study may involve facts, procedures, and semantics, all tied together by an event or events. Episodic knowledge is important in maintenance when a technician shares precise thinking and sensing associated with a specific diagnostic event. Walking into a particular room, what does it smell like, what does it feel like—what diagnostic thoughts does the situation inspire that may not be part of the documented procedure. Episodic knowledge is often best capture through storytelling protocols.
These forms of knowledge exist in three different states. Explicit knowledge has already been recorded in a tangible form (document, voice recording, video). Implicit knowledge exists in the head of a technician but it has not yet been recorded anywhere. Tacit knowledge is knowledge known to a technician but even he or she may not know they know it, or they may have a difficult time finding a way to express it.
To have a good handle on a knowledge domain, organizations must recognize and embrace all forms of knowledge capture or risk incomplete approaches to diagnosing, solving, and avoiding problems. Knowledge exists in ecosystems that evolve. Most organizations should include a combination of community and content so they can capture the changing nature of knowledge and the continuous learning of community members. Communities are also the best way to work through expression issues with tacit knowledge.
One example of knowledge that might not be considered critical at first: a company runs the same pumps in different locations. The pumps come with a recommended maintenance schedule. Senior maintenance staff knows that equipment in warmer, humid climates requires more frequent attention than equipment in cooler, dryer locations. They share this knowledge with new employees, but they don’t reference it in any of the manuals provided by the original equipment manufacturer. While this information may eventually become apparent from sensor data collection, it could take months, if not years, to accumulate enough data to back into such a heuristic.
Once sense-making systems get put into place, It is crucial that as much knowledge as possible be incorporated into these systems to rapidly validate or modify it to reflect actual operational situations.
Knowledge+Data Saves Lives and Money
In a food processing plant, a temperature sensor goes off telling the on-duty supervisor perishable ingredients at one location face immediate risk of spoilage. While the plant monitors the refrigeration system temperature, the actual equipment itself only undergoes periodic inspection. This failure could result from condenser fins restricted with dirt, reducing their airflow—or it could be that stressed bearings failed due to plugged inlet screens in the compressor sump or clogged lubrication to the crankshaft—or perhaps the suction pressure was lowered, increasing the compression ratio, resulting in high discharge temperatures.
Many of these issues could have been solved ahead of failure by monitoring sensors on crucial subsystems rather than looking only at temperature. In another situation, an equipment failure might result in the loss of life rather than the disposal of spoiled food.
Looming and immediate threats stem from not having instruments in place to capture the data needed to monitor facilities remotely, and in realtime. This is true for all types of industrial equipment from pumps that move the liquid around food processing plants to rotary transfer machines. The lack of data blinds even the most experienced employees to overstressed equipment facing potential failure conditions.
The lack of data blinds even the most experienced employees to overstressed equipment facing potential failure conditions.
Industrial organizations and those who operate controlled facilities will always need their technicians. By investing now in knowledge capture, they will be able to better leverage data as it becomes available and be better prepared to train the next generation of technicians to improve on this knowledge.
Access to sensor data through analytics systems and dashboards will empower current technicians to operate more efficiently. It will also allow them to validate their assumptions and practices ahead of any future automation.
Additional reading from Serious Insights:
Did you like thinking about knowledge and aging industrial infrastructure? For more Serious Insights read this post:
Daniel W. Rasmus, Founder and Principal Analyst of Serious Insights, is an internationally recognized speaker on the future of work and education. He is the author of several books, including Listening to the Future and Management by Design.