Predictive maintenance is about collecting, integrating and processing data sets from different sources, notably data sets that could provide insights into the current and future status of the assets of your plant floor.
In this post, we present five different modalities that can serve as a basis for implementing a predictive maintenance approach. Beyond predictive maintenance, these data sets could also boost your condition monitoring and asset management operations.
As part of our everyday driving from home to work and vice versa, we all know pretty well that the vibration felt in our steering wheel could be an indicator of mechanical problems.
The same holds for components and equipment in your plant, which has led to the extensive use of vibration monitoring as a means of assessing assets’ condition.
In principle, vibration is measuring the level at which an elastic body or medium moves beyond its position of equilibrium. Hence, vibration is in practice perceived as an oscillation or trembling motion, while vibration monitoring is about measuring motion around a point of equilibrium.
Machines and equipment are comprised of many parts, which have their own unique pattern of vibration (i.e., their vibration is characterized by various levels and frequencies). As a result, vibration data collection needs to take place for several components (e.g., oil pumps, generators, hydraulic pumps, pistons, valves), which might have similar frequency values.
Following data collection, software for vibration analysis, such as signature analysis software, can be used in order to accurately detect the parameters (i.e., displacement, velocity, acceleration) that define the magnitude of the vibration.
With the magnitude of the vibration and individual contributing parameters at hand, the condition of the equipment can then be assessed.
A typical vibration monitoring solution uses transducers, like sensing devices, in order to measure the vibration and translate relevant information to some electronic signal. Instead of mounting transducers to the monitored part, handheld devices can be used to transform the vibration to an electric signal.
In several cases, handheld devices come together with the instrument that measures the vibration magnitude, given input from appropriate probes. Alternatively, transducers can directly interface with a computer running the analysis software.
Even though oil analysis is one of the oldest techniques for monitoring the condition of an asset, it is still used extensively.
In principle, oil analysis has three aspects:
The rationale behind conducting an oil analysis is that as the equipment starts to wear, heat and oxidation lead to the formulation of small microscopic particles within the oil.
In practice, oil analysis is based on a number of physical and chemical related tests, including viscosity monitoring, moisture/water analysis, monitoring acids and bases produced by heat-generated oxidation, as well as the detection and analysis of wear metals in the oil.
Conducting these tests requires specialized equipment and therefore, several plant operators opt to outsource this task to a laboratory, rather than doing it in-house. No matter whether the analysis is conducted in-house or is outsourced, measurements can be entered in a database in order to enable their use as part of a BigData deployment for predictive maintenance.
Operating equipment produces a wide range of sound patterns, including patterns in the ultrasonic frequency spectrum which are not audible to humans.
Ultrasonic wave emissions have a frequency level above 20 KHz and are indicative of the equipment’s condition. High-frequency ultrasonic waves are short and expose directionality, which facilitates their separation from background plant and machines’ operating noise.
When there are changes to the condition of the mechanical equipment, ultrasound sensors are able to detect changes in the directional sound patterns. In this way, ultrasound sensors can provide insights about under-lubrication, over-lubrication and timely detection of wear.
One of the main benefits of the ultrasound sensing modality is that it can lead to early detection of asset degradation, being in this respect even more efficient than vibration or infrared modalities. Furthermore, ultrasound sensing can be used in cases where vibration cannot work efficiently as a result of the very high energy nature of the emitted elastic waves.
Temperature sensing another sensing modality that can indicate problems with the condition of electrical and/or mechanical components in the plant.
In the case of electrical components, high temperatures are associated with reduced operating life due to damages in insulation and increased resistance of conductor materials. Likewise, in the case of mechanical equipment, rising temperatures indicate overload or a need for lubrication.
In general, overheating is a signal that should alert plant operators, since if it is not investigated, it can lead to equipment damage and personal injury.
Temperature sensors, such as infrared thermometers, can be mounted in the plant floor in order to not only enable maintenance applications, but also assist production quality auditing, environmental monitoring and safety checks.
In several cases, hand-held, non-contact, infrared thermometers can be also used in order to measure temperature in hot, moving or even not easily accessible objects.
Equipment condition insights can also be derived based on the analysis of visual images that depict variations in the infrared radiance of objects’ surfaces. Infrared imaging is affected by high or rising heat from resistance in electrical circuits.
The process of generating the visual images of the equipment is called thermography. Note that the success of maintenance applications based on thermal imaging presupposes that accurate images are obtained.
This requires not only the use of modern IR cameras, but also experienced operators that know how to capture images that can reveal condition information. The task involves using and orienting the devices in a proper way, while at the same time avoiding conditions that could lead to inaccurate or even false data.
Beyond the proper capturing of data, expertise in the analysis and interpretation of data is also required.
Each one of the above-listed modalities can be used on its own for the implementation of condition monitoring and maintenance programs, including both preventive and predictive maintenance.
In this era of industry digitization, BigData technologies can enable the combination of data sets from all the above modalities as part of a holistic approach. This practice can lead to accurate predictions of the health status of assets, including hidden patterns of assets’ degradation that can be hardly detected based on any of the above modalities alone.
Consider what kind of maintenance data you collect today and how you could combine data to enhance accuracy and/or improved predictive maintenance programs.
John Soldatos holds a Phd in Electrical & Computer Engineering. He is co-founder of the open source platform OpenIoT and has had a leading role in over 15 Internet-of-Things & BigData projects in manufacturing, logistics, smart energy, smart cities and healthcare. He has published more than 150 articles in international journals, books and conference proceedings, while he has authored numerous technical articles and blogs posts in the areas of IoT, cloud computing and BigData. He has recently edited and co-authored the book “Building Blocks for IoT Analytics”.