In part 1 of this series, we talked about the elements of a Condition Monitoring system and their relationships. We stressed the importance of thinking about it as a continuous cycle that drives bottom line improvements to your overall Asset Management strategy.
The five steps involved in implementing an Asset Health Condition Monitoring Program are as follows:
- Identify the assets and their models
- Define and understand the variables
- Select monitoring equipment and establish a data collection and storage system
- Define and follow a consistent condition assessment and health ranking system
- Integrate the condition monitoring system into your existing maintenance strategy
1. Identify the assets and their models.
The first step is going to set the tone of the rest of your program. It will help you identify what variables you need to measure, what data you need to collect and how to transform this into valuable decision-making information. Depending on the reliability and availability goals of your organisation, knowing the health condition of a particular asset will help you decide the type and timing of the maintenance that needs to be performed on it.
First you will have to select the assets that will be monitored. A good place to look to get a sense of which assets are worthy to be included in this program is your equipment criticality matrix. Typically, those assets with higher associated risks and higher financial or safety consequences of failure will make a good case to be included in the Condition Monitoring program. Validate your selection criteria with the appropriate stakeholders within your organization including Finance, Operations and Maintenance. Achieving early buy-in from the team will make it easier to get the resources you’ll need for implementation in the future.
Once you know which assets need to be monitored, you have to decide which condition and health assessment models you are going to use. I would venture to state that these models are one of the most important components of a Condition Monitoring system. The models are the “translators” that will help you transform the data being collected into a useful diagnosis that will indicate the current “health” of the asset, whatever health means (more on this later). Without them your system is that, just data.
These models might take various forms and I would encourage you not to think about a specific embodiment of the model but rather the output that it should produce. A model is a conceptual representation that helps us visualise this “black box” that takes data as an input and produces a condition assessment as an output.
The model you choose will depend on the type or class of equipment for which you are trying to assess the condition of.
Over the years since the industrial revolution, the various engineering disciplines have reached varying degrees of understanding on different classes and types of equipment. This means that for each class of asset there will be more or less advanced condition assessment models available to you. The best way to approach this selection process is to consult subject matter experts in the various classes of equipment that you plan to include in your program.
These models might take the shape of a simple “if” rule, a software tool, or even be an organisation or expert that will interpret the data that you give them produce a judgment about the current condition and health of the equipment.
The output of this step is a list of assets and associated condition and health assessment models.
A simple example would be that of a chute. The internal walls of the chute are fitted with wear plates which have the function to protect the walls of the chute and maintain its mechanical integrity. As the plates wear, their thickness decreases until they are too thin to perform their function reliably and have to be replaced.
In this case, the condition of the chute liners is determined by its thickness. Each plate will wear at a varying rate depending on factors such as: ore abrasiveness, position of the plate within the chute, ore speed, average lump size, etc.
|100% life (new)||150 mm|
|75% life||130 mm|
|50% life||110 mm|
|25% life||90 mm|
|End of life (replace)||70 mm|
A simple way of describing this is: CONDITION = f (WEAR PLATE THICKNESS).
The model should be capable to describe the condition of the asset when the factors of the model are known. In our example it would be something like the following table.
What we need as an output of the model is an assessment of the condition, which in our case is directly related to the wear plate thickness. Often the condition of the asset will have to be described by more than one factor with more complex relationships. My advice would be to make the model only as complex as necessary to achieve an acceptable degree of confidence in our assessment of the condition. Thickness is a sufficient descriptor of a wear plate’s condition and that’s where we stop. We could add more parameters and make the model more complicated but we won’t because it’s not necessary.
One more thing, once you select a model for a particular class of asset, stick with it. Or at least make sure that any changes are properly documented. Five years from now, when you have moved on, the person managing the system has to be able to obtain consistent health assessments as additional data is collected and understand what “tweaks” or changes have been done to the system over time. By doing this, you are building robustness into the system.
After having identified what assets and models you will be using, the next logical step is to properly understand what parameters you are trying to measure and why. In this sense the models themselves will be a great guide in terms of what inputs are required to make them work.
One of the common pitfalls you have to avoid is going “measuring happy” or as one colleague used to put it, turn your plant into an “engineer’s playground”. We engineers love to play with data and make nice trend lines and charts in Excel™, even if at the end we don’t really know what to do with them.
In order to establish an adequate Condition Monitoring system, it is essential that the parameters to be measured are carefully selected and defined. If a variable is to be worthy of capture and storage, it has to be a significant indicator of the asset’s condition or play an important role in the overall health model for a particular piece of equipment.
The chosen variables will have to be clearly mapped to a model, even if it’s a simple one. The simple rue is: if there is not a clear relationship between the variable being measured and the model, don’t include it.
Continuing with our chute example, we saw in the previous step that the condition of the chute is a function of its thickness. The wear plate thickness would be then the obvious variable to monitor. However, it might also be valuable to monitor other variables which might enrich our knowledge of the system. In our case we might want to monitor the qualities of the ore (such as its abrasiveness)or the amount of ore moving through the chute.
This acquired knowledge, which will be very specific to your plant and operational circumstances, will give you the additional capability to not only assess the current condition but also predict the wear rate and therefore future states of the asset condition over time. This would start your journey into Condition Based Maintenance, but this is a topic for another article.
An important point here is not to confuse the variables that make a condition monitoring system with the measurements that you need as part of the automation, protection and control functions of your plant. The parameters measured for automation, protection and control purposes can be incorporated into a Condition Monitoring program but not all of them have to be used. Don’t feel compelled to collect and store a piece of data just because it’s there, available. Some measurements and variables have a specific function within the context of a particular system, but that doesn’t necessarily mean that they have to be used as, or even that they are, good condition monitoring inputs.
This step will add the next layer of detail to your system, which will incorporate all of the associated input variables of the models. With this in hand you are ready to jump on to the next part, selecting monitoring equipment and establishing a collection and storage system.
3. Select monitoring equipment and establish a data collection and storage system
If you have followed through the previous two steps, by now you should have:
- A list of assets to be monitored,
- The models that describes the condition of those assets and,
- A set of variables that need to be fed to the models to produce an assessment.
With this information in hand you’re now ready to go shopping. The variety of vendors and functionality of the devices they sell can be overwhelming. Also, you should keep in mind that not everything would or should be monitored by a remote device, in some instances it might be justifiable, and even advisable, that the some of the monitoring functions are performed by people rather than machines. Having said this, we must acknowledge that the economic and safety benefits and thus the trend towards automation are evident.
Although it would be naïve for me to try review each possibility in this article, we can certainly review a number of pointers to keep in mind when sourcing the monitoring equipment:
- Functionality. If you have followed the steps so far you are in a great position to objectively review and compare monitoring equipment options. You know what variables you’d like to extract from the equipment and you know exactly what functions the monitoring equipment should have to fulfil your needs. This would prevent you from “over-specifying” this equipment as well as to not get distracted by the myriad of options that you will be offered by the equipment vendors.
- Obsolescence. In the context of monitoring equipment, obsolescence plays an important role in as much that the monitoring equipment has to be available for the life of the asset that is being monitored. In terms of obsolescence you should ask questions like: what technology is this monitoring equipment based on?, does the equipment manufacturer have any plans to release a new line of equipment within the life of the asset, and if so, is this equipment going to be “backwards” compatible?. The situation that you are trying to avoid in the future is for you to have to either pay dearly for spare parts that are considered obsolete or even worse, have to replace your monitoring system with a new one, having to incur in retrofitting costs with the added headaches of adapting the new signals or communications standards to your existing Distributed Control System (DCS). Ideally, once you make a choice of monitoring equipment, it should stay functional for the life of the asset that it’s monitoring.
- Training. Another factor to consider, although it might not be a deal-breaker, is to whether the vendor offers adequate training on the operation and maintenance of the equipment. Realising the value that you are supposed to get out of this equipment depends in some measure on the capacity of your personnel to properly operate it and maintain it.
- Portability. Last but not least, there is the question of whether the equipment you choose is permanently fitted to the asset or there are portable versions that you can carry around to measure the variables you require from the asset to determine its condition. In terms of portability you would need to strike a balance between the cost of the monitoring equipment itself, the need for “real time” monitoring or just spot measurements and the cost of having personnel go around the plant taking measurements. In some instances, such as partial discharge (PD) measurement on large electric motors, it might pay to use a hybrid system, where probes are permanently fitted to the motor but the PD reading instrument is connected to the probes at regular intervals determined by Reliability Centered Maintenance (RCM) principles.
All in all, you should try to find a balance between the functional requirements you have for this equipment, the cost of implementing such as system and the reliability and safety benefits gained by its implementation.
4. Define and follow a consistent condition assessment and health ranking system
All the effort you have done so far would not amount to much if all the collected data is not transformed into actionable asset management tasks.
This is where you will put all that technical expertise within or outside your organisation to good use. It is at this point of the process that a solid and repeatable system to determine the condition of the asset has to be defined and decided.
You will find that different people, including external experts, will have their own catchy commercial name for this part of the system, but at the end it all means the same: transforming data into knowledge. What is really useful for this whole system to work as intended is not the temperature of the bearing, but the best possible estimation of the current condition or time-to-failure of that bearing. The temperature of the bearing is data, the condition or time-to-failure is knowledge.
Furthermore, it is not only important that you have a consistent knowledge generation structure but that it is also sustainable in the long-run, well beyond your retirement date, the ranking system has to keep operating as intended.
Independently of which particular algorithm you select to process the incoming data to transform it into knowledge, it is essential that your decisions and the inner workings of the ranking algorithms are well documented and integrated to your overall document control system. It should allow anyone that is part of the team making this system work, to understand how all that data that is being collected is being turned into actionable asset health and condition knowledge.
Ultimately, this will also allow your organisation to best digest and integrate new asset health ranking algorithms as they become available.
5. Integrate the condition monitoring system in your existing maintenance strategy
How many times have you heard the Condition Monitoring team say “we find the issues, but then Maintenance doesn’t do anything with them and machines end up failing anyway”. This thought is disheartening to everyone in your team as it conveys a sense of wasting efforts in something that doesn’t produce tangible benefits to the organisation. In more common terms, people will feel that they’re “spinning their wheels”.
It wouldn’t be worth having gone all this far to waste the potential benefits of the system you have implemented.
Each step that we’ve talked about has a role to play and value in its own right, but if I had to choose a step that was the most important one, it would be this one, where everything is put into practice.
The overall success of the system hinges around the capacity of your organisation to opportunely acquire asset information turn it into knowledge and act in consequence, preventing early failures and eventually extending the life of the asset. The outcomes of the condition monitoring system have to be an integral part of your maintenance planning and execution activities.
It is my hope that the brief recommendations that I’ve shared in this article series have provided you with a better perspective on how to put together and execute a plan to implement an effective, efficient and sustainable Asset Health and Condition monitoring structure.
By no means was this series was intended to be an in-depth or academic analysis of an asset condition monitoring system but rather I tried to share some practical advice and a logical sequence of the steps that you might follow to implement such as system and hints on common pitfalls to avoid.
The observing readers amongst you might have noticed that the system described in these articles would fit nicely on the “Asset Knowledge Enablers” block of the overall PAS-55 (soon to be ISO-55000) framework, allowing all functions of the business to have a confident idea of the condition of the business’ assets and their capacity to reliably perform their function.
An adequately designed and implemented Asset Condition Monitoring programme will enable your organisation to more accurately and opportunely allocate capital based on sound and reliable asset knowledge, delivering a much needed edge in the competitive world we live in.