A new Approach to Asset Maintenance
by Frederick Mannings, Data Engineering Lead
In today’s hyper-competitive and technologically advanced business landscape, the maintenance strategy you employ is more than just a piece of the operational puzzle – it’s a cornerstone of efficiency, profitability, and longevity.
There is often no one maintenance strategy that fits all scenarios. The best strategy for your operations may be a balance between reactive, preventive, and predictive maintenance. Understanding and applying these strategies effectively can mean the difference between a thriving and resilient business and one that’s perpetually trying to catch up.
Unlocking the perfect mix of Reactive, Preventive and Predictive maintenance will turn unexpected breakdowns into planned interventions, and unforeseen costs into calculated investments. This blog post will delve into how to find that magic balance and why it’s a crucial activity for any business in terms of the bottom line, and how the optimal service strategy for your company’s situation may not be what you think.
Know How your Assets Fail
Understanding the intricacies of how your assets fail isn’t just good practice; it’s a vital factor in crafting a robust maintenance strategy. A cogent awareness of failure patterns and modes can equip businesses to pre-empt and manage issues, instead of being blindsided by unexpected breakdowns. It can also provide confidence in the minimal necessary force required to provide operational excellence to customers – a core requirement to any strong service pipeline.
When dealing with failure modes in the real world, there is rarely just one. Assets can fail due to a myriad of reasons ranging from wear and tear, misoperation, environmental stressors, to unforeseen catastrophic events. Each failure mode has its unique tell-tale signs, and being familiar with these can aid in recognizing and addressing issues before they escalate. However, one aspect of asset failure that is often overlooked and is critical to the successful design of a service strategy, is simply the distribution of failures – how often failures occur in any given time.
To start identifying what mix of Reactive, Preventive and Predictive your company needs to apply, it’s crucial to understand this distribution of failures in conjunction with the ways your asset can fail.
The core reason for why understanding how your assets fail (i.e. the distribution of failures) is critical, is because of the relationship between the gross cost associated with servicing an asset vs the revenue generation capabilities of the asset over life. For example, two failure modes may have the same MTBF (Mean Time Between Failures), but could have a drastically different spread in failure distribution. Simply knowing the MTBF is not enough to design a service strategy as the same MTBF may describe both failures modes that occur more often in early life, and failures that occur equally over earlier and later life. But clearly, the gross cost associated with servicing such assets for these two assets would be drastically different over a fixed time window.
It’s critical to get a picture of the failure distribution. For example, here are two frequency distributions with the same MTBF; one describing failures normally distributed through time, and one describing failures distributed through time by a constant rate (exponential distribution that characterises a Poisson Point Process).
Here the MTBF has been set to 300 cycles for both processes. with this additional info of nominal scale for the normal distribution. Both time between failure distributions can be characterised by the same MTBF. And clearly, both distributions of failure are very different and would demand different service strategies, particularly if the gross cost associated with service varies significantly with asset life, and the time which over which cost forecasting is performed is less than the 99th percentile of ‘Time Between Failures’ for both failure processes.
This direct impact of failure mode on service liability and therefore net revenue generation cannot be overstated. Unplanned downtime due to asset failure occurring soon after service (which is a more prominent feature of the exponential distribution) can lead to production halts, customer dissatisfaction and ultimately, financial losses. Knowing how your assets fail really is the first crucial step to optimising your service strategy.
Failure Modes
As previously mentioned, assets fail in a broad range of ways. An asset may not outright fail, but may enter into a degraded state that causes frustration to your customer with decreased revenue generation capabilities. When your asset fails, how often does it degrade rather than completely cease to function? And does it continue to perform revenue generating activities in this degraded state? This is where knowing the way in which your asset fails can play a crucial role in understanding how you should service them when in different failure modes.
If you have an understanding of how your assets failure mode impacts revenue generation, customer satisfaction and service cost/time, you’re in a strong position to start the process of optimise your service strategy over your suite of assets.
Understand your Service Pipeline
The nature of your servicing pipeline plays a significant role in determining the viability of each maintenance strategy. Having a coherent awareness of your service lead time, service time and service cost enables your business to assess the liability associated with each mode of failure.
Additionally, understanding the states that your assets go through during a service cycle will set your business up for success when optimising your service strategy. Each fleet of assets behaves differently, and has a different relationship to the action of failure or service. For example, we may summarise the state that a system is in into the following:
- Operating
- Degraded
- Non-Functioning
- ServiceRequired
- Servicing
Once a fail has occurred and your asset has left the ‘Operating’ state and entered either Degraded, Failed or Requires Service, how long does it stay there before being serviced? And does it continue to generate revenue / provide value to the customer? Or, say, once in ‘Requires Service’ can the asset spontaneously fail and thus require a full re-work?
These are crucial questions that should be understood to form a clear picture of the cycles that an asset goes through when being serviced. Additionally, understanding what actions transition your asset through the maintenance cycle will lay the foundation for designing a winning service strategy.
Here, we see a state transition diagram for an example asset. Our asset can only transition into a Degraded or NonFunctioning state from the Operating state. In a similar manner, we can request servicing directly from the Operating states, like with a Preventive Maintenance strategy. Running this exercise of mapping out the states of your system will help gain a deeper understanding of how the asset behaves throughout the maintenance cycle as well set the stage for optimising for service strategy over your population of assets.
Understanding the dynamics of your typical maintenance cycle is pivotal for tailoring a robust maintenance strategy. Grasping the nuances of your lead time, service time, and service cost equips your organisation to effectively gauge the risks associated with each potential failure mode and build the capability to virtually trial service strategies and evaluate, at zero risk, the liability to your business.
Build your Optimal Service Strategy
We’ve laid the groundwork required for optimising your businesses service strategy. Having an acute awareness of how your assets fail and the intricacies of the maintenance cycle that is required to service them, will put you in good heading to optimise over service strategies. In this section we’ll discuss how.
Simulating Reality
Once the asset state model is in place, the next step is to simulate it over the spectrum of service strategies. This is typically done in two steps:
- Embed quantitative information about asset states and transitions into the state model
- Simulate the model over the domain of service strategies, many times to get a clear picture of all possible realities
Although these steps are well defined, they can also be the most challenging to execute. Nonetheless, their power should not be underestimated, as different asset characteristics and models can demand drastically diverse service strategies to minimize cost. We will see by running the simulations, that choosing an incorrect strategy for your business can have a significant negative impact on your profit margin.
It’s crucial to assign values associated with each state and transition in your state model. These values can be monetary, such as service costs, or revenue, or they can reflect less tangible aspects such as reputation or customer satisfaction. Attaching these values helps to quantify the implications of each state and transition, informing the overall strategy optimisation.
Finally, it’s important to remember that there are costs associated with each type of maintenance: reactive, preventive, and predictive. Reactive maintenance, dealing with failures as they occur, can lead to higher immediate costs and potential lost revenue. Preventive maintenance, scheduling regular inspections and servicing, incurs consistent costs but helps to prevent more serious issues. Predictive maintenance, utilizing advanced techniques to predict failures before they occur, involves an upfront investment in technology but can prevent unexpected downtime and associated costs.
Applying these costs in the simulation agent and running this many times to simulate many potential realities will insure that you approach a service strategy that is robust and does not over expose the business to excess unforeseen cost. Without doubt, this strategic approach to maintenance planning is an investment in the longevity and success of your business.
So let’s put this approach to the test. In the following sections we have mocked up 2 scenarios where we utilise the tools we have developed at Predixus to estimate net revenue for a fleet of assets that have very different failure and service characteristics:
- One reflecting a situation where failures are common and cheap to repair, but service availability is limited
- And another scenario where failures are less common, expensive, but the service is there when you need it
Scenario 1
In this scenario we have the following characteristics of our asset and service strategy:
- Failures are common
- Repairs are cheap and quick to perform
- However, it takes a while for service to become available
- We have a large fleet of assets
- When assets fail, they stop working completely untill repaired
As the assets become non-functional when they fail we assume that their revenue generation capabilites become 0 at failure. The numbers used to produce the simulation are as follows:
- MTBF: 80 cycles (exponentially distributed)
- Service Lead Time: 1 cycle
- Service Time: 10 cycles
- Number of Assets: 30
- Probability of Failure over Degradation: 100%
- Number of cycles simulated over: 500
- Nominal revenue generation per state values
Configuring the asset state machines with the above configuration in <product brand> and running them to produce 100 possible realities per maintenance strategy gives the following result:
Interestingly, preventive maintenance does not provide a large competitive advantage over reactive for this scenario. In fact the overlap between the two distributions means that there are some scenarios where reactively maintaining the systems would give greater net revenue than preventively maintaining!
Nonetheless, predictive is the clear winner in this scenario! Note also, how predictive maintenance significantly shrinks the spread in net revenue, characterised by the short tails in the distribution. This is a clear demonstration of how predictive maintenance can help businesses manage financial risk associated with service strategies.
Scenario 2
Here we have a slightly different scenario:
- Failures are semi-common
- When an asset fails, it will continue to operate in a reduced capacity, producing less revenue than when in an optimal state
- Repairs are immediately available and cheap
- Service is not trivial, and it takes a while to complete
- This is a small fleet
These are the numbers associated with this simulation:
- MTBF: 80 cycles (exponentially distributed)
- Service Lead Time: 1 cycle
- Service Time: 10 cycles
- Number of Assets: 30
- Probability of Failure over Degradation: 0%
- Number of cycles simulated over: 500
- Nominal revenue generation per state values
The resultant failure distributions are counter intuitive at first:
A predictive maintenance strategy is on par with a reactive maintenance strategy, with preventive significantly under performing. The reason why make sense after some thought. As these assets continue to be able to perform revenue generating activities once failed, it’s counter productive to service too soon. Additionally, the short service lead time worsens this effect.
In this scenario, predictive maintenance would only show an improvement over reactive, if it mitigated customer dissatisfaction that results from a spontaneous degradation. In any case, this scenario highlights the importance of performing such an analysis on service strategy as it shows there is not a hierarachical nature to the three service paradigms.
Unlock Your Optimal Strategy
Building an effective maintenance strategy is more than a complex juggling act—it’s a nuanced science that requires an intricate understanding of your assets, careful attention to service logistics, and a quantitative approach that perfectly balances reactive, preventive, and predictive maintenance. At Predixus, this is where specialise – in using state of the art tools to understand the conditions that significantly enhance net revenue, reduce service cost, and improve customer satisfaction.
In the end, creating a robust maintenance strategy isn’t merely about surviving in a competitive business environment; it’s about flourishing and setting new standards of success for your business. Supported by our simulation tools, businesses are well-equipped to not just navigate their maintenance challenges but also to convert them into strategic advantages that ensure operational longevity and prosperity. Truly, this is the power of leveraging asset management in the age of advanced data and technology.