Predictive Maintenance in Aviation - What is it & why does it matter?
What is Predictive Maintenance
If we think about predictive maintenance, it is much like a pyramid.
We are at a point, in the aviation industry where we are quite good at descriptive and analytical things such as reliability reporting… what is happening today and what happened in the past? We’re also beginning to better see what will happen if current trends continue but now, we’re trying to look at what is going to happen next. If we know what happened in the past and if we know what is happening now, can we use that information to look further ahead and predict what will happen in the future?
Let’s put all that in the context of aircraft maintenance: knowing what happened in the past, knowing what is happening today, and using that in order to think about what could happen in the future.
Not everyone will agree on this but, in this context, aircraft maintenance is, in essence, not that complicated: it’s either reactive or it’s proactive, that’s all. Reactive is failure based; something breaks down so it has to be fixed. Proactive aircraft maintenance offers two options; it can be preventive maintenance, A-Checks, C-Checks, an overhaul of components, etc. or we can do predictive maintenance. The big difference between these two is that preventive maintenance, the current position of the industry, is very much age-based, i.e. at certain intervals of hours, flight cycles, and/or time, there are checks to be performed and components to be overhauled… With predictive maintenance, the idea is to still perform maintenance but make it condition-based; so, rather than doing maintenance based on a certain fixed number of flight hours, flight cycles, or calendar dates, we determine which condition of aircraft systems and components would justify performing maintenance actions. In short, reactive maintenance is too late as in AOG (aircraft on ground) situations; preventive maintenance means that you’re doing some jobs too early; but predictive maintenance tries to do maintenance right on time: There is a sound philosophy behind doing maintenance based on condition monitoring.
If condition monitoring is that important for predictive maintenance, how can we monitor and measure the condition of aircraft systems and of components installed in the aircraft to ensure that maintenance is carried out on time rather than being reactive and waiting for them to break down?
This is the million-dollar question of our time. The answer is very straightforward: there are two options.
The first option is to put as many sensors in an aircraft as it is possible to do, as these sensors will allow users to measure the condition of systems and components. A temperature sensor will measure the temperatures reached by a component at various stages in the aircraft’s engine’s operation; a pressure sensor measures pressures in hydraulic systems, etc.; so as many sensors as possible should be fitted. This is what is happening with new generation aircraft: A350, B787, B737 MAX… they’re full of sensors. However, it isn’t possible to fit a sensor on everything and not everything can be measured with sensors.
For those aircraft systems and aircraft components where sensors cannot be fitted, option 2 is to use statistical modeling in order to calculate what the condition of these systems and components could be at a given moment in time. If we put a complex mathematical definition against that, it’s determining, in time, what the deterioration percentage ‘X’ will be over time. It uses modeling mathematics in order to calculate how we expect the condition to progress. That is, in essence, what we are doing with predictive maintenance.
Predictive maintenance involves the use of information such as sensor data, and statistical modeling to calculate what the condition of systems and components could be at a given moment in time to predict maintenance needs in advance. It helps airlines to determine when maintenance should be performed.
Why Predictive Maintenance Matters in aviation
Pretty much since the introduction of the Boeing 787, people have been talking about predictive maintenance, data collection in aviation, and suchlike. The reasons that it has become such a hot topic are the arrival of next-generation e-enabled aircraft, reduced costs for data storage increased availability of data through mobile and smart devices with the tools to access this, catching up with and spill-over from the consumer market (social media developments and consumer behavior analysis), the drive to reduce maintenance costs and more efficient use of resources.
Understanding the real value
In short, there are a lot of good reasons why predictive maintenance matters in aviation and, going back to the questions and answers at the top of the article, not least is reducing costs and increasing aircraft up time. So, there is a very valid use case for predictive maintenance and predictive analytics.
The History Interestingly enough
I feel that the real value of predictive maintenance exists in something completely different which relates to our current maintenance philosophies in aviation and the ever-increasing shortage of ground engineers. In order to explain this viewpoint, we need to go back to the introduction of the Boeing 747.
You might find this surprising that I’m now referring to an aircraft type that was introduced in the 1970s. But stay with me. Age-based maintenance is, in essence, a maintenance philosophy. So, when the Boeing 747 was introduced, something else important happened in the aircraft MRO industry. With the 747, MSG Maintenance Programs were introduced, moving away from older to new maintenance philosophies. As a result, certain sets of aircraft components were classified as ‘On-Condition’ components and others remained age-based centric with hard-time requirements and life limits. In addition to this, airlines, under the MSG-3 philosophy, needed to monitor the performance of the ‘on-condition’ parts and the effectiveness of their maintenance. As such, modern-day aircraft reliability management was born. This is actually still very relevant today as most aircraft maintenance programs are based on MSG-3 philosophy with hard-times, life-limited parts, age-based maintenance, and a number of aircraft components where there are no requirements associated with them.
The facts, the real value
In order to monitor the effectiveness of that maintenance program, it is necessary to perform reliability analyses. That’s the one thing that comes with MSG-3: if there are components that don’t have any maintenance requirements against them, it is still at least important to monitor whether maintenance is being carried out effectively. That is why ‘reliability’ was introduced. And those maintenance programs are still in the industry today: so, a significant amount of tasks in an aircraft maintenance program consist of functional checks, component replacements, etc.; all based on age-based maintenance. But what does all this have to do with the real value of predictive maintenance in aviation? A typical maintenance planning document (MPD) would be broken down as in figure 5 with just 15 percent of work related to zonal/structural maintenance, 25 percent to components, and a huge 60 percent related to systems.
Why now?
Technology and computing power available today allow us to develop sensors and reliable statistical solutions for condition-based monitoring. But there’s another material reason why predictive maintenance is becoming increasingly important. That has to do with the forecast, from ICAO, for a global shortage of Aircraft Engineers. This will require us to rethink the way we do aircraft maintenance. If we are still in a situation where we are effectively applying maintenance philosophies, age-based maintenance, that are, essentially, not applicable to those 82 percent of aircraft systems and components, and there is a global shortage of engineers looming, it means people are performing maintenance activities which are not the most effective to maintain that aircraft and there is a shortage of those people anyway. A combination of issues that cry out for improvements in maintenance philosophies. This is what predictive maintenance is about, being able to get to a point where we can do condition-based monitoring on that 82 percent of aircraft systems and components for which we know that maintaining them based on cycles or hours or calendar dates simply doesn’t make sense. We do it today because the MPD says we have to do it, but we know it doesn’t make sense. If that is already the the case today, what might we be able to do to get started on transforming to predictive maintenance
If you look at your own maintenance program today, and run some analyses on it, would you see similar distribution like above chart? Interestingly, for only around 18 percent of those systems and components is the age-based maintenance philosophy the right way to maintain those aircraft systems or components? For the remaining 82 percent, condition-based maintenance is much more effective. (Rio R. 2015) In other words, there is potential to further reduce maintenance costs if we are able to monitor the condition of 82% and only replace/repair when the condition indicates that a particular component needs to be replaced. This shows us that there is quite a high level of potential value that still sits in maintenance programs today, regardless of the aircraft type being operated.
What you need to get Started with Predictive maintenance
We, at EXSYN, see this as three different and distinct phases from the Reportive Airline to the Monitoring Airline to the Data-Driven Airline:
The above diagram reflects the stages of the ability of an airline to analyze what happened in the past and what is happening today; the different steps in what we might call the evolutionary ladder in order to be able to successfully adopt predictive analytical or predictive maintenance technologies. First, it is necessary to look to the past and also know what is happening today, which makes up a Reportive airline. The Monitoring airline goes one step further; they know what happened in the past, they know what is happening today and they use that information to look for trends and what might happen going forward. The third step in adopting predictive maintenance or predictive analytical methodologies is to start looking beyond today. To use information from last year, yesterday; monitor what is happening today, and identify trends with mathematical models and sensors in order to look beyond today and to what might happen with certain components and how their condition can deteriorate over time. This implies that data is quite important: if we’re going to monitor the conditions of components in order to maintain them it’s also important to know which factors will influence that condition: which environmental conditions can influence it, and which system use parameters can impact those conditions. It might be important to know several different things that an airline might not know today in order to successfully adopt predictive maintenance technologies.
It’s important to know about the airline’s aircraft and fleet utilization and everything else that is needed for aircraft reliability monitoring, but it’s also necessary to have access to flight data recorder information. These are what we call ‘operational parameters’ that show how a particular aircraft has been operated or to which conditions it has been subjected. They’re the operational parameters that, in turn, influence the condition of aircraft components. That would require quite an extensive data model but, again, it’s a push towards a new maintenance philosophy. However, it comes with one big challenge and, interestingly enough, with all the different airlines with whom I’ve worked across the world, there is a common denominator that was confirmed in a recent industry survey. For most airlines, if a technical director or head of maintenance and engineering is asked, they’ll tell you that their number one focus is predictive maintenance and that they’re putting a lot of resources, time, and attention into that. Ask what is the biggest issue that respondents face and the largest response is ‘data accuracy, the data is not that good. So, on the one hand, predictive maintenance is the highest priority while, on the other hand, data accuracy is the biggest issue that airlines face. That’s ironic in its own right; trying to use a lot of analytical, condition-based monitoring technologies based on data that users don’t think is wholly reliable which, in turn, might influence the accuracy of some of these analytical models. Of course, businesses will want to know when they might see a return on the investment into predictive maintenance and early evidence suggests that benefits of improved aircraft uptime and reduced maintenance costs, especially for unscheduled maintenance events are the early wins. The bigger benefits come when the airline can look at that 82 percent of maintenance programs and to improve those items which will deliver significant cost savings but that are hard to quantify as the change will be to a completely new situation in terms of aircraft maintenance philosophies.
There are some early win advantages but the main driver for those who are really serious about adopting predictive maintenance technologies is to start with data accuracy today to have early-win advantages, focusing on the accuracy and availability of data. That will be the foundation of being able to do successful predictive analytical modeling to get to the point where maintenance on the aircraft can be done based on the monitored condition of those aircraft and systems. The biggest challenge is not so much data sharing, because that is just a matter of time and the removal of barriers that constrain data sharing in the industry. It will be the change that has to take place in the industry that will bring the greatest challenge. We’re accustomed to doing maintenance too early in the preventive mode and will be moving to condition-monitored maintenance which means we’ll need to rely on and trust the systems and algorithms that monitor the actual condition of aircraft systems and components. That might feel like giving up a bit of security, so we need to be one hundred percent sure that the outputs and conditions monitored are accurate.
Takeaways
There are five important things that I hope you have gained.
Sensors and statistical modeling are the core essence for condition monitoring or for condition-based maintenance.
A looming shortage of aircraft engineers will push MRO / M&E functions to rethink how aircraft maintenance is done.
Technology allows us to only perform maintenance when the condition indicates the necessity to do so. In essence, what we are talking about is a push of the industry towards a completely new maintenance philosophy: a move towards condition-based maintenance.
The data required to be able to properly use such models within an airline might be of questionable quality and is scattered amongst multiple different stakeholders (OEM / Airline / MRO) so, if this is to be done properly by the industry then information needs to be shared with each other: otherwise, any initiative that might be embarked upon in the area of predictive maintenance, will be destined to fail from the outset.
Click on the picture below to view the presentation:
If you do have any questions or would like to receive more details about predictive maintenance and how EXSYN can help you, feel free to contact us: