Predictive maintenance is crucial in the aviation industry as it enables airlines to optimize fleet efficiency. With predictive maintenance, aircraft and their components undergo maintenance only when warranted, based on estimations of their rates of degradation, rather than routinely when they may not need it. It represents both cost and time-saving benefits, so it is widely adopted by the sector. Using it, carriers can ensure all aircraft are available during peak times to capitalize on demand and schedule maintenance during quieter times of the year. This also minimizes the risk of aircraft requiring unplanned maintenance and disrupting schedules.
In contrast, preventative maintenance is when events are scheduled to occur on a set date in the maintenance cycle, whether or not it is actually needed. It ensures parts are kept in good condition, but could potentially cost airlines extra both in time and money but also the unnecessary cost of parts being replaced too soon.
Airlines often need to decide whether carrying out predictive or preventative maintenance is the most efficient option, taking into account numerous other factors such as scheduling, aircraft lifecycle, overall fleet movements and MRO availability.
Going one step beyond this, it is important to understand how to best utilize the huge amount of data afforded through either method. For example, with Aerogility, we use AI to help airlines take their abundance of data created from sensors and turn it into complementary, actionable insights to enable forecasting of maintenance needs, elevating their predictive capabilities.
Model-based AI technology
To facilitate this, there is a demand for software that enables fleet planning teams to play out different maintenance scenarios to not only save time and costs, but to also prepare for any eventuality. Using model-based AI technology from companies like Aerogility, for example, enables the creation of a digital twin of an operator’s organization. In turn, this allows for the generation of operational simulations, which complements predictive and preventative maintenance systems.
Customers can use model-based AI to fly their fleet over different timeframes to forecast life consumption on the airframe, powerplant and other major systems. The simulation considers multiple factors and optimizes when a scheduled maintenance or inspection should be carried out. Planners can also introduce modification or upgrade programs and can include plans for unscheduled events.
The outbreak of Covid-19, for example, and the unpredictability of the past two years has fueled the need for more visibility and flexibility in terms of planning for temporary, disruptive changes and highlighted how crucial it is to plan for the unexpected.
When a maintenance event occurs, the simulation brings the aircraft into a hangar and all the required skills, inventory and toolings are utilized and their associated timings and costs are all recorded through model-based AI technology.
Benefits of model-based AI
The benefit of model-based AI to help process data from predictive or preventative maintenance systems, is that the simulation process is more transparent and the outputs more explicable compared to some data-driven AI, which means you can easily test the reliability of your data. This makes it easier for users to understand the results and make a safe, trusted decision – an important element of gaining acceptance for AI in the aerospace industry. It also gives a much broader view of fleet operations.
Using model-based AI to help with data processing for predictive and preventative maintenance can also reduce aircraft downtime and identify potential problems before a failure happens, which is a significant safety benefit.
Data considerations
There are two factors to bear in mind when using predictive or preventative maintenance. Firstly, aircraft fleet managers and maintenance teams already have access to a considerable amount of the data that enables them to do their jobs, but these come with limitations. Spreadsheet tools, which can only work in two dimensions and can be lengthy and complex, are still used extensively to model predictions and produce forecasts. Secondly, there is the issue that past data is not always an accurate predictor for the future of managing a fleet of aircraft.
Model-based AI, like Aerogility’s, provides advanced software technology for business simulations, radically improving forecasting capability. This enables managers to make sense of very complex and dynamic future landscapes – and to help them work through their optimal decisions. It optimizes the key drivers of a fleet’s operational economics; engines, airframe, landing gear and other high cost, life limited parts for maximum aircraft availability.
It can provide a very powerful holistic insight into how much the fleet will cost to maintain, the required hangar capacity and how many skilled resources are needed in order to carry out the defined maintenance programs. Comparing the dynamics and costs of in-house or outsourced alternative simulations may also be appropriate. It is also important to bring in factors such as the timing and cost of modifications and upgrades.
Vast amounts of data
In modern aircraft, flying a fleet every day produces a vast amount of data, and being able to process that data to produce meaningful connections and results in your maintenance plan is an ongoing challenge in the aviation industry. Model-based AI systems, like Aerogility, help make sense of an abundance of data. When you process your data through a transactional system, such as AMOS or TRAX, and data is then seeded to the model-based AI system, and the relevant and meaningful outputs of your simulation are then displayed. However, you will still be able to view your relevant data – a three-year simulation, for example, will include performing all the required maintenance events consuming life, spares, kits, tools and skilled resource. Projected data can then be used to render different plans and schedules and in turn drive key performance indicators supported by charts and graphs. Of course, there will always be uncertainties.
The great thing about model-based AI is that maintenance teams can consider a series of what-if options and understand the behavior of different trade-offs. By optimizing your maintenance planning you can help maximize your availability, drive down whole life costs and improve your yield and productivity.
Forecasting future scenarios
Using this type of technology to help with both predictive and preventative maintenance is excellent for forecasting the future impact of discontinuous or disruptive events.
Covid-19 is a good example of where, very rapidly, previous business assumptions were thrown into disarray. For fleet planners, these short-notice changes have an impact on what was once a relatively stable long-term maintenance plan. Planning teams, which may be reduced in size or working fewer hours, cannot afford the time and effort of manual methods. They are using model-based AI to adapt quickly to changing circumstances by automatically generating new and re-optimized schedules for each new scenario they are hit with.
Over the past two years, the airline industry has faced many uncertainties and difficult decisions have often been required. However, using model-AI software, companies have been able to utilize their most relevant data to aid some of the crucial decisions regarding aircraft.
Looking further ahead to the next ten years, it is like that there will be a convergence process between data-driven and model-based approaches to AI. The predictive insights gained from big data are very powerful, but aviation companies don’t have unlimited maintenance and engineering resources. They need to optimize their options to align with how they plan to respond to a new fleet or asset status.
Phil Cole is the business manager for civil aviation at model-based AI provider Aerogility. On any given day, Cole could be testing a new release, traveling abroad to demonstrate Aerogility’s software to a prospective customer, or helping a customer solve their complex scheduling and planning issues. Cole began his career as a project manager at Keane NTT, an IT services unit of NTT Data Corporation. After gaining eight years’ experience, he joined computer software company, Lost Wax Media, in 2006 and joined Aerogility just over 10 years later.
