Flight schedules are probably airlines’ most critical factor towards commercial success and for decades flight scheduling analysts have been tasked with the critical but challenging role of making sure they are profitable. Our Strategy team explore the issues in our AI flight scheduling report below.

Simple but sophisticated solutions exist; carriers must overcome resistance to change that prevents them from finding and implementing these solutions.

Our research shows a strong willingness from airlines to adopt best-in-class flight scheduling solutions; this is a significant or top priority for 80% of LCCs we surveyed.

Airlines must also have a strategic vision for how AI- and ML-powered scheduling can be applied in other aspects of their operations. Once the value case for embracing AI scheduling is proven, potential areas for further value creation can be explored.

Flight schedules

  • Arranging flight schedules is probably the most critical factor for determining if an airline thrives or flounders.
  • Increasing aircraft utilisation1 is the number one strategic goal of all the low-cost carrier CEOs with which we have spoken. 
  • Why? The airline industry is highly capital intensive and operates on razor thin profit margins. This is especially true for low-cost carriers (LCCs), for which the margin per available seat kilometre is only around 0.8 Euro cents, compared to 2.6 Euro cents full-service carriers; approximately 70% less. 
  • Abrupt changes in the broader operating environment can quickly make such economics untenable. 
  • This reality came into stark relief during the COVID-19 pandemic of 2020 and 2021, during which 63 airlines failed or restructured, including LCCs such as Norway’s Norwegian Air, the UK’s Flybe and South Africa’s Mango Airlines.2

Source: KPMG analysis of selected basket of low cost fuel airlines, 2019 annual company reports

Use of aircraft

Maximising airplane utilisation by increasing the number of sectors (a flight leg between two points) an aircraft can serve over a given period is achieved through optimising the flight schedule.

Not only does maximising aircraft utilisation drive topline uplift, but it also yields improvements in bottomline performance.

  • Aircraft operating costs: Boeing has calculated how airlines can reduce aircraft related operating costs by increasing airplane utilisation. A 5% reduction in aircraft related operating costs is achievable by increasing aircraft utilisation by 20% when the average flight distance is 500 miles (800 kilometers) or less, which is a typical flight length for low-cost regional and domestic carriers.3
  • Cabin crew costs: Improving aircraft utilisation can directly increase daily flight hours per crew member, and improvements in crew productivity can lead to significant improvements in operating costs. Prior to the pandemic, Lufthansa modelled a 0.20 Euro cents improvement on CASK for its Eurowings subsidiary between 2019-2022 by taking various steps to improve crew productivity, of which one lever was flight schedule optimisation.4

Planning and scheduling

Network planning and scheduling analysts play a critical role in taking care of the mind-boggling task of creating an optimised flight schedule, given the multitude of constraints they must consider.

  • Many low-cost carriers try to eliminate complexity from their operations by adopting tactics such as only using aircraft from a single manufacturer and using point-to-point networks rather than a hub-and-spoke model.
  • However, when it comes to flight scheduling, complexity cannot be fully avoided: analysts need to consider the availability and usage terms of landing and departure slots, consider oscillating seasonal demand, review the probability of late departure at locations, check for the availability of flight crew, ensure the plane is at a certain location for set period for maintenance and inspection, and a myriad of additional variables.
  • Flight scheduling planners have the unenviable job of being tasked with cutting through this complexity and building profitable flight schedules.

Computers and software

Since the 1970s, virtually all airlines’ schedule planning functions have adopted computerised support to help analysts plan networks and set flight schedules. 

  • Aviation has been one of the most technologically innovative industry sectors since the mid-twentieth century. From the early application of mainframes for ticketing, through to pioneering satellite communications, introducing the concept of dynamic pricing for revenue optimisation, and running some of the earliest big data platforms and analytics for frequent flyer programs. 
  • Flight scheduling software was first introduced in the 1970s to help analysts with the job of creating flight schedules. This software receives data inputs from flight scheduling analysts and after (often lengthy) computation, helps suggest a schedule. 
  • Such tools have played a role in helping the global aviation industry to function at the scale it has achieved; it reached just under 40 million flights per year in 2019. 

Software solutions under pressure

However, present flight scheduling software solutions are creaking under the pressure of the contemporary operating environment.

  • The growing scale and complexity of the variables and data-points relevant to optimisation of a flight schedule is leaving many in the industry to question the capabilities of the existing computational tools employed to build schedules. 
  • Scheduling teams complain that existing flight scheduling applications – which are meant to alleviate this pain – have also become complex, bloated, unwieldy and slow. For example, a frequent complaint is that the UI/UX of these applications is typically characterised by clutter and a furious array of options and toggles. 
  • This isn’t helped by the fact that competition in the space is not so intense as only a few vendors serve the market and there is perhaps a lack of incentive for vendors to radically innovate. 

Use of standard business software

Consequently, many scheduling teams revert to filling out tabular data in standard business productivity software like Microsoft Excel and Google Sheets to create and/or adjust schedules. 

Do planners at your airline ever use standard office productivity software at any point during the flight schedule planning process?
e.g. Microsoft Excel, Google Sheets, LibreOffice Calc, etc.5

There are three shortcomings of this way of working:

In terms of basic operational readiness, many airlines will spend months or weeks tweaking a draft schedule before finalising, due to the manual and cumbersome nature of the software packages they use. 

Simply by using the standard industry solutions, building a schedule for 300 to 400 aircraft could take us two to three months of manual work: moving flights around, trying to get the right times – that’s a lot of time wasted.

Network optimisation and Planning Manager
Low-cost carrier

  • Until recently, building clean sheet scheduling (bottom-up creation of new schedules rather than relying on input from previous seasons) simply wasn’t possible due to technical limitations of the software packages. 
  • This still means, for many carriers, that flight schedules are rolled over from one season to the next with little variation. 
  • An unfortunate side-effect of this approach is that flawed schedules can easily become deeply embedded in airlines’ operations 
  • Because of human intervention and overwrites, flight schedules become error laden (Raymond Panko, a University of Hawaii professor, discovered that, in general, 88% of Microsoft Excel spreadsheets have 1% or more errors in their formulas)6 and unduly influenced by schedule planning teams’ unconscious biases. 
  • Consequently, it is not unusual for schedule analysts to lack confidence in the schedules they prepare. 

Data analysis is absolutely key to plan a network and a schedule; but everyone is using the same tools, data and processes, and therefore coming to the same conclusions. This means that presently the difference is made by human intuition, which is fallible.

Senior Vice President, JetBlue

  • In terms of more proactive operational readiness, the available tools and ways of working are too cumbersome for airlines to prepare for various operating scenarios by running alternative schedule simulations in advance. 
  • This means that carriers are less able to react with agility to sudden demand and operating environment shocks – e.g., the recent COVID-19 upheaval suffered by the industry 

Instead of tinkering with a master version of the flight schedule for weeks or months, we feel we could get more benefit if we could do more scenario planning to really fine tune getting the right flights to the right routes, or, for example, to identify the best hubs at which to base our aircraft.

Network optimisation and Planning Manager
Low-cost carrier

Network and route planners have told me that the COVID-19 period was the most exhilarating period of their careers - but by the same token, they had been working from six in the morning until midnight manually reconfiguring umpteen different versions of the flight plans - which is a recipe for stress, burnout, and mistakes.

John Strickland
Veteran independent aviation industry analyst

Is better technology the answer?

Can airlines maximise value creation in their schedules with better technology? 

Airline analysts are doing the best job possible with the tools at their disposal; it’s just the tools at their disposal were not built for the demands of modern airlines; particularly the lean, ultra-efficient, route rich LCC model. 

The job has gotten a lot more complicated and the skies above Europe have gotten so congested that, when it comes to creating and fine-tuning our schedules, we can’t keep going the way we used to. The old systems and ways of working don’t suit our needs anymore.

Network optimisation and Planning Manager
European low-cost carrier

We therefore ask, can technology offer a better solution? 

Artificial intelligence and machine learning

Airlines can maximise aircraft utilisation by using artificial intelligence (AI) and machine learning (ML) powered tools to automate and optimise their schedules.

Airlines should adopt AI and ML to optimise schedules

  • So, we’ve established that the scheduling products on the market today lack ease-of-use, automation, and optimisation. They are largely manual and cumbersome to work with. Conversely, airlines and schedulers are required to be more agile (the old days of spending weeks or months developing a suboptimal schedule are over), they need to work through more complex operating permutations, and they need to remove human biases and intuition from the equation. 
  • The best way to achieve this is by truly embracing emerging technology in the advanced fields of artificial intelligence such as machine learning, deep learning, and neutral networks.
    “Using data from different sources is key but it can only be truly actioned through advanced analytics that move beyond simple code and into techniques such as advanced algorithms, deep learning and complex neural networks.” 
    Director, Digital Business, European low cost carrier
    “Many existing schedule optimisers claim they use artificial intelligence, but they just use standard computer algorithms that really have nothing to do with AI – but I do believe that we are on the precipice of the true application of AI in scheduling and that it will bring about radical improvements”
    Senior Vice President, Qatar Airways 
  • Artificial intelligence is not new in aviation – it’s not even new in airline operations. For example, the industry already uses AI to estimate potential demand for a route that is not currently served by crunching huge volumes of search and social media data. 
  • However, advances in artificial intelligence are only now beginning to be applied in flight scheduling, and the significance of this development is only beginning to be understood within the industry. 

Do you have a good understanding of the benefits of applying AI (artificial intelligence) and ML (machine learning) in schedule optimization?

Results of adopting new tech

Adoption of AI and ML by airlines would result in: 

  1. Revenue increase, by achieving additional flight sectors; 
  2. CASK decrease, by improving operational efficiencies such as crew scheduling; 
  3. Airlines being able to free-up scheduling analysts to do more cerebral scenario planning and war-gaming leading to increased confidence in the quality and agility of network design. 
  • Of course, AI flight scheduling will offer most potential upside to low-cost carriers. Full-service carriers operate under a different set of operating parameters and demand drivers. They may need to respect banked slots to ensure connections for long haul travelers or ensure certain departure and arrival windows to respect the working hours of business travelers. 
  • Nevertheless, AI in flight scheduling should not be dismissed by full-service carriers. It may still unlock incremental improvements in their flight schedules. And it can jumpstart conversations in these carriers around other operational functions and activities where AI could be applied.
    “For low-cost carriers taking opportunistic approaches to network planning each year, AI will undoubtedly have a significant impact. For fullservice carriers, there will be benefit of course, because computers will always be able to find ideas that humans can’t conceive. But the benefit may be more marginal due to embedded operating constraints and historical norms.”
    Senior Vice President, Qatar Airways 

Ryanair case study

Using Optifly’s optimisation algorithms to optimise their schedule

Context: At the beginning of the pandemic, Ryanair sought new approaches to help it achieve its aggressive growth targets of carrying 225m passengers by 2026. 

Challenge: A crucial part of this ambitious goal was to use technology to maximise the productivity of its fleet. However, the typical scheduling processes had been taking the company many months to settle, even with the help of (non-AI) software. With 500 aircraft, 90 bases, 225 airports, 36 countries and up to 3,000 daily flights, incorporating many operational constraints and commercial objectives, building a commercial and consumer-appropriate schedule had become a near-impossible task. 

Solution: Ryanair selected to partner with Optifly, a next-generation AI driven software, in order to increase the number of sectors that their aircraft fly each day. 

Optifly offers a cloud-hosted web-application that works in conjunction with Ryanair’s existing scheduling software. Users can create clean-sheet schedules, whilst having the ability to turn up and down dials on numerous different constraints to help build both a highly utilised schedule which is operationally reliable and maintenance compliant. 

Ryanair initially rolled out the software across its Italian network of 90 aircraft (18% of their network) for summer 2022. 

The current approach of building schedules twice a year is dead. We’re going to move to a much more optimised, agile, AI-driven model that allows airlines to rapidly respond to developing trends in consumer demand and build schedules and route networks that address that demand as quickly as possible. The two key benefits that Optifly provides its airline partners are agility, so the speed, control and flexibility with which to try different scenarios and build schedules very quickly. And the second one is the productivity increase that allows an airline to get more out of its fixed assets, ensuring its aircraft operate more flights every day.

Michael Healy
Director of Network Optimisation, Ryanair

Results

Owing to Optifly’s ability to build schedules at speed, Ryanair’s planning team has been able to run multiple scenarios throughout each working day and ensure that each schedule is meeting its operational and commercial objectives. Ultimately, the team managed to increase seat capacity by 38% for summer 2022 vs summer 2019. This is 3x the seat capacity increase achieved by the airline for the season across the broader network. 

On the back of this successful outcome, Optifly has now been rolled out across Ryanair’s entire European fleet for its winter 2022 schedule. 

Source: SRS Analyzer, Cirium

With the track record that Optifly has of successfully building scheduling algorithms across other industries, with the improvements and increase in sectors that we’ve seen across our Italian bases for our Summer ‘22 Season and the continued performance improvements we’ve seen throughout our Winter ‘22 schedule, we’re extremely positive and excited about the potential of this product running across our whole network.

Ryan Dooney
Senior Manager Longterm Planning, Ryanair

Although the overall value created by the pilot has not been revealed by the partners, Optifly, which also works with Eurowings (Lufthansa Group’s low-cost carrier), reports that it generally achieves a 5-8% increase in utilisation gains across the airlines with which it works. 

Source: SRS Analyzer, Cirium

Planning for the future

What next? Airlines must dismantle barriers to adopting solutions and start planning for future applications.

Simple but sophisticated solutions exist; carriers must overcome resistance to change that prevents them from finding and implementing these solutions

  • Our research shows a strong willingness from airlines to adopt best-in-class flight scheduling solutions; this is a significant or top priority for 80% of LCCs we surveyed. 

  • So, what’s holding them back? Our survey showed that some of the main concerns center around implementation risks, such as the perceived risk involved with migrating from an incumbent scheduling system. 

  • Yet such concerns may be misplaced. The Optifly solution which features in the case study, for example, is a stand-alone, cloud-based application which does not require integration with any internal system; everything is managed from the industry standard IATA SSIM (Standard Schedules Information) file. 
  • However, concerns of this nature are understandable. While the Aviation industry has pioneered some marvellous technology over the years, as a regulated industry, there’s a certain innate hesitancy around adoption of emerging technologies. Rushing and being experimental can have catastrophic implications and bring down the wrath of regulators. Paradoxically however, there is also a massive driver for players to embrace innovation given the hypercompetitive nature of the sector. Successful implementation of new solutions by respected carriers can act as a bell-weather for others to fall in line. 
  • According to one industry veteran “We’re very much a me-too industry; we copy each other left and right. So, if one carrier has great success with something, then other carriers will adopt it.” 
  • Changing demographics within the industry are also a signal that airlines might be shifting their risk tolerance regarding AI driven decision-making systems. 

Following the COVID-19 pandemic lots of airlines here in the US offered early-out programs which were accepted by many of the older industry veterans. Younger people who’ve grown up with technology, are now assuming more management roles and are more likely to embrace technology like artificial intelligence across airline operations compared to the previous leadership.

Senior Vice President
US-based LCC

Building a strategy across the business

To conclude, airlines must also have a strategic vision for how AI and ML powered scheduling can be applied in other aspects of their operations.

Once the value case for embracing AI scheduling is proven, potential areas for further value creation can be explored. 

  • Slot trading: One extended use case relates to the slot allocation process. The International Air Transport Association (IATA) coordinated twice-yearly slot conferences allow airlines and other parties to meet face-to-face to review their slot portfolio. Here airlines frequently barter and trade with other airlines to exchange slots. Airlines mockup routes before they attend, and AI scheduling will certainly help expedite and optimise that process. But as Ben Leon points out, there is a real-time on-the-ground application also,
    “Because of the near-instantaneous output of an AI scheduler, if an unexpected opportunity arises while an airline is attending a slot conference, it could quickly run a scenario to see if it would be worth going for.”
    Ben Leon, Chief Commercial Officer, Airline Management Group 
  • Other use cases: Airlines could eventually use AI driven schedule optimisation to manage fuel consumption, maintenance and ground handling schedule optimisation. 
  • Irregular operations: Inclement weather, strikes and short-term schedule changes are a pain for everyone involved from passengers to airlines – but meeting these challenges head-on and dynamically rescheduling using AI powered flight schedule optimisation, perhaps for a sub-set of the network, could help minimise delays and reduce unexpected costs.
    “An algorithm can help ensure that when an aircraft arrives to a hangar city for a maintenance check there is a connecting aircraft on the ground at the same time so we can swap the crews over and let engineers get their checks done.”
    Network optimisation and Planning Manager, European low-cost carrier

Footnotes

  1. Measure of aircraft productivity. Typically presented in block hours per day. A block-hour is the time from the moment the aircraft door closes at departure of a revenue flight until the moment the aircraft door opens at the arrival gate following its landing.
  2. Dr. Stuart Hatcher, “IBA records 63 airlines failures and restructurings 2020-2021”, IBA, January 20 2022
  3. Mansoor Mirza, “Economic Impact of Airplane Turn-Times”, Boeing Economic and Financial Analysis Group
  4. Lufthansa, “Capital Markets Day 2019 – Slide 61”, June 24 2019
  5. Our survey of senior network and flight planning executives of ten separate low-cost carriers was fielded in June 2022. These carriers represent a combined fleet of over 1,800 aircraft (June 2022), annual revenues that exceeded USD35 billion in 2019, and operate in the following regions: North America, Europe and Developed Asia-Pacific.
  6. Raymond R. Panko, “What we know about spreadsheet errors.”, Journal of End User Computing, Vol. 10, No. 2, Spring 1998

Get in touch

The pace of change is challenging leaders like never before. To find out more about how KPMG perspectives and fresh thinking can help you focus on what’s next for your business or organisation, please get in touch with Chris Brown, Head of Strategy. We’d be delighted to hear from you. 

More in Aviation