The Next Wave In Airport Efficiency: Predictive Analytics

April 6, 2016

No one knows your airport operations better than you. You make decisions based on your arrivals and departures and the most accurate Terminal Aerodrome Forecasts (TAFs) available. When weather meets congestion you have prescribed actions to deal with the situation.

But, delay propagation impacts both your activities and the airlines you serve, not to mention the passengers you serve. What if there was a way to translate a forecast’s impact on your airport operations up to 12 hours in advance? Predictive airport analytics takes weather forecast data to the next level, by predicting weather and congestion impact on your operations. Staying ahead of a Ground Delay Program (GDP) means avoiding costly diversions for airlines, and keeping your operations running smoothly.

Airports and airlines alike can benefit from more efficient flight operations. Since weather is the leading influence of flight delays, including airport operational impact due to weather in your decision making can improve operational efficiency. From fuel loads to the deicing queue, time and money can be saved by accurately predicting future operational conditions. When you can plan a full 12 hours out using reliable predictive analytics, your entire operational behavior changes in terms of staff, processes and severe weather preparedness.

Solutions, such as WSI Fusion Predictive Airport Analytics by The Weather Company, an IBM Business, make this possible by leveraging big data analytics and machine learning methods to predict airport operational conditions up to 12 hours out from current time up to 5-times sooner than previously possible.

Today, when airlines make your airport the hub of their operations, a delay at your facility can have ripple effects through the carrier’s entire network and cause displacement of aircraft, passengers and crews.

Let’s start by looking at three common airport operational metrics; Airport Congestion, Runway Configuration and Taxi Times.

  • Congestion — Airport congestion is a leading cause of en route and ground delays, especially when it is unanticipated. Airport congestion predictions provide early insight into future congestion, empowering decision makers to take action to reduce the impact on operations.
  • Runway Configuration— Airport runway configuration prediction provides decision makers with early insight into runway layouts and timing of runway configuration changes, which can aid flight route planning for reduced fuel burn and flight time.
  • Taxi Time — Airport taxi time prediction offers future taxi in and taxi out time at airports for specific airlines. Fuel planning and ramp resource management can benefit greatly from such insights.

Predictions of these metrics provide actionable insights for airlines to integrate into existing operational processes to optimize fuel loads and route selections, mitigate delay propagation and prevent potential crew timeout due to congestion.

Simply knowing about anticipated delay events further in advance can diminish their impact and, in some cases, make them totally avoidable. When you couple weather forecasts with demand capacity prediction and flight arrival and departure data, patterns emerge. Using data science to systematically decipher such patterns leads to reliable prediction of critical airport events so that airports and airlines can, consistently, utilize such insights to proactively manage delay events and mitigate potential impact.

Data Science Is the Magic Behind Predictive Analytics 

To predict airport operational conditions, analytics use machine learning models and numerical analysis based on historical behaviors and patterns to project future conditions. For example, if, from historical data, an airport always changes its runway configuration when the wind changes to certain direction, and based on the weather forecast, the wind at the airport will change to the direction after 2 p.m., then the model can predict a runway configuration change after 2 p.m. Runway configuration changes cause both approach patterns and taxi times to change. As a result, airlines, knowing such operational conditions ahead of time, can plan their routes and fuels accordingly to optimize their flights to the airport.

By anticipating the operational conditions at an airports and conduct proactive actions to mitigate subsequent delay propagation, airlines and airports can significantly lower both direct and indirect operational costs and achieve better operational efficiency.

The airport congestion prediction chart for SFO shows both flight demand and predicted airport capacity values for each time period. On the top of the chart, a traffic light color system provides a quick glance into the congestion level at the airport. The green line indicates the congestion level at SFO and the scale is shown to the right of the y-axis. A measure of 10 indicates that the airport is expected to be at its full capacity (100 percent utilization of the capacity). In Figure 1, the analytics predicts that SFO will have departure flow congestion between 19:00 UTC and 22:00 UTC. Because the departure flight demand during this time period exceeds the predicted airport departure capacity (especially between 20:00 UTC and 21:00 UTC), taxi out delays are likely to occur. Airlines might choose to delay flights, hold flights at gate, fuel additional taxi out fuel, or cancel flights to avoid extended taxi times.

Calculating Potential Savings

In the abstract, the benefits of predictive analytics are clear. Let’s dig a little deeper to understand where savings are derived.

Cost savings can be quantified across four categories:

  • Fuel Burn — Excess fuel carried due to uncertain surface operations can drive fuel burn inefficiency.
  • Flight Time — Runway queuing and taxi-out/taxi-in delays due to congestion can add up to significant operational costs when conditions are not anticipated.
  • Inefficiency Propagation — A single delay doesn’t just impact that flight. Late arrival may lead to departure delay for the next flight mission; passengers may miss connections and crew timeout can delay flights downstream.
  • Intangibles — Although difficult to quantify, brand erosion, decaying industry confidence and various other operational costs are incurred when operational efficiency is not optimized.

An airline with 650 daily flights is expected to receive an average of $1.3M annual saving from fuel and flight time using predictive airport analytics.

Evaluating Predictive Solutions

Selecting the right predictive solution is critical in obtaining correct impact estimations. When looking at a predictive solution, it is critical to examine the following four attributes:

  • Superior weather forecast capabilities are a must. The best predictive models will fail if inaccurate weather data is fed into the system. Select a solution that has proven, global weather forecast capabilities.
  • Building an accurate model takes significant skill and robust set of data points. Look for solution providers who have put in the hard work to build sustainable algorithms and machine learning techniques.
  • No operation can afford to rework their entire operational framework. Look for a solution that is already integrated into existing systems.
  • Seek solutions that have taken sufficient time to validate and verify the accuracy of their predictive model. Quality solution provides will be able to explain their testing methodologies.

Three Critical Benefits of Predictive Analytics

  • Reduced Unplanned Fuel Burn — With better taxi time, configuration and airport congestion predictions, carriers are able to burn less fuel during taxiing and carry less contingency fuel during the operations. As a result, carriers can reduce unplanned fuel cost.
  • Reduce Excessive Flight Time — With better taxi time, configuration and airport congestion predictions, carriers are able to avoid unnecessary taxi out delay by taxiing toward the right departure runways. In addition, airlines are able to select the routes aiming toward the predicted arrival direction to avoid vectoring in the terminal airspace. For example, by acquiring better situational awareness at JFK airport, Delta saved an estimated 228 hours in excessive taxi time in just one quarter.
  • Reduce Delay Propagation — Carriers can reduce initial flight delays by taking proactive actions using insights from predictive airport analytics, further mitigating the downstream delay propagation impact.

As adoption of this next wave of efficiency driving technology grows, airlines, airports and passengers alike will benefit. Early insight into congestion, runway configuration changes and taxi time at key airports will improve pro-active communications, optimize ground operations and enhance the flight planning process.

Mark D. Miller is senior vice president and general manager of Decision Support at The Weather Company, an IBM Business. Miller joined The Weather Company in 1998 as a senior product manager for the Weather Pro product line, serving the broadcast and cable markets. He has also held the positions of director of content and new media for and director of aviation products. Since 2008, he has served as the The Weather Company ’s vice president and general manager of the aviation and government businesses. Miller oversees the company’s suite of aviation solutions serving global civil, military, commercial, business, and private aviation markets as well as the company's Energy & Risk division.