With so many airlines in financial trouble, a common thread seems to be emerging among the rising stars of the industry. Upstarts like America West, Southwest Airlines and Jet Blue are working to harness technology to shave costs and heighten efficiency.
America West Airlines, based in Tempe, Ariz., for example, recently completed the rollout of an analytics system that has helped them forecast their reserve flight crew needs. The business intelligence software by SAS Institute Inc. of Cary, N.C., targeted two aircraft fleets for staffing reductions, while increasing staffing on a third fleet to avoid expensive flight cancellations.
”Within the first year of implementation, we made several changes in our pilot staffing model,” says David Seymour, vice president of operations, control, and planning at America West.
These changes saved the airline $3 million, and helped make it the industry leader in on-time performance.
Airline Business Survival
America West began operations in 1983 with three aircraft and 280 employees. The company has found a niche as a low-fare, full-service business that has blossomed into a 13,000-employee airline that services 95 destinations across the U.S., Mexico, Canada and Costa Rica with more than 900 daily departures.
Like most major airlines, America West schedules and assigns full flight crews for hundreds of daily departures, and uses staffing forecasts to predict employment levels and training needs for the upcoming year. Until recently, the airline followed a standard industry process: Keep a straight percentage of reserve crew members on standby. The reserves act as a pool of on-call pilots and flight attendants ready to cover for any scheduling snag or for crew members who take an unscheduled day off.
This type of system, however, is prone to errors, and the utilization rate for reserve pilots is rarely well tracked.
”Airlines using this method tend to not invest a lot of time and effort into their reserve error rate,” says Seymour. ”Prior to the implementation of our forecasting system, there was no consistent tracking of how these reserves were being used.”
The number used by America West, for example, was based on a percentage of pilots required. Some airlines keep a 20 percent reserve cushion for a very simple reason — it is a lot cheaper to pay a couple of extra pilots than it is to cancel flights. But if you schedule too many pilots, you could lose a big chunk of your profits.
An Analytical Approach
IT executives at America West asked SAS to help them develop a reserve crew forecast system that would help them avoid expensive flight cancellations and save in overstaffing costs. The goal was to schedule the minimum number of reserves required to protect the operation against canceled flights.
”At what point do we stop staffing and take the risk of cancellations versus overstaffing just to be sure that those flights are covered?” asks Seymour. ”It’s very cost-effective to know, within a few heads, at what point you can expect either to pay premium rates or to cancel flights based on your crew levels.”
The problem that analysts at America West and system developers at SAS had to deal with was how to accurately determine how many reserve crews would be needed for each one-hour block throughout the day.
But how exactly do you predict when a plane and its crew will be stuck in Columbus, Ohio during a snowstorm? How can you foretell when a flu bug will hit Phoenix and keep dozens of pilots in their beds for the day?
When America West’s Seymour first started working on this project in 2002, the obstacles to accurate forecasting included a lack of historical data and a technical infrastructure that made current data difficult to retrieve. America West developed forecasting models that pull from a large assortment of data collected throughout the enterprise. The SAS system, known as Base SAS, is hosted on Windows 2000.
”When we started testing, I was just quite shocked at how quickly it validated,” says Seymour. ”We were able to produce a solution with accurate results from the beginning of the project.”
Base SAS is designed for data access, transformation and reporting. It includes a programming language, programs for data manipulation and reporting, and enables organizations to gain a single view of its data. Base SAS can be used for simple descriptive statistics that include mean, sum, variance and standard deviation to more advanced data correlation and cross-correlation, frequency analysis and detailed data distribution analysis.
The SAS database at America West receives data like flight cancellation histories, weather patterns, training schedules, available vacation time, previous reserve crew levels and detailed flight schedules. By incorporating a large number of variables, America West has increased the usefulness of its models.
According to Seymour, the old method was inaccurate and often led to overstaffing. The new approach, in contrast, predicts reserve crew numbers to within one head per month.
The results have been fewer flight cancellations, $3 million in savings, and an airline that leads the industry in on-time performance.
”If we plan accurately and we’re staffed at the right levels, we typically have zero cancellations attributed to crew issues,” says Seymour. ”Currently, America West is the only airline using this method of forecasting to enhance reserve crew planning.”