The Challenge of Airline Air Day of Operations Recovery (Part 1 of 2)
This is the first in a series of two blogs that discuss the challenges airline companies face when there are disruptions to service and how these issues can be resolved in real time.
When it comes to technology usage, airline companies are quite sophisticated, with solutions for information technology, customer management, finance, procurement, asset management, human resources, and back-office processes.
But there is one area where there is a need for more effective technology – and that’s in day-of-operations, and in particular, in the area of interruptions to service.
Planning for the unplanned
Unplanned events, such as severe weather conditions, unscheduled maintenance, safety risks, or strikes, are unfortunately a cause of frequent disruptions – which are highly inconvenient and often costly. The innovative planning and scheduling tools airlines use now typically are challenged when it comes to solving these kinds of disruptions and therefore, the airlines can’t execute recovery problems as quickly as they would like.
The primary cause of this dilemma is that there is too much data from too many sources – and there are too many legal and regulatory rules and regulations – which makes it difficult to have a real-time single source of truth. Consequently, airlines or air traffic controllers can’t readily identify where aircraft, crew, and passengers are right now in this moment – with or without disruptions. Without that knowledge, determining compliant new schedules during disruptions becomes nearly impossible.
Too much data – and too many rules
For instance, there is complex operational and planning data – such as flight routes, aircraft, crew members, maintenance, and passengers – that is scattered in a variety of resources. In addition, there are compliance, business, and legal requirements governing each of those entities as well. Airlines have attempted to build in-house applications and data models needed to determine optimal rescheduling options. But few if any have succeeded in building a tool sophisticated enough to handle the immense complexity that compliant resolutions require.
Therefore, most airline planners and aircraft controllers must rely on a manual and tedious process that is error-prone, time-consuming, and a compliance risk. While the process is being worked out by hand – which can take hours or even days – flights are delayed or cancelled, causing significant inconvenience for passengers, flight crews, and maintenance teams. Not to mention, these kinds of delays affect an airline’s brand, and they also cause a significant revenue loss.
Real-time visibility and transparency needed
The ripple effect of disruptions and the almost inevitable limited recovery experienced today is a result of a lack of transparency across an airline’s network of data. Each change (whether for a plane, crew member, or passenger) is associated with the extensive compliance regulations and business rules mentioned above. Hence, a manual rescheduling process takes on a ripple effect – each change affects another and another in the complex interwoven web of data.
To accommodate all these nuances, airlines need a tool that incorporates:
- A centralized data hub that collects and integrates the data on flights, aircraft, crew, and passengers with real-time, up-to-the-moment visibility into all data sources
- Scientific algorithms that model all the rules and regulations governing each entity
- A flexible fast engine to process all of the above instantaneously
- Optimized and embedded rules algorithms and processes for producing new compliant schedules
So what would that look like?
When data science and SAP HANA combine forces
The SAP PSG Data Science organization took on the task or resolving disruptions in airline schedules using an operations research approach. They developed functionality that uses embedded rules algorithms and processes running on the SAP HANA database to enumerate new airline schedules for aircraft, crew, and passengers in near real time.
There are three major components to this new functionality:
The collection and integration of all the data, right now, in one place. To rapidly resolve disrupted schedules, an airline needs to see comprehensive data on three pertinent dimensions – planes, crew, and passengers – all at once, as it changes in real time as. This is where the in-memory computing power of an SAP HANA database comes in. This functionality uses SAP HANA as a data hub to integrate data from various systems, including MRO, HR, flight and grounds operational, reservations, airline partnerships, and connecting flights, into a single source of truth.
The modeling of each and every existing rule. Airlines are governed by an enormous amount of regulations, such as Federal Aviation Agency (FAA) mandates for crews in regards to such things as flight-time limitations, layovers, and compulsory training. Typically encased in hundreds of pages of documentation, these rules are not readily available in any computerized system today. Then there are airline-generated business rules for passengers, such as VIP status, rewards programs, and flying preferences. In addition, there are closely monitored dictates for aircraft safety, such as mandatory maintenance schedules, repairs, and inspections. SAP has developed a rules processor with scientific optimization criteria that utilizes all of these rules and regulations and then overlaid them on the data hub to help determine real-time compliant resolutions.
Optimization ideal rescheduling scenarios. With integrated real-time data and a comprehensive data model, the SAP data scientists then created algorithms and an optimization engine to manipulate the pertinent information to arrive at potential solutions. Business intelligence tools are then used to enumerate a sophisticated, optimized rescheduling of aircraft, crew, and passengers within seconds or minutes. The results can be manipulated easily so that other options can be considered as well, with instant visibility into the current situation and continuous alerts, regardless of the underlying data source, affecting scheduling options.
This process allows for a much faster recovery of disruptions but does it work in real life? To see the full power of this functionality, the next blog
(Day of Operations Recovery in Action (Part 2 of 2)) will take a deeper look into how an actual recovery unfolds.
This problem, of having a very large volume of rules needing to be applied to large data sets in a short period of time is a key problem that occurs in many industries - good to seen a solution presented here.