Process Discovery: initions creates AI and dashboard for Fraport

A data-based forecasting model for growing flight volumes

Fraport faces the challenge of improving and further developing a data-based forecasting model for the growing volume of flights. With the claim "Gute Reise! We make it happen!" Fraport ensures smooth operations at 31 airports on four continents. In doing so, the company focuses on passengers and cargo. Customer satisfaction is the top priority, so Fraport's goal is to further increase efficiency in the loading and unloading of aircraft.

In times of COVID-19 and the subsequent resurgence of air traffic, Fraport had to face new challenges. As the need for personnel continues to grow, and skilled workers are brought on and trained on an ongoing basis, Fraport aims to ensure the best possible allocation of the existing workforce. Planning requires a suitable forecasting model that considers rapidly changing conditions in air traffic and the associated services – while maintaining a consistently high level of quality.

The changing processes and requirements for aircraft loading and the restart of air traffic allowed Fraport to optimise its forecasting capabilities in terms of load staff utilisation. The workload during peak loads is higher because traffic is bundled by the airlines. The available data, such as flight times, flight detail data (position, aircraft type, airline, etc.), as well as passenger and baggage flows, should be open and clearly displayed according to a new, improved forecasting model based on demand.

The existing data-based forecasting model for loading staff utilisation has proven useful but has yet to be able to fully take into account criteria such as individual utilisation of different skill groups or imbalances in staff composition. To enable even more accurate forecasting, initions' data engineers are working on an advanced solution that will allow staff to be deployed as efficiently and optimally as possible on the current traffic day. This new method will enable accurate data analysis to determine how much loading staff is needed at any given time and which positions should be filled.

Process Discovery Workshop with the customer

In discussions with the client, we identified the need for visual data representation as a challenge. Our questions were thus clear. How do we forecast capacity or demand for loading personnel in the near future based on data? Which AI model is suitable for this? How do we provide the data in near real-time, and how do we process it? How do we use data analytics for process optimisation, e.g., to identify bottlenecks?

With the business department and the customer's development team, we achieved efficient process discovery through close feedback loops by sequencing solution proposals and testing. Our data engineering and AI experts were thus able to develop an optimal solution for this issue.

Data Strategy Assessment of the existing data

We overcame the challenge – a need for more viable data representation – by preparing the data and providing a dashboard. The digital solution deployed balances demand with availability and improves the deployment of loading personnel at the airport.

Data engineering and analysis with Databricks

Using the Databricks AI solution, our AI experts achieved improved data preparation. Data collection and analysis enable meaningful forecasting for up to four hours into the future. Fraport thus uses data in a targeted manner to create forecasting models for loading staff utilisation. The solution is used for process optimisation in the distribution of personnel in the loading area. Another advantage of the data preparation is that the database can be used in this form in general and not just for this project. The customer can thus draw further benefits from the data.

Visualisation with Grafana

We visualise the data analysis performed with Databricks with an innovative Grafana dashboard. The graphical representation allows easy data reading and provides a better decision-making basis.

Process Discovery for efficient planning

With a good database and a precise AI, Fraport makes efficient decisions for the aircraft loading and unloading area. The descriptive dashboard makes this possible after only 4.5 months of project runtime.

In further projects with Fraport, we work closely with the developers on the customer side. Among other things, we are using the existing database to predict when aircraft are ready to take off. This time is a central planning parameter calculated using data from many channels. Predicting aircraft take-offs as accurately as possible serves Fraport's customer-experience in the areas of cargo and passengers.

Process optimisation with Databricks and Grafana

Working closely with Fraport, initions' Data Scientists completed the discovery stage and then implemented a successful Data Engineering project using AI tools in just 4.5 months. By collecting and analysing data using Databricks and Grafana's clearly structured dashboard, Fraport can better plan the deployment of loading personnel at the airport. The data collected can also be used for further process optimisation within the Group.