By Gary Flood

December 20, 2023

(Image by Jan Vašek from Pixabay )

One of Northern Europe's busiest airports, Denmark’s Copenhagen Airport (CPH), says it has started to achieve significant operational value from Artificial Intelligence (AI).

This is thanks to its new airport wide AIRHART platform, which is fuelled by a real-time data engine that collates inputs from passengers, carriers, and the airport’s many commercial partners.

CPH says AI is already starting to enhance the airport's ability to predict and proactively address delays, disruptions, and other unforeseen events.

The result is what CPH calls an ecosystem that helps all stakeholders - a “Total Airport Management” solution.

AIRHART will eventually be used by 4,000 people at the airport, from not just CPH’s own team but a total of 46 different stakeholders and partners.

That’s ‘eventually’ because not all processes are running on the system yet—only airside operations and “certain elements” of the airport’s security, baggage and passenger handling, crowd control, and information systems.

This new way of working with real-time operations is also seen as core to CPH’s commercial ambition of increasing capacity from its current 30 to 40 million passengers per year.

Changing airport operations

Becoming operational this year, AIRHART’s data-driven airport management replaces 100 background legacy systems into one cohesive and efficient network, says Mehdi Motaghiani.

Motaghiani is the CEO of Smarter Airports - a new joint venture between Copenhagen Airport and a local Danish-headquartered IT consultancy, Netconomy, which operates the system.

He retains a position as a partner in the latter company.

Motaghiani says:

We went into this collaboration because, despite Copenhagen Airport being one of the most efficient airports in the world, it had reached its limits in terms of achieving digitization and optimization.

That was with both the current technology stack it had, but also what it was available in the market. So, we sat down to look at if, by bringing forces together and utilizing our relative strengths, could we fundamentally change the way that airport operations get done?

As a result of the collaboration, CPH (also known as Kastrup) says it can now use a mix of real-time data, algorithms, and AI to integrate of large amounts of data on traffic handling, flight times, check-in, and security.

These AI and data analytics applications combine their ‘knowledge’ across the airport so that ground handlers can configure their own real time dashboards to access information relevant for their tasks and ensure the right action is taken at the right time.

The aim: optimize all processes relating to airport operations for the benefit of passengers, airlines and businesses operating in and around the airport.

He says:

Instead of looking at millions of lines of data and being bombarded with information which could be mostly irrelevant, airport operators want to do management by exception error. They want something intelligent that can predict something is happening, but also predict the best solution for any event that is about to occur.

Another planned use of the data is to gain data to both reduce energy consumption by, for example, adjusting light and heating in real-time for terminals that have less people in them at any one time.

Another project aim: lower CPH’s overall carbon emissions, says Motaghiani.

He says:

We're already doing that by more precisely and more intelligently planning the routes of the aircrafts how they are actually moving on the apron; do they have to come to a standstill and then use inertia to get moving - energy consumption that we can improve by more intelligent planning.

He adds:

To optimize as complex a place as Kastrup, you need to optimize a lot of small bits and bytes in the chain of events going on across multiple stakeholders. So, with this platform, we’re not only replacing systems but creating a new data and process ecosystem, then opening that ecosystem for everyone to utilize and use and work in instead of working in different systems and silos.

This is for all the internal processes and satisfaction for employees and everyone anchored to this ecosystem and this supply chain, but everything ultimately has to result in passenger satisfaction, because you're reducing waiting times, you're reducing delays, and all of that positively impacts passenger experience.

Throwing ML at the baggage handling roadblock

Netconomy and the airport started working on what finally evolved into AIRHART back in 2015.

The consultancy was called in that year when CPH hit maximum luggage handling capacity over the summer.

Instead of simply building more physical handling storage, the Airport decided to see if it could instead optimize existing luggage handling capacity.

The partner modeled this business process and created a proof of concept (PoC) system that used data to better support handlers.

The test was if software could better predict when luggage gets a timestamp, as this is the key to working out the best workflow for handling it.

Importantly, says Motaghiani, the luggage handling application wasn’t just a normal ERP system, but a machine learning model that continually improves delivery by learning from its own past performance.

And soon, the system was making better and more accurate timestamp predictions than experienced airport staff.

In 2016, the PoC became the first of the smart systems that now comprise AIRHART went into production.

The idea of a learning system was carried over into all subsequent system development.

For example, an AI models has been written that predicts the precise taxi time for each aircraft which is similarly continually trained on operational data and historical CPH traffic patterns.

An early design principle that is still core to the project is creating multiple AI agents that have a single purpose - like ensuring that luggage is arriving exactly as promised, while another knows exactly how weather affects aircraft landing.

However, says Motaghiani, CPH didn’t just want an ML system; it also wanted a way to prepare for an unknown future.

That could only be delivered, the partners agreed, via a flexible and composable IT stack with no potential future vendor lock-in.

This has been achieved, he says, by use of a combination of jointly created software and several open source enterprise tools, including Kubernetes and Kafka.

The technology stack is also containerized and deliberately cloud-agnostic, so can run on any cloud.

Next steps for airport AI

Smarter Airports says that in its first year of Total Airport Management, manual processes are being gradually reduced through optimization and the first wave of use of intelligent decision support.

Now, the AI agents are being scaled up to enhance the level of intelligent decision support that everyone in the platform ecosystem is getting, and ultimately the passenger. 

Smarter Airports says it wants to further develop the idea of Total Airport Management and is already in ongoing dialog with several of Europe’s major other airports about future AI-based collaborations.

Motaghiani says:

A passenger doesn't want a self-service app where they can see a lot of different things. They want something intelligent to tell them exactly what they need to know, and only that.