When a roll container loaded with milk cans leaves the conveyor system every thirty seconds in Aylesbury, UK, and has to find its place in the 9,000-square-meter cold storage facility within a very short time, it shows how precisely Arla Foods' internal logistics are coordinated today. The plant is one of the largest fresh milk dairies in Europe. More than 1.6 billion liters of milk are processed, packaged, and distributed here every year. Such a tightly scheduled material flow can only be managed with highly reliable transport robotics. Arla has been relying on an automated guided vehicle system (AGV) from ek robotics since 2014. The Hamburg-based company is one of the most experienced providers in the industry and has been developing customized automation solutions for more than 60 years.

In addition to manufacturing mobile robots, the company focuses on software development, which coordinates all transport and storage processes as a central unit and is tailored to the individual requirements of customers. At Arla, 90 custom-built transport robots have been reliably moving roll containers between production, cold storage, and truck loading for over a decade. During this time, more than 15 billion liters of milk and over 44 million transport orders have been carried out without personnel and fully automated with an efficiency that is crucial in the food industry.

When ERP reaches its limits and the control system takes over

After ten years of continuous operation, Arla took the next step in January 2024: control of intralogistics in the cold storage facility was completely transferred from the previous ERP (Enterprise Resource Planning) system to the AGV control system from ek robotics. “Although a control system is always a central component of an automated guided vehicle system and takes over functions such as traffic control, order distribution, and loading strategies, the situation at Arla was particularly diverse,” recalls Ronald Kretschmer, Chief Sales Officer at ek robotics.

Controlling the complex material flow in the cold storage warehouse pushed the existing goods management system to its performance limits. In order to avoid capacity bottlenecks and reliably map the growing process complexity, it was decided to also integrate warehouse management into the control system from ek robotics. As a result, it is no longer the ERP that bears operational responsibility for the individual transport processes, but the control system from Hamburg. For Arla, this means a noticeable reduction in system limits: The ERP system now only recognizes the cold storage facility as a “black box”; all internal processes, from storage to truck provision, are handled by the control system.

“ERP systems are optimized for production planning and master data management, but quickly reach their limits when it comes to highly dynamic transport processes,” explains Julian Schwarzat, Team Lead System Development at ek robotics. "Our Transport Manager is designed precisely for this task: it is not only responsible for the scheduling and load management of the vehicles, but also for the entire organization of the block storage areas and the connection of peripheral devices such as gates, elevators, chargers, and scanners.

This creates an overall system that bundles transport and storage logic in a single control instance and controls material flows much more efficiently." Another advantage is the modular architecture of the system. New functionalities—such as the connection of additional storage areas or the integration of further peripheral devices—can be implemented without downtime. This flexibility makes the control system interesting not only for dairies, but also for industries such as the pharmaceutical or beverage industry, where similar requirements exist in terms of timing, cold chains, and traceability.

New milk every second

Using various conveyor technologies, the roll containers loaded with milk canisters leave the production area at intervals of between 15 and 30 seconds. Once they arrive at the cold store, robot arms distribute them to the collection points, where the transport robots take over. At this point, the control system already knows which of the approximately 80 different milk products is involved, what the best-before date is, and when it is needed. Based on this information, a decision is made as to whether the milk is taken directly to a truck gate, temporarily stored in a buffer area, or positioned in a block storage aisle. The software ensures that identical types are not stored exclusively in one block.

Otherwise, too many vehicles would head for the same area when an order was placed, resulting in congestion. Instead, the system intelligently distributes the products across different segments. During peak times, the control system allows flexible temporary storage at any location, while during quieter periods, the roll containers are redistributed in a targeted manner. This data-based control also creates a high level of transparency regarding stocks and shelf life. The control system provides real-time information to adjacent systems, thus also supporting traceability and quality management. This aspect represents a decisive added value for the food industry.

Material flow simulation as a dress rehearsal in the digital twin

From a technical perspective, the interface between the previous merchandise management system and the AGV control system remained unchanged; only the host servers were replaced. Arla set up its own simulation server in advance, which was used to conduct intensive test series of the transport processes over several days. Borderline cases were identified, evaluated, and incorporated into the process logic.

“Simulations allow many borderline cases to be tested, but around ten percent of all special cases can only be identified in live operation. That's why we closely monitored the system on site and gradually optimized it further,” recalls Julian Schwarzat. Since the dairy operates 24/7, only a two-hour window was available for the actual migration. Despite these tight constraints, the changeover was completed without any production stoppages. Today, the digital twin is used not only to validate updates, but also as a planning tool for future expansions. New layout variants or additional vehicle fleets can be tested virtually before physical investments are made. This turns the control system into a tool that supports continuous improvement.

Operation and reliability

The AGV operates under challenging climatic conditions: at temperatures of three to five degrees Celsius, the floor regularly becomes slippery with moisture. These conditions pose challenges for the vehicles: in some places, they lose traction and malfunction. To ensure safe operation nonetheless, employees can define zones in the layout, reduce speeds, or block off areas during cleaning. A total of around seven people look after the system in shifts around the clock. During planned ERP downtimes, the behavior of the system in the event of power or server failures is tested, as is how feedback to the warehouse management system is processed and how manual orders must be entered during the transition period. These scenarios show that milk logistics have absolute priority: the robotic arm that removes the roll containers from the conveyor system can only buffer four units. Therefore, even in the event of system failures, the AGV must ensure that the cold chain is not interrupted.

Intelligent traffic management makes the difference

The changeover led to a demonstrable increase in transport performance of around 25 percent. This optimization is the result of significantly improved traffic management. Previously, bottlenecks occurred regularly because vehicles were heading for the same aisles at the same time. Intelligent distribution of the fleet has enabled more efficient use of the block storage area. Especially at utilization rates of over 90 percent, precise control is crucial to avoid congestion while still ensuring that aisles are sorted by type. Thanks to lithium-ion batteries, charging planning remains flexible: during peak times, almost all vehicles are ready for use, while during quieter periods, several robots can recharge in parallel. Just five to seven minutes of charging time is sufficient for continued operation, and the batteries are fully charged after 15 minutes. Here, too, the importance of a software-supported overall strategy is evident. Charging cycles, traffic routes, and priorities are closely interlinked. This not only increases productivity but also improves energy efficiency—a point that is becoming increasingly important in terms of sustainability and carbon footprint.

An AGV project with a signal effect

With 90 vehicles, the Aylesbury facility is one of the largest AGV installations in the world. Expansions would be technically feasible, but Arla's experience shows that the actual performance of an AGV system does not lie in the sheer number of vehicles, but in the quality of the control system. “Hardware moves loads, but the intelligence of the software determines the efficiency of the overall system,” summarizes Ronald Kretschmer. The long-standing collaboration between Arla and ek robotics is now considered a prime example of the cooperative development of logistics processes. Regular testing, close coordination between IT, production, and logistics management, and the ability to adjust configurations even during operation have created a system that has become an integral part of the dairy.

For the industry as a whole, the project is proof that intralogistics software is increasingly complementing or even replacing the role of traditional ERP systems. Many companies face similar challenges, and the results achieved in Aylesbury show the potential of consistently software-driven intralogistics. Arla is considering equipping other locations with this technology in the future. One thing is already clear: the switch from ERP-controlled to control system-based intralogistics has not only increased efficiency, but also laid the foundation for future automation steps.