BSI supports HHLA with AI project for predictive maintenance

Working with the software company BSI, Hamburger Hafen und Logistik AG (HHLA) examined the promises of predictive maintenance for the maintenance of wire ropes on harbor cranes – resulting in remarkable success.

Hamburger Hafen und Logistik AG (HHLA) is a leading European logistics company with a tight network of container terminals in Hamburg, Odessa, and Tallinn. The company’s business model looks to innovative technologies and is committed to sustainability. In the predictive maintenance space, HHLA works with BSI and uses the BSI AI platform.

Learning machines can reduce maintenance costs

With various focal points and characteristics, HHLA uses artificial intelligence (AI) in several projects to test new applications. One application of particular interest from an economic perspective is predictive maintenance, i.e., the generation of reliable forecasts for the service life of plant technology and its expectable deterioration.

Let’s take steel wire ropes as an example. They are subject to heavy daily usage on the HHLA’s container bridges. In 2019, 138 such wire ropes measuring a total length of 113.2 kilometers were replaced at HHLA’s Burchardkai container terminal alone. However, when exactly is the right time to replace or, at least, inspect the wire ropes? Determining the right point in time can reduce the costs associated with maintenance and replacement.

To date, steel wire ropes have been inspected manually at regular intervals, with the collected data being compared against standardized threshold values. The resulting comparison shows whether the steel wire ropes need to be replaced. The timing of the replacement can be inconvenient, however, and might, for example, coincide with the unloading of a ship. This causes additional costs and delays in the operation.

Also, since the service life of wire ropes varies quite a bit, the anticipated number of ropes has to be procured in advance. Then, the ropes have to be warehoused until the replacement date, using up valuable terminal space, and they might even sustain damage before they are installed.

Optimization of the accuracy of AI predictions

HHLA Technik developed a joint project with BSI’s software specialists to more effectively predict the expected service life and thus the optimal replacement date. Suitable machine learning (ML) modules were selected, and the team had access to the operational data for container bridges as well as to rope maintenance data of the past six years to train a neural network. The stated goal was the determination of individual replacement dates two weeks in advance.

ML helps to identify patterns or conditions in existing structures, and neural networks are used as tools when there are complex relationships between many variables in the existing data. Being able to recognize patterns in such complex datasets is one of the strengths of neural networks.

After two days of training by BSI, the HHLA Technik project partners were able to run the neural network by themselves. Only a few small adjustments later, it turned out that the accuracy of the predictions matched the actual service life of the wire ropes very well. Ulf Bockelmann, HHLA Technik Director, considered the attained results “impressive” and stated: “We have to examine more closely how ML can best be used in our field. For example, I can conceive of modifying the maintenance intervals depending on the load parameters. Based on the service life prediction, we will be able to schedule the maintenance measures in a manner that is more operationally compatible going forward. In the best case, we would further narrow down the reasons for increased wear and tear and infer countermeasures from it. We do have a way to go until then, though.” Benjamin Heusser, who supported the project as a machine learning engineer at BSI, was delighted: “The combination of the industry knowledge the HHLA team brought to the table and our ML expertise provided ideal conditions for a successful project.”

How industrial customers can benefit from AI

Potential applications for artificial intelligence for B2B customers abound. Predictive maintenance is not the only benefit of machine learning for industrial customers. They can also achieve compelling results in smart manufacturing, i.e., the improvement of the production process, quality control, or energy management. BSI’s AI platform is easily integrated into existing system landscapes and provides an abundance of smart automation options. BSI is eager to apply the experience it has gained in ML projects with industrial customers to other B2B applications.

High-resolution press images