Flexible yet accurate workforce planning in the logistics warehouse is of particular importance in seasonal peak times. Big data can assist with making reliable forecasts – but workflows and processes can also be made more efficient by targeted evaluation of data.
One of the first learning systems was able to beat a world champion at chess; today we encounter artificial intelligence in autonomous vehicles and voice assistants on the smartphone. All of these rely on “big data”, or enormous volumes of data. Data and learning algorithms have long since also been applied in logistics – in the most diverse areas in the warehouse, for example.
Accurate forecasts for workforce planning
In 2016 there was already 2.8 million square meters of storage space in Hamburg and about 2.5 million square meters in Frankfurt and the Rhine/Ruhr region – and with the rapidly growing online retailing and e-commerce, this trend is rising. Efficient planning of space and personnel is only possible with the use of highly technical systems and targeted evaluation of data volumes.
It’s all about asking the right questions: how can the warehouse space be most usefully stocked, how much storage space will be needed in the future, and how many workers will be needed at specific times? The question as to which goods should be stockpiled where is a task for AI algorithms. Using historical data, algorithms can calculate where to store which goods, because one article is for example more in demand in a particular rural area, while another needs to be kept in city warehouses.
But even the associated workforce planning could in the future benefit greatly from big data and learning algorithms. Without a reliable forecast, there is the risk of assigning too many workers or ending with a shortage of staff during the specific period.
Reliable forecasts are valuable especially at peak times
This challenge has been taken up by Moritz Gborglah, Head of Sea & Air Operations at Hermes International. The Frankfurt branch of Hermes International regularly receives airfreight, especially clothing for the mail order company Otto, which is then forwarded from there to the destination warehouse. “The warehouse is only informed how much merchandise to expect and in which parcel sizes and weights five or even three days before the goods arrive. At such short notice, targeted workforce planning is of course no longer possible”, explains Gborglah. The question was, why does the warehouse only get the details when the goods are packed and shipped and not when the order is placed? This takes place four to eight weeks before delivery, which would make planning easier.
By tapping into different data sources with order data from the past and connecting them to the current order, it was possible to create a reliable forecast. “We were able to predict how many employees would be needed on the day to handle the incoming quantity of goods and their weight. Especially with part-time employees who are hired at peak times, this is particularly valuable, since we only book the workers we really need”, says Moritz Gborglah.
Now that it has been demonstrated that this forecast can be reliably generated, the actual implementation of this forecast for workforce planning will be initiated in Frankfurt. “We want to proceed with this quickly; for us this was a test run with which we wanted to show that you can easily start a trial, even without lengthy preliminary planning – which has now proved successful”, said Gborglah.
Precise and efficient workforce planning as well as optimisation of processes and workflows with the help of technologies and digitisation are gaining importance not only in the warehouse but also in other areas of logistics – especially since there is a shortage of skilled workers. The transport logistic trend barometer 2019 shows that a shortage of skilled workers is one of the biggest challenges that companies face.
Noise evaluation: will the machine be shut down soon?
With the right number of employees working in the warehouse, further data collection can make their work more effective while protecting their health and well-being. For example, the Fraunhofer spin-off “Motion Miners” anonymously records large volumes of movement data using data bracelets. From the data collected in this way, unnecessary journeys can be avoided and, for lifting movements that are clearly detrimental to body and health, shelves or boxes can be lifted to a height that makes the work easier on the back.
And finally, it is not only the people in the warehouse that reap the benefits: with so-called “predictive maintenance”, artificial intelligence can also help the machines in the warehouse by predicting breakdowns and thus ensure that work in the warehouse continues successfully and without interruption. It evaluates the noises made by the machines and recognises whether they will break down soon, notifies the workers and even explains how to remedy the situation. Furthermore, only parts that are actually worn are replaced and not those that can still last for many hours of work – and this has been impossible to differentiate until now. There is a huge market for predictive maintenance – in a study published in July 2018, the consulting firm Roland Berger predicted a growth of up to 40 percent annually by 2022, when the market is expected to be worth 11 billion US dollars. So far, too many companies have limited themselves to merely collecting data and shied away from spending money to really create value from this field.