The scheduling and forecasting system at Anita Dr. Helbig GmbH
Felicitas Heid-Davignon (Anita Dr. Helbig GmbH) and Dr. Bernd Reineke
Hardly any other industry is as dependent on the correct assessment of sales potential as the clothing industry. Anita Dr. Helbig GmbH, Brannenburg, therefore decided to use the “Diskover SCO” (Supply Chain Optimizer) forecasting and scheduling system to support sales planning in order to predict customer expectations as accurately as possible using distribution-free forecasting methods. The special feature of this installation is the so-called black box, which works autonomously in the background.
As a stock manufacturer, Anita Dr. Helbig GmbH, a specialist supplier of corsetry and swimwear, relies on good planning quality in order to avoid unnecessarily high stock levels and still be able to deliver the items ordered by customers on time. In the past, sales figures were calculated by averaging past values. As a result, product trends were only recognized very late and the quality of planning declined as the product range changed more frequently. This led to increased inventories of non-turnover items and a lack of availability for items in high demand.
To improve the situation, the company decided to support the planning process with a specialized planning tool, the Diskover SCO system from Abels & Kemmner GmbH. Before the system was implemented, a detailed analysis of the sales items was carried out. By simulating several scenarios, the readiness to deliver and the inventory level were optimized. In particular, the addition of distribution-free methods showed significantly better results than simulations without these control variables.
Although Diskover SCO was initially only used to optimize demand forecasts, it was also possible to implement special requirements, such as the regular adjustment of batch sizes, for example taking packing units into account, within a short space of time. After coordinating the data structures and additional functionalities, the system was put into operation in just a few weeks. Microsoft Access was selected as an inexpensive database solution, which is perfectly adequate in terms of both the expected data volume and performance and also offers simple options for evaluating the data flexibly and in line with requirements.
For optimization purposes, the scheduling and forecasting solution receives interface data such as master data, movement and order data in the form of ASCII files with a pre-agreed structure. Optimization runs lasting several hours are then carried out once a month, usually at the weekend. After optimization, the results are transferred back to the higher-level ERP system based on an IBM AS400 computer in the form of text files. Diskover SCO therefore works as a black box in the background without direct intervention by the dispatcher.
Due to the high degree of automation, this solution can be operated without great effort. And since the software from Abels & Kemmner acts as a black box, no training is required. The monthly optimized forecast data is made available in the leading ERP system without changing the front end. Consequently, nothing has changed for the dispatchers. Only the delivery readiness and the inventory situation improved continuously. All this is achieved without much intervention in day-to-day business and simply by selecting the right scheduling procedures and parameters.
http://www.anita.com/
https://ak-online.de/
Distribution-free processes
belong to the statistical forecasting methods. They should always be used if the fluctuation in consumption values per period is not normally distributed. Experience shows that only around 25% of all consumption series are normally distributed, i.e. three quarters of the items to be forecast do not have a normal distribution. Such items can therefore only be forecast to a limited extent or not at all using the standard methods of conventional forecasting or ERP systems, as these usually only offer methods that assume a normally distributed demand (e.g. mean value method, exponential smoothing). If the normal distribution is not given, conventional forecasting methods produce very inaccurate results and lead to incorrect delivery readiness levels.
