Planning uncertainty in the terry cloth and home textiles sector has a new answer for Frottana Textil GmbH & Co. KG, the manufacturer behind the MÖVE brand. The tool, built by Fraunhofer IWU working with Logsol GmbH, uses neural networks to analyse past sales records and generate monthly forecasts, replacing manual and Excel-based planning processes across the business.
- The tool automatically identifies trends and fluctuations in demand across the year, providing a transparent and traceable basis for sales and order planning decisions.
- The model's accuracy holds even for months with stronger-than-usual fluctuations, capturing larger deviations rather than smoothing them out.
- Employees can actively review, adjust and enrich forecasts with their own expertise, combining AI-based analysis with human knowledge to increase acceptance within the company.
BEHIND THE PROBLEM: The terry cloth and home textiles sector, largely made up of medium-sized enterprises, faces persistent planning challenges driven by seasonal demand peaks in spring, the holiday season and Christmas. Many companies still rely on Excel spreadsheets, handwritten lists and the experiential knowledge of long-serving staff. A growing shortage of skilled workers, combined with age-related departures, has accelerated the loss of institutional know-how, while manual data transfers from ERP (Enterprise Resource Planning) systems add avoidable costs. Frottana Textil is among the companies now deliberately embracing digitalisation to place planning on a solid data foundation.
- Demand peaks across the year create major scheduling challenges for manufacturers that also need to maintain supply against a stable base demand between peak periods.
- Some companies continue to recreate Excel spreadsheets every month for thousands of products, or resort to handwritten lists, despite holding extensive historical sales and production data.
- The growing shortage of skilled workers and age-related departures result in the loss of valuable institutional know-how that has historically underpinned planning decisions.
- Extensive historical sales and production data often goes unanalysed in companies that rely on manual processes, compounding planning inefficiencies rather than resolving them.
- In the SmarMoTEX project, Fraunhofer IWU developed material flow simulation tools enabling virtual process modelling, bottleneck identification and alternative production strategy testing without disrupting live operations.
THE ACCURACY CASE: The forecasting tool has demonstrated high accuracy even with a limited data foundation. Results show that 82.7% of sales fluctuations are explained by the model, including accurate representation of stronger deviations in individual months. This level of performance was achieved using only four years of historical data and without differentiation by sales channels, regions, promotional effects or COVID-related special influences.
- At an average of 340 units sold per month, the forecast shows a typical deviation of only about 38 units from actual sales figures, a performance achieved without differentiation by sales channels, regions or promotional effects.
- The model accounts robustly for stronger deviations in individual months, particularly in accounting for the impact of forecasting errors.
- The combination of AI-based analysis and human expertise has significantly increased acceptance of the tool within Frottana Textil, with employees actively reviewing and enriching forecasts alongside the automated system.
- Employee onboarding becomes faster under the digitalised process, and absences of experienced staff can be more readily compensated for by less specialised colleagues.
