Client challenge

Since 1924, ENKA GmbH & Co KG has produced premium viscose fabric – ENKA® Viscose, which is used to make medical and technical garments – using a proprietary trademarked process. The yarn comes from wood, and producing the multi-filament viscose is a complicated task involving multiple cellulose concentrates that require numerous technical settings on the manufacturing machines. Data are collected for 360 process parameters, 365 days a year, in some cases as often as every second.

ENKA had built up more than a decade’s worth of this data and understood that detailed analysis could improve its process-dependent KPIs. The company also understood there were benefits to integrating a traditional and modern approach to manufacturing but the company needed to work with digital experts to develop the big data and machine learning tools to optimise the production process.

Approach

ENKA sat down with PwC’s interdisciplinary team that included statisticians, bioinformatics specialists and mathematicians. The aim was to identify how to reduce the error and reject rates and optimise the use of resources, all of which would have a direct, positive effect on the bottom line. The first step was for PwC’s data scientists to clean the company’s data and put together a machine-learning model. The team trained what is called a support vector machine using the historical readings and the results of quality measurements.

The second step was for the specialists to develop an algorithm that used all the inputs to determine the ideal setting for viscose yarn production. This algorithm repeatedly combines and modifies the process parameters which the ENKA specialists and the PwC team then discussed and interpreted to establish the optimal production process.

Impact

The outcome is a system that has reduced errors and optimises the resources that go into making ENKA^® Viscose. It uses the expertise of ENKA’s employees and the digital tools to configure the individual production steps in the manufacture of the yarn. When a variation in quality is found, it is now easier to identify and assess because the key process parameters are known and can be reconfigured. This makes a major contribution to optimising the KPIs and thus to the company’s success.