Industrial Statistics

Complex production processes have to be optimally designed and permanently monitored. The statistical methods of Numerics help ensure stable process outcomes and quality of data. Deviations from the target values ​​are to be revealed and underlying root causes being identified as soon as possible.

With control charts Numerics can monitor processes for quantitative and qualitative characteristic values ​​visually. Preceding quality control and production monitoring steps, and existing production processes can be optimized by means of methods of industrial statistics. These methods are based on statistical test planning (DoE, Design of Experiments) to identify the optimal settings of the process parameters. Experiments are performed such that potential influencing factors are varied simultaneously to obtain the maximum amount of information from a smallest possible number of trials. Advantage: processes are more stable and less prone to errors and at the end run more economically.


Manufacturing processes being optimized by statistical methods: often findings can be drawn retrospectively from data resulting in optimizing the business. In addition to multivariate methods such as “tree regression,” graphical methods are applied. This approach eases root cause analysis of flaws in individual production parts. Represented visually, possible patterns and relationships can be uncovered quickly and remedy actions  taken thereby increasing the quality and efficiency of the industrial processes.

Experimental Design or Design of Experiments (DoE) is a method to quantitatively determine correlations between potential factors and outcomes with as little effort as needed. From a set of potential factors the relevant variables are being determined. In addition, optimum factor levels settings can be determined to match the target value (i.e. minimize or maximize of yield). Both tasks will be solved with as little experimental effort as possible, which is with the smallest possible sample – characterizing the art of optimal DoE. Our methods include: factorial plans, fractional plans (i.e., restricted interactions), and robust experimental design (Taguchi).

By means of the measurement system analysis, the basic suitability or capability of a measurement system is determined. The measured values have to occur within the pre-specified upper and lower tolerance limits. Using selected criteria, the process capability can be objectively assessed. As deviations emerge from the operator, the badge materials or the device characteristics can be revealed in a straightforward manner.

The Statistical Process Control (SPC) is  implemented as an ongoing monitoring process. Control charts for monitoring mean (average location) and variability are applied (x, s – card) or for attribute characteristics such as failures p- cards. For small samples r – cards can also be used (range, i.e., minimum – maximum). We work with specialized software tools to implement such an automated process control based on scripts to be run on a routine basis.