Smart Sensing of Forging via Tonnage Analysis
Provided by Judy Jin, University of Michigan; Andreas Sterzing, Jürgen Steger and Tim Lehnert, Fraunhofer Institute (IWU)
The background of this FIERF-funded research project is that tonnage signal analysis provides rich information about forging process operating conditions. It has the potential to predict part quality for achieving intelligent forging process control. Generally, the tonnage signal profiles are affected by the design of the part and die as well as process operating conditions (e.g., material properties, workpiece and die geometry, temperature, positioning error, etc.) If, for example, a significant tonnage change is detected, it would usually reflect that process conditions have abnormally changed, and the part quality may also be severely affected.
The research team conducted experimental studies to investigate how three typical process variables (workpiece geometry, temperature and die position) affect the tonnage signals. The experimental tests were conducted at Fraunhofer IWU. The testing tool is used to forge a small connecting rod for chain saws, with the capability of adjusting the position of the tool on the press table to simulate the eccentric loading during the tests.
The tests were conducted by studying three important factors under different testing levels: part temperature at four different degrees, part geometry with three different angles and eccentric errors with four different distances from the center. To consider the inherent part-to-part variation, five replicates of parts were conducted under each of the testing conditions. During the tests, the press-tonnage force profiles are collected by the strain gauge sensors installed on the left and right tie rods of the press. However, only the maximum force over each operational stroke is provided by the available press monitoring system.
In order to obtain the tonnage profile of the whole cycle, the specially designed in-die sensor was used during the operating stroke. The following analyses will be conducted for both the peak press tonnage signals and the whole in-die tonnage profile signal. Analysis of the results shows the advantage of utilizing the full tonnage-signal profiles, rather than only the peak press-tonnage data, in identifying the root cause of the process change.
Analysis of Results
The ANOVA (analysis of variance) method was used to analyze whether and how much each factor affects the peak force of the total press tonnage (Fig. 1).
Peak press tonnage alone may be insufficient to detect process changes. Therefore, a further study was conducted using the multivariate monitoring method on the whole tonnage profile.
The comparison between these two analyses shows that use of the tonnage signal profiles can provide better sensitivity than peak tonnage alone in separating different test conditions. Therefore, the use tonnage profile signals instead of single peak tonnage to achieve a better process monitoring and diagnostic performance is recommended.
This paper and many others will be presented in their entirety at the 31st Forging Industry Technical Conference Sept. 19-20 at the Nationwide Conference Center in Columbus, Ohio.