An engineer at a final assembly OEM was called to investigate a product that had passed through production with a loose bolt. The investigation concluded that the bolt had been tightened properly, according to the recorded data. The assembly had passed all required checks and reached the prescribed torque. The engineer had a problem: How could this bolt, that met their required challenges, be loose at the end of the assembly?
The engineer started their investigation by looking at the torque trace, a graphical representation of the tightening. Investigation of the torque trace showed a different picture than a typical assembly and a properly tightened bolt. The engineer could quickly see that the beginning of the tightening looked very different from a typical assembly.
The first task was to modify the process to catch this issue in the station. The engineer estimated that 50,000 products had passed through this station with the current process. The prospect of either recalling all 50,000 products or manually reviewing 50,000 torque traces was a daunting one, and would cost the company significant resources. Fortunately, new data analytics capability from ToolsNet 8 allowed the engineer to avoid either of those costly measures.
The engineer used the Box Analysis function in Atlas Copco’s ToolsNet 8 data analytics tool to automatically evaluate all 50,000 torque traces for the affected assemblies. By using this function, they were able to catch nine torque traces that matched the characteristics of the loose bolt they identified. The torque data was tied to the serial numbers of those nine products, confirming that they could be clearly identified, which saved the OEM from having to consider recalling 50,000 vehicles. There was a clear monetary gain, but more importantly, this helped the OEM protect their brand name.
ToolsNet 8 helps customers find assemblies with missing washers, cross threads, material inconsistencies or other irregularities. The Box Analysis feature compares a torque trace to a user defined reference trace, and to up to ten user-defined criteria to identify deviating assemblies. If an assembly deviates from the predefined criteria, ToolsNet 8 presents the deviation, along with the identifier (such as serial number), program name and result time. The software automatically evaluates all torque curves for the selected application, eliminating the need for manual review. This is also useful for process improvement, or to identify the products affected by a recall.
ToolsNet 8 is also capable of monitoring incoming production data and notifying users by email whenever Box Analysis finds a deviation.
Some manufacturers do not systematically collect data in a central location, and most have a reactionary approach, running reports after an issue occurs. Many are starting to see the importance of collecting data and performing proactive data analysis. ToolsNet 8 continues to be successful in helping production engineers automate data analysis, improve production processes, and be proactive.