Defining and assessing part complexity

Abstract

Companies require information about their customers to maintain an attractive portfolio of products and services. This information is essential in different departments for a variety of tasks along the product life cycle. By providing this information based on data, decision-making processes can lead to more accurate outcomes and improved overall results. This thesis explores part complexity and its relevance in better understanding customer needs throughout the product lifecycle. The industry partner of and example used in this dissertation is TRUMPF, a German machinery and plant engineering company that builds machine tools for the sheet metal processing industry. Sheet metal processing contains process steps such as laser cutting, bending, and welding to manufacture sheet metal parts. The application possibilities are manifold, with use cases in industries such as automotive, construction, renewable energy, aerospace, and many more. Many researchers agree that part complexity represents the manufacturability of a part. However, we identify three major research gaps in this field: (1), there is no consensus regarding the research method of assessing part complexity but co-existing methods differ in how they investigate part complexity. (2), the part complexity influencing part characteristics have not been thoroughly researched, even less for our field of research, sheet metal processing. (3), only two part complexity use cases have been identified in the literature, despite the increasing demand for data-based information along the product life cycle. This thesis addresses these three research gaps with these contributions: (1), we develop a methodology for assessing part complexity and demonstrate its applicability by putting this research approach into practice. This methodology combines both qualitative and quantitative methods. (2), we conduct a computer-assisted self-assessment to let experts label the complexity of 80 parts. To facilitate this self-assessment, we develop a labelling tool that implements a visualization of the parts and provides additional part information. For complexity labeling, we implement a Likert scale ranging from 1 (least complex) to 5 (most complex), deliberately omitting the middle option “3” to encourage more definitive responses. Participants are also required to provide a written explanation for their chosen complexity rating. The participants of the computer-assisted self-interview are experts for the production unit that we chose as an example for our research endeavor. Furthermore, as an evaluation mechanism, we repeat a subset of 10 parts in each of the three weeks of the complexity labeling to assess the consistency of the participants’ labeling over time. Second, we observe a consensus of the labeling participants in some of the repeating geometries and a clear correlation between distinct part characteristics and the reasons given for the assigned complexity ratings. (3), we identify part complexity influencing part characteristics based on the results of the aforementioned research approach. (4), by conducting a focus group, we explore the application possibilities of part complexity for the three main stakeholder groups: product and portfolio management, research and development, and sales and consulting. These results add to the two use cases of part complexity that have already been identified in the literature and demonstrate the usefulness of part complexity as a contributor to data-based customer information along the product life cycle.