journal of materials processing technology 209 ( 2 0 0 9 ) 619–624 journal homepage: www.elsevier.com/locate/jmatprotec Design improvement using process capability data Kevin Dominic Delaney a,* , Pat Phelan b a Mechanical Engineering Department, Dublin Institute of Technology, Dublin, Ireland b Manufacturing and Operations, University of Limerick, Ireland article info Article history: Received 25 January 2007 Received in revised form 8 February 2008 Accepted 20 February 2008 Keywords: Tolerance allocation Process capability data Process capability database Electrical connectors abstract Historical process capability data (PCD) is commonly used during product design. Such data can be stored in a formal database known as a process capability database (PCDB). PCD is primarily used during product design to allocate tolerances or forecast and manage manu- facturing variation. However using a PCDB for these purposes alone does not exploit its full potential. This paper describes a method by which PCD can be applied to predict product performance variation early in the design process, thus saving time and reducing subse- quent redesign costs. More specifically it describes the use of PCD during the initial design stage to predict how the product will operate, in terms of functional reliability, after it has been manufactured. An illustrative example based on the design of an electrical connector is provided. In this case connector beam normal force is a strong indicator of connector reliability and the example shows how the variation of this normal force can be predicted during initial connector design. © 2008 Elsevier B.V. All rights reserved. 1. Introduction According to Nickolaisen (1999) design engineers have tra- ditionally used a combination of trial-and-error, design carryover, and the opinions of process experts to assign product tolerances. Various approaches to dimensioning and tolerancing were outlined by Nickolaisen to justify the devel- opment of a formalised system of dimensional management. Due to the level of rework and engineering changes involved such traditional approaches can be expensive and time con- suming for engineering companies trying to get their products into the market before competitors. These competitive pres- sures require companies to take a more scientific approach to assigning tolerances and many now adopt a more data-driven approach to ensure that the tolerances assigned are process capable. Lucca et al. (1995) wrote an early paper to propose a for- mal database to store actual process capability data related to sheet metal parts and use this data systematically to assign ∗ Corresponding author. E-mail address: kevin.delaney@dit.ie (K.D. Delaney). achievable tolerances during the design of future products with similar geometries. Referring specifically to the high volume manufacture of precision metal stampings for con- sumer products this work highlighted the potential benefits that could be gained by implementing such a formal system. Building on this concept, King (1999) described how to create a process capability model (PCM) utilising process capability data. This work was inspired by the realization that six sigma quality levels are achievable only by improving the quality of new designs in addition to process improvement. A method to develop process capability models from real measurement data was outlined and expanded to include a method to val- idate and check the results of the predictive model created. Further developing this concept, Bauer (2002) proposed an indexing structure and a user interface for such a PCM. In addition to the potential benefits of applying PCD early in the design process this work listed some of the practical issues to be addressed when building, using and maintaining a PCM. 0924-0136/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.jmatprotec.2008.02.059