An industry case - Prediction of Glass Cartridge ...

Object Details


International Symposium on Robust Design (ISoRD'14) Technical University of Denmark, DTU

An industry case - Prediction of Glass Cartridge Robustness in Assembly line Loading
Author:Torben Hansen (Device Research and Development, Competency Centre, Novo Nordisk A/S, Denmark)
Bo Kristiansen (Device Research and Development, Competency Centre, Novo Nordisk A/S, Denmark)
Date: 2014-08-14     Track: Main     Session: 10:30-12:00 Oral Session 1 - Robustness in Design

Each year Novo Nordisk produces multimillion injection devices incorporating drug contained glass cartridges. These cartridges will inevitably be subjected to various loadings in both line feeding systems and in the device assembly rigs. It is obviously crucial to preserve the structural cartridge integrity and avoid any form of cracking and fragmentation of the glass for the full life time of the devices. The robustness is quantified by a safety factor against cracking. As shown in figure 1, it implies that both assembly line loadings and the strength of the sub-supplied cartridges are determined along the location of max stress in relation to the rotatory position of the weakest region. These figures can be used to specify the loading for the incoming inspection and prevent future device designs from overloading the cartridges. The cartridge glass is brittle with a low cracking energy and sensitive to impact loading. The glass strength is determined by microscopic manufacturing related imperfections (which have no influence on the performance and integrity of the final device) and is both loading mode and rate dependant. Thus a conventional material property such as the ultimate stress cannot be used to calculate a safety factor. Matters are further complicated by loadings being tolerance dependant and by the fact that each cartridge has a randomised position of the weakest region relative to the maximum loading. This calls for a statistical based calculation of the safety factor. Figure 2 shows an example on a safety margin without distribution overlap. In some cases a limited distribution overlap might be acceptable.