Abstract:
The manufacture of pharmaceutical products and medical devices is increasingly carried out by automated manufacturing systems. These systems are typically electromechanical devices controlled and monitored by software written in a high-level language, communicating with the equipment via a motion controller. These controllers convert the digital software outputs to the analogue signals required for machine control, while providing feedback to the software system.
The industries concerned are subject to a number of federal and international regulations. A key common requirement of these regulations is that all manufacturing equipment (including software) is validated for its intended use according to an established protocol. The validation of such equipment can be time-consuming, and hence can represent a substantial cost to a company in terms of lost production, as all validation activity must be carried out on the production equipment. In addition to this, developing software modifications also requires access to production equipment.
This research investigates the feasibility of carrying out the equipment validations described above using software models of the physical equipment and controller hardware. With suitable models in place, the validated controlling software could interface with the models in exactly the same way as it would with the real thing. The models and software could hence be validated in their own right. This would reduce the cost of validation by allowing at least some of the validation effort, as well as development and testing, to happen off-line.
This investigation comprises three main strands:
1. A study into the potential for acceptance of such a validation strategy with the bodies which enforce the relevant regulations.
2. A critical analysis of the existing applications of software modelling in similar and/or contrasting contexts.
3. A proof-of-concept prototype comprising software models of controller hardware and simple electro-mechanical systems which interfaces with a controlling software application.