Design and implementation of a hardware reliability simulator for multiple component systems.

Reliability is a critical feature in systems which perform a critical safety or economic role as failure can result in catastrophic loss. There is a need to demonstrate that such systems will perform as required but the ability to do this is restricted by the available techniques. Conventional approaches to reliability assessment fall into two broad categories both of which are limited in their application. Consequently there is a requirement for an alternative method of reliability evaluation and this forms the basis of this thesis. It is proposed that accelerated simulation using a novel computer architecture and a tailored algorithm is a viable means of overcoming the drawbacks of the current techniques. The initial stage of the development of SPRINT II, a computer dedicated to performing reliability simulation, is the identification of key features of both the machine architecture and the algorithm it supports. The characteristic entities of reliability models and the way in which their attributes are modified during simulation are used to develop the algorithm in detail. The algorithm was implemented in software in order to demonstrate that it performs the intended task. Complex maintenance policies have not previously been modelled in detail by reliability assessment techniques. The features of maintenance policies important to reliability assessment are identified and are included in the algorithm for SPRINT II. The approach taken by the machine architecture is to support a separate dedicated processor for 5 identified simulation functions within the algorithm. A special purpose hardware unit was designed to implement each in what is argued to be the most suitable way. The hardware is specified such that is capable of replacing the corresponding software routine thus offering an incremental approach to simulation. The correct operation of the simulation algorithm is demonstrated by evaluation of a number of standard problems. The performance of the dedicated hardware is assessed and it is believed that the performance is accelerated by a factor of up to 50 times that of software simulation.

Grantor dc:publisher.institution

Robert Gordon's Institute of Technology

Year dc:date.issued

1991

Author dc:creator

• Kermack, Calum S.

Advisor dc:contributor.advisor

• A.J. Miller

• Hardware reliability simulator
• Multi component systems
• Reliability evaluation
• Software simulation
• Novel computer architecture
• Bespoke algorithm
• Maintenance policies

Language dc:language

en

Identifier

oai:rgu-repository.worktribe.com:2807425
https://doi.org/10.48526/rgu-wt-2807425

OAI identifier oai:identifier

oai:rgu-repository.worktribe.com:2807425