PhD for Nick Jordan

Nick Jordan, Founding Co Director of EPC Services, was awarded his PhD by South Bank University, London in November 2006.

The research was entitled  "Protection Co-ordination with the D.C Traction Network" and was supervised by Dr  Ray Pettitt.

The subject of the thesis was protection co-ordination within a D.C traction system, it was focused upon a short-circuit fault positioned remote from the detecting circuit breaker as this is the most onerous condition to detect reliability. This research considered  the 3rd  D.C rail conductor system, however with modest parameter adjustment it may be equally applied to either a 4th rail conductor or overhead centenary system.  The  D.C traction system is a relatively complex distribution network as it consists of dynamic loads generally traversing a non-ideal ferromagnetic conductor system, which is also subject to real time movement control. The D.C protection strategy employed has a fundamental influence upon the initial design and subsequent operational performance of the network. The principal factors to be considered in a protection study include assessment of the sustainable load (within given control and parameter constraints), the fault characteristics remote from the feeding circuit breaker and the method by which circuit interruption may be initiated.

This thesis describes a process for determination of key system parameters under various feeding configurations. It includes a new modelling technique for the twelve pulse traction rectifier which has been verified by correlation to a six pulse rectifier when completely uncoupled. Detailing modelling of internal and external parameters of the conductor system within the remote fault interval is described taking into account parameter sensitivity. As a result a number of discoveries have been made in relation to performance of both the traction rectifier,  conductor system and protection system under a remote fault condition.

The MathCAD models developed have been subject to practical validation, a high degree of correlation has been achieved to real-time system measurements. The models have also been subjected to sensitivity analysis to determine the effect of any change in either network configuration and/or parameter variation to the simulated results.The research conducted has reduced uncertainty of remote fault parameter prediction.

To further decrease uncertainty in system performance a new D.C protection philosophy based upon application of a pattern recognition artificial neural network has been outlined. This system automatically adjusts the protection setting thereby improving fault detection and its discrimination from a legitimate load.