Abstract:
The Load Commutated Inverter (LCI) fed synchronous motor has gained wide acceptance as a high power variable speed drive due to its excellent efficiency and reliability characteristics. The standard LCI drive configuration comprises a fully controlled six pulse thyristor bridge rectifier (using conventional SCRs) which is connected to a similar fully controlled thyristor inverter by a DC link inductance. The inverter is attached to a wound rotor synchronous machine. In this Thesis a new, algebraically simple, steady-state, linear model of an LCI fed synchronous motor drive with a finite DC link inductance is presented. The behaviour of the DC link inductance is included in the machine equations in an implicit fashion, thereby avoiding the difficulties involved in direct incorporation. This is facilitated by deriving a simple equivalent circuit, representative of the DC link and inverter, at voltage ports which connect the power electronic network and the machine. The power electronic analysis is based on a simple backward Euler integration algorithm and considers the inverter SCRs to be two state variable resistors. The machine is modelled by Park's equations. Initial currents for a steady state operating point are found by iteration; direct calculation being found to be insufficiently accurate. The first iteration to the initial conditions and the source voltage / field voltage ratio are calculated by a state space method which assumes infinite DC link inductance. A practical simulation time step is selected by simulating an ideal current source in the DC link. A constant mechanical firing-angle LCI drive system (using a 3 kVA microalternator) was constructed in order to experimentally verify the· new modelling strategy. The drive featured a filtered DC source voltage and a forced commutated start up strategy. An excellent level of agreement between the theoretical and measured results was achieved when the machine was operated in its linear region. Significant discrepancies did not arise until the machine was heavily saturated. A comparison between the new LCI drive model, a state space model which assumes infinite DC link inductance, and a third model which considers the machine to be a voltage behind a subtransient reactance was performed for an LCI drive with constant inverter margin angle control. Both cylindrical and salient pole machine configurations were examined. There was a good level of agreement between the two time domain models except in the predicted levels of torque ripple. Other differences only became apparent when the DC link inductance was very small. The "voltage behind the subtransient reactance" method was only slightly less accurate than the infinite DC link inductance method except where the machine subtransient reactances were relatively large. Significant differences between the three models were noted, however, when a machine without damper windings was studied, due to the increased impact of DC link current ripple. The versatility of the new technique was demonstrated by the inclusion of source voltage ripple for some operating points. The new modelling strategy was also applied to other drive system configurations - an induction motor supplied by an inverter with 120° conduction and a similar ir.verter fed Permanent Magnet (PM) motor drive - and a sample operating point was calculated for each system. The initial condition iteration process was not convergent for the induction motor drive; thus the new model is better suited to machines with standing back emf components. A simple predictor-corrector algorithm was used to predict a sample step response for the above PM motor drive and it is hoped that this method may have application in dynamic analyses of LCI drives.