Faults in power systems cause voltage sags, which, in turn, provoke large current peaks in gridconnected equipment. Then, a complete knowledge of the inverter behaviour is needed to meet fault ridethrough capability. The aim of this paper is to propose a mathematical model that describes the behaviour of the currents that a three-phase inverter with RL filter inject to a faulty grid with symmetrical and unsymmetrical voltage sags. The voltage recovery process is considered, i.e., the fault is assumed to be cleared in the successive zero-cross instants of the fault current. It gives rise to a voltage recovery in different steps (discrete voltage sag), which differs from the usual model in the literature, where the voltage recovers instantaneously (abrupt voltage sag). The analytical model shows that the fault-clearing process has a strong influence on the injected currents. Different sag durations and depths have also been considered, showing that there exist critical values for these magnitudes, which provoke the highest current peaks. The analytical study is validated through simulations in MATLAB TM and through experimental results.

No abstract

Transformer Rectifier Units (TRUs) are a reliable way for DC generation in several electric applications. These units are formed by multiple three-phase uncontrolled bridge rectifiers connected according to two main topologies (parallel and series), and fed by a phase-shifting transformer, which can have different configurations. Fault diagnosis of the uncontrolled bridge rectifier diodes is one of the most important concerns on the electronic devices, nonetheless, rectifier units are inherently not protected in front of Open-Circuit (O/C) faults, which cause malfunction and performance deterioration. In order to solve this drawback, the proposed fault diagnosis method is based on the O/C fault signature observed in the DC-link output voltage of TRUs rectifier. It allows detecting the O/C diodes of parallel and series TRUs with different phase-shifting transformer configurations and for the most usual fault scenarios. Moreover, it also helps the prediction of diodes that could be exposed to failure after the fault, which provides corrective maintenance for the TRU development. The proposed method is illustrated from MATLAB TM numerical simulations of a 12-pulse TRU, and is validated with experimental tests.

Average value of the DC-link output voltage is a variable of interest in multi-phase uncontrolled bridge rectifiers. The aim of this paper is to present a new, effort-saving procedure capable of providing an accurate value of this variable, a value which can be later corrected considering the usually omitted voltage drops. The proposed method, based on the Cauchy's formula (1841), allows the limitations of the existing methods to be overcome and can be used under supply voltage balance and unbalance conditions. Time-domain simulations and experimental tests were conducted to show the usefulness of the method and validate its accuracy. Under supply voltage balance conditions, the new method allows results as accurate as those provided by analytical expressions available in the literature or time-domain simulations performed by any software to be obtained. Moreover, under supply voltage unbalance conditions, this method outperforms analytical expressions available in the literature and at least equals time-domain simulations performed by any software in terms of accuracy of the obtained results. Therefore, under supply voltage balance and unbalance conditions, the proposed method makes the mathematical effort required to elaborate analytical expressions or the computational effort required to perform time-domain simulations unnecessary. In addition, the new method provides suitable estimates of values experimentally determined.

Three simplified models for adjustable-speed drives are compared with measurements on an actual drive where the\ud inverter control strategy remains unaltered during the event (also known as continue mode of operation) to assess drive\ud behaviour prediction during balanced and unbalanced sags. In particular, the models calculate the dc-link voltage and ac line\ud current dynamic evolution, assuming that neither the drive nor the system protection will trip. The comparison reveals minor\ud differences in behaviour between the three models and the actual drive, although the constant current model exhibits slightly\ud a better agreement. To verify whether differences in the simplified models remain small, extensive ranges of sags are\ud simulated, and drive sensitivity to sag type and depth is illustrated. As the straightforward models studied predict the\ud measured drive dynamic behaviour in severe network conditions with similar accuracy to other unnecessarily complicated\ud models, they are very useful to reduce computational burden and overall complexity for sag immunity as well as dynamic and\ud steady-state network studies. On the other hand, laboratory measurements on the actual drive when the inertia ride-through\ud mode of operation is active illustrate that the simplified models are not valid when the inverter control strategy is modified\ud during the sag.Peer ReviewedPostprint (published version