IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 56, NO. 7, JULY 2009 2697 Extended Operation of Cascade Multicell Converters Under Fault Condition Pablo Lezana, Member, IEEE, and Gabriel Ortiz Abstract—Multilevel converters are an interesting alternative for high power drives, due to their good quality output signals. De- spite their advantages, the large number of components required increases the fault probability. Among the multilevel topologies, the cascade multicell converter presents advantages when oper- ating under internal fault conditions, due to its high modularity. Previous works proposed to compensate the unbalanced operation due to a fault by changing the canonical fundamental output phase shift to precalculated angles, depending on the fault condition. This solution assumes that, if the maximum output phase voltage on each leg is used, the maximum line-to-line voltage will be at a maximum as well. This paper shows how this assumption is not always valid and presents the optimum angles and modulation in- dexes that must be used in order to obtain the maximum balanced load voltages. Index Terms—Converters, fault tolerance, multilevel converters. NOMENCLATURE x Electrical variable. x Phasor of electrical quantity x. X Magnitude of electrical phasor x. N Load neutral point. n Inverter neutral point. v Y Load voltage for phases a, b, and c, where Y ∈ {A,B,C}, respectively. v y Inverter voltage for phases a, b, and c, where y ∈ {a, b, c}, respectively. A - B - C Number of operative cells on phases a, b, and c, respectively, where {A,B,C}∈ Z + . I. I NTRODUCTION A FTER two decades, multilevel converters are still in con- tinuous development in fields such as Static Compen- sators [1]–[3], photovoltaics [4]–[6], topologies [7]–[9], and modulation and control techniques [10], [11]. All the multi- level topologies, namely, neutral point clamped [12], cascade multicell (CM) [13], [14], and flying capacitor [15], require a large amount of components [16] in order to distribute the voltage (and, hence, the power) among them. An associated issue related to a high number of components is an increase Manuscript received December 11, 2008; revised February 4, 2009. First published April 14, 2009; current version published July 1, 2009. This work was supported by the Chilean Research Fund (FONDECYT) under Grant 1085111. P. Lezana is with the Departamento de Ingeniería Eléctrica, Universidad Téc- nica Federico Santa María, Valparaíso, Chile (e-mail: pablo.lezana@usm.cl). G. Ortiz was with the Departamento de Ingeniería Eléctrica, Universidad Técnica Federico Santa María, Valparaíso, Chile. He is now with the Power Electronic System Laboratory, Eidgenoessische Technische Hochschule, Zürich, Switzerland (e-mail: ortiz.gabriel@gmail.com). Digital Object Identifier 10.1109/TIE.2009.2019771 Fig. 1. (a) Eleven-level CM converter. (b) H-bridge cell fed by a diode bridge. in the probability of internal fault. However, most multilevel converters allow themselves to be reconfigured in order to work in an under-rated operation mode [14], [17]. This paper is focused on the CM converter operation after an internal fault condition has been detected, either by sens- ing each power switch or using a more sophisticated method [17]–[19]. A good characteristic of CM converters is that faulty cells can be isolated from the system by using an external switch [T in Fig. 1(b)] that even allows the faulty cell to be replaced by a new one without turning off the system [20]. Then, the problem becomes how to obtain the highest power level with the remaining operative cells. In [21], the redundant states of the CM converter are used to avoid the switching states that are no longer available due to the fault. As space-vector modulation is used, the well-known hexagon obtained from the α-β transformation changes its geometry, and depending on the fault, a nonregular hexagon or even a rhombus is obtained. Then, a diminished maximum bal- anced stationary voltage vector, defined by a circular trajectory, is reached. A different approach is used in [20], where triangular carrier- based PWM modulation is used. In this paper, the fundamental output voltage phase shifts are used to recover the balanced operation. As the phase shift between the inverter output volt- ages is no longer 120 ◦ , a fundamental component is injected into the common mode voltage between the inverter and load neutral points, which not only increases the load voltages but also the voltage stress on the motor bearings, which can lead to a parasitic current sent through them [22], [23]. 0278-0046/$25.00 © 2009 IEEE Authorized licensed use limited to: Universidad Tecnica Federico Santa Maria. Downloaded on August 13, 2009 at 14:16 from IEEE Xplore. Restrictions apply.