1
Introduction to Fault-Tolerant Machine Drives

1.1 Background of Fault-Tolerant Machine Drives

Advanced electric drive systems are increasingly being used in a wide range of applications from industrial automation, household appliances, and transportation to oil and gas, mining, and renewable energy industries, where efficient and reliable electric-to-mechanical energy conversion or vice versa is essential. Extensive research activities on electrical drives have been undertaken in both academic and industrial organizations [1]. A typical electric drive is composed of a power converter, a control unit, and an electric motor, generally known as an electric machine, as shown in Fig. 1.1. The power converter contains typically power electronic devices (i.e. Insulate Gate Bipolar Transistor (IGBT), Metal-Oxide-Semiconductor Field-Effect Transistor (MOSFET), SiC, and diode), gate drives, and passive components (capacitor and damping resistors, etc.), which are responsible for driving the motor. The control unit consists mainly of microprocessor and its associated electronic circuitry in addition to various sensors. The electric motor delivers controllable torque to a mechanical payload by converting electrical power into mechanical power or vice versa.

As advances emerge fast in the areas of materials, electric machine design and manufacturing, power electronics, microprocessors, and sensing, electrical drives are capable of delivering desirable features such as high-power density, high efficiency, low emission, and good controllability, compared to other counterparts, namely, mechanical, hydraulic, or pneumatic drive/actuation systems [2]. Aircraft employing electrical actuators, electrical propulsion, and power generation in the form of more electric, hybrid, and full-electric aircraft can leverage the merits of weight saving, economical fuel consumption, low CO2 emission, increased functionality, and less maintenance [3]. Another emerging example is the electric vehicle (EV) replacing traditional internal combustion engine (ICE) for low CO2 and low harmful pollutant emissions [4, 5]. However, high reliability is also an essential requirement for these safety critical applications, which should be addressed at the system design stage [6].

Figure 1.1 Illustration of a typical electric drive system.

In the aforementioned safety critical applications, the electrical drives are expected to continue operation if a fault occurs, or at least being fail-safe without catastrophic damage [7]. Otherwise, the unexpected fault may cause casualties and huge economic losses [8]. Thus, fault tolerance should be considered to attain the reliability requirement for the targeted applications.

Fault tolerance means that the system is capable of performing at a satisfactory level of operation in the presence of fault. It is a common requirement that has been investigated in various areas, such as fault-tolerant computing systems [9], distributed power systems [10], and high availability internet servers. In the scope of electrical drives, the fault tolerance mainly means it is capable of maintaining the original or an acceptable output torque or power level after a fault. The acceptable level defines the minimum output, which should be considered at the primary design stage of such systems.

1.2 Frequent Faults in Electric Drives

An electrical drive is a complex electromechanical system composed of an electronic controller, a power converter, an electric motor, and sensors. These components are exposed to electrical, thermal, mechanical, and environmental stresses as well as chemical corrosion. Fault may occur in each of these components. Studies in [11, 12] have been carried out to investigate the failure distribution of electric machines. The results of the survey show the bearing faults account for the majority of the failures, as much as 51%, followed by stator winding faults, up to 25%. Other faults such as rotor bars and end rings in induction machines, shafts, and other unidentified failures take up the remaining percentage in Fig. 1.2(a). The investigation data in [13] also illustrates that electrical winding failures amount to a failure rate of 1.4 × 10-7 failures per hour in military-grade machines and 1.0 × 10-6 in industrial machines. Since these surveys are mainly focused on induction machines, permanent magnet (PM) failure in permanent magnet synchronous machines (PMSMs) is not included. In fact, partial demagnetization is a frequent fault for PM machines due to a strong armature reaction field, overheating, and excessive mechanical stress and vibration [14].

Figure 1.2 Fault distribution in electrical drives: (a) machine side and (b) converter side.

A similar industry survey was conducted on failures in converters in [15]. The survey indicates that the most vulnerable component is the switching devices, followed by capacitors and gate drive circuitry. Open circuit in one phase due to device and connection failures is also a frequent fault. Failures associated with resistors and inductors are quite rare and only observed in a few applications, as shown in Fig. 1.2(b). The survey result shows most IGBT/MOSFET device failures result from thermal and power cycling, with a typical failure rate of 2.78 × 10-6 failures per hour. Additionally, the controller and sensors may also experience faults during operation. Nevertheless, it should be noted that the probability of the microcontroller and sensor faults is much lower.

As mentioned above, many potential faults may occur in the system. In this chapter, the principal device and electromagnetic faults under consideration that may occur within an electric drive are shown in Table 1.1.

On the machine side, the winding insulation degrades gradually due to electrical, thermal, and mechanical stresses and finally develops into open-circuit or short-circuit failure. The short-circuit failure can be classified as interphase and intraphase short circuit, which occur between phases or within a single phase, respectively. The intraphase fault is usually caused by turn-to-turn insulation failure. In particular, an intraphase fault involving a few turns, also known as a turn fault, is reported as the worst-fault scenario since only a few turns are short circuited. The resultant fault current is massive and the excessive hotspot temperature may lead to catastrophic failure. Partial demagnetization is another common fault in PM machines due to the excessive armature reaction field, overheating, and a high level of mechanical stress and vibration. It may cause torque reduction and increased torque ripple, etc.

Table 1.1 Potential faults occurring in an electric drive.

Machine side Drive side Winding open circuit Switch device open circuit Winding interphase short circuit Switch device short circuit Winding intraphase short circuit DC-link capacitor failure Demagnetization Controller/sensor failure Uncontrolled generation failure at high speed

On the drive side, the switch device is also subjected to open-circuit and short-circuit failure due to electric and thermal-mechanical stress during repeated switching on and off operations. DC-link capacitor is exposed to combined electrical and thermal stress during inverter operation and hence contributes to a considerable failure rate in electric drives [16]. Gate drive failure gives rise to similar consequences of switching device failures and may be incorporated into the switch device fault mechanism.

Besides, another possible fault is the uncontrolled generation, particularly for PM machines. If the power converter fails when the machine is rotating at high speed, the electromotive force (emf) may be much higher than the DC-link voltage and consequently cause uncontrolled rectification via the freewheeling diodes in the power converter. This may damage the DC-link components if the generated power is excessive and cannot be absorbed [17, 18].

So far, most of the fault-tolerant electrical drives focus on the faults described above [8, 13], since the most frequent bearing failure can be significantly reduced by regular maintenance, online monitoring, and replacement, whereas the controller and sensor faults are less likely.

1.3 Design Requirements of Fault-Tolerant Machine Drives

The requirements for the fault-tolerant systems in distributed power systems have been investigated in [10]. The methodology for fault-tolerant electrical drives follows relatively similar principles. The principal guideline is one fault in the system should be isolated in a subunit and has limited effect on the remaining healthy part, which can be in place to maintain uninterrupted operation. As extensively discussed in literature, four design criteria for fault-tolerant electrical drives are summarized.

1. Partitioning and Redundancy: A fundamental specification for fault-tolerant system is that a single fault would not disrupt the whole system. Therefore, the fault must be confined to a relatively independent subsystem. This implies the system...