Proyecto de Pavimentos Sector Aeronáutico 158

Apparently, the most established brushless wind generators are the SCIGs, Li, Chen and Polinder: 2006 [15]. This constant speed system is already discussed in subsection 1.3.1. They were eventually replaced by DFIGs, which though are not entirely brushless in nature, but are able to produce higher energy yield due to a wider operating speed range.

Also, PMSGs as discussed in the preceding subsection can be considered as brushless; but be- cause the PMs are positioned internally within the rotating member, they become vulnerable to tem- perature rise and certain motive forces, which can result in demagnetisation and mechanical stress, respectively. Besides, the high cost of high–energy PMs is another major disadvantage of PMSGs in spite of remarkable benefits such as potentially reduced generator active mass, as well as capabilities for high power density and efficiency, Cao, Xie and Tan: 2012 [4] and Madani: 2011 [26].

Switched reluctance generators (SRGs) are next in line to be considered as brushless wind gen- erators, and the first to be highlighted as stator–active machines with a robust rotor structure. In addi- tion, because they have both salient rotors and stators, they are also called double salient machines. SRGs are simple to construct (bodily), mechanically robust and can withstand high temperature con- ditions, but they produce loud acoustic noise as attributed in Mademlis and Kioskeridis (2005) [33] and Ahmad (2010) [34]. Besides, they have a complicated control strategy in that, “they obtain their excitation from the same voltage bus that it generates power to”, citing Husain, Radun and Nairus (2002) [35]. Karthikeyan (2009) [36] suggested that the suitable drivetrain topology for SRGs are LS drivetrains; this, of course, would make them very bulky and expensive for WPG systems.

In Potgieter’s (2014) [85] thesis, some details on the emergence of non–conventional wind generator concepts is provided. Eventually, one such candidates, the slip synchronous permanent magnet generator (SS–PMG) design, was preferred for his study because, according to him, it has improved simplicity, robustness, reliability and grid compatibility compared to most conventional wind generator systems. However, the researcher is aware that SS–PMG’s are usually configured for fixed speed wind turbine systems, coupled with a complex and bulky architecture, which is acerbated by the assembling of two direct–drive generators, among others.

Wang and Gerber (2014) [86] investigated on the use of magnetic geared PM (MGPM) ma- chines for wind generator drives. Their findings, among other things, show that such generator topol- ogies have potentially higher efficiency, better thermal management and higher reliability compared to conventional geared wind generator drives. However, the complexity and PM–reliance of such systems is again bound to increase their manufacturing costs.

Yet, there are new and emerging stator–active brushless wind generator concepts such as the PM flux switching machine (PM–FSM) studied in Ojeda et al (2012) [37]. PM–FSMs are not alone in this category, as others such as PM flux–reversal machines (PM–FRMs) and the PM doubly salient machines (PM–DSMs) have been similarly designed for wind energy applications in Saou, Zaïm and Alitouche (2009) [38] and Fan, Chau and Cheng (2006) [39], respectively. These categories of brush- less machines have similar features which distinguish them from other machines such as, a robust

rotor structure, ease for thermal cooling due to stator–mounted components and the capacity to de- ploy less copper usage due to a concentrated winding configuration, among others.

Interestingly, PM–FSM is being resurrected after it was initially invented by Rauch and John- son (1955) [40], which unlike the PM–DSM was first devised by Liao, Laing and Lipo (1995) [41] and the PM–FRM by Boldea, Serban and Babau (1996) [42]. These brushless machines tend to site their active components (be it PM blocks or DC coils), as well as the phase windings, mainly in their stators as similarly contrived in SRGs. Thus, they result in less cumbersome rotor structures as de- picted in Fig. 1.7.

Different from SRGs, their field sources are physically distinguishable by either PMs or field coils in the stators, making their design less complicated. Considering that cooling measures are also easier to deploy, they are further preferred for HS fault–tolerant applications. Also, they have very good flux weakening8 capabilities, which is mainly beneficial in automotive applications, Fasolo, Al- berti and Bianchi: 2014 [69], Zhu, Shen and Howe: 2006 [133] and Štumberger et al: 2006 [134].

An overview on these stator–active PM machines is already provided in Hua et al (1999) [43] and Cheng et al (2011) [44], hence the researcher is not going to go on and on about it. In Hua et al (1999) [43], it was simply about PM–FSMs––their features and what makes them interesting for a number of application needs; their different excitation modes such as PM–FSMs (only PMs), WF– FSMs (only wound–fields) and HE–FSMs for hybrid–excited systems (both PMs and wound–fields); their electromagnetic performance; as well as their different stator and rotor topologies.

In Cheng et al (2011) [44], the discussion was generally on all three stator–active machines, with focus on their concepts, operating principles, machine topologies, electromagnetic performance, etc. They were able to provide a summary on the advantages and disadvantages of all three machines, based on a qualitative comparison, replicated here in Table 1.3 for emphasis.

Fig. 1.7. Cross–sections of stator–active machines9_{: (a) PM–DSM, (b) PM–FRM, and (c) PM–FSM.}

8 _{Flux weakening is a control strategy usually employed to extend the operating speed range of electric drives, while maintaining constant power. In}

cases of coils e.g., WRSGs, it is achieved by controlling the field current magnitude to adjust the back–EMF which is speed dependent, while in other cases, e.g., PM–FSMs, it is characteristically achieved by forging a compromise between the mechanical torque and output power, when the speed is adjusted.

Table 1.3. Comparison of different novel stator–PM machines [44]

Design issue PM–DSM PM–FSM PM–FRM

Phase flux Unipolar Bipolar Bipolar

Energy conversion loop 1st and 2nd quadrants All four quadrants All four quadrants

PMs location Stator back iron Sandwiched by stator teeth Stator teeth surface

PMs consumption Low High Medium

EMF waveform Trapezoidal Sinusoidal Trapezoidal

Cogging torque Low High Medium

Torque density Low High Medium

Control mode BLDC BLAC BLDC

It is clear to see from Table 1.3 that, PM–FSM provides the best qualities in terms of torque density and bipolar flux linkages among the rest, thus making it a promising candidate for the pro- posed geared MS wind generator drives. Little wonder, therefore, why Cheng et al (2011) [44] con- cluded their appraisal with the prediction of a bright future for PM–FSMs in areas such robotics, au- tomobiles, wind energy drives, to mention a few.

Just like in Cheng et al (2011) [44], Zhang et al (2009) [45] also embarked on a comparative study on all three stator–active PM machines but this time, it was based on the formulation of their respective general power equations. After establishing the equations, which were used for the com- parison, they equally confirmed in their findings that the PM–FSM option possesses the highest pow- er density among the three machines.

But, as highlighted in Table 1.3, a major setback in PM–FSMs is their characteristically high cogging torque and torque ripple, which is due to its double salient structure. Even when compared to other conventional machines, they do not fare any better in this regard, according to the studies in Hua et al (2008) [46] and Pang et al (2007) [47]. Shao et al [103] provided in tabular form, a compar- ison of stator–PM machines e.g., PM–FSMs, with typical rotor–PM machines, reproduced herein as Table 1.4. Clearly, it can be further seen that the airgap field harmonics in PM–FSMs is very high, another factor usually attributed to the effects of high cogging torque and torque ripple in PM–FSMs. To this end, numerous studies have been initiated in an attempt to address the problem of high cog- ging torque and torque ripple in FSMs, with some reasonable success already established.

For instance, Sikder, Husain and Ouyang (2015) [48] used flanges together with rotor pole shaping to reduce cogging torque, while Wang, Wang and Jung (2012) [49] employed rotor teeth notching schemes but at a slight detriment to the average torque. Also, Xu et al (2011) [50] applied several techniques such as rotor pole–pairing, rotor pole–notching, rotor pole–chamfering and rotor pole–skewing, just to ensure that they topple the problem of cogging torque in PM–FSMs, once and for all. On the other hand, Xu (2014) [51] proposed a novel decoupling model–based predictive cur-

rent control algorithm to diffuse torque ripple effects as against popular techniques which are based on electromagnetic optimal design, whereas Hwang et al (2016) [52] employed the winding function theory on the back–EMF and MMF airgap harmonics in their attempt to diminish torque ripple in PM–FSMs, to mention a few.

Again, another debilitating problem in PM–FSMs is their high PM utilization as also implied in Table 1.3, which can greatly increase their cost, when designed using very expensive PM materials like rare–earth PMs. One way to tackle this problem is by employing rare–earth–free materials like ferrite PMs or wound–fields. The feasibility of using rare–earth–free excitation sources is suggested in the light of the following:

 PM–FSMs have flux focusing10 _{capabilities as mentioned in Hua et al (2005) [53], which can} be exploited by the use of lower performing rare–earth–free materials such as ferrite PMs or wound–fields, leading to a more economical design at reasonable performance constraints.  Although with higher demagnetisation risks compared to rare–earth PMs, ferrite PMs used in

PM–FSMs are to come without any fear of demagnetisation during normal operation, due to a special magnetic circuitry produced in PM–FSMs, which constructively align the field source MMF and the stator current MMF, Amara et al: 2005 [54] and McFarland, Jahns and El– Refaie: 2014 [55].

 PM–FSMs are low maintenance machines due to their brushless nature. Hence, the problem of brushes and slip rings, in the case of their wound–field designs, are totally eliminated.

Table 1.4. Comparison between PM–FSM and rotor–PM machines [103]

Design issue PM–FSM Rotor–PM

Airgap field harmonics High Low

Back–EMF harmonics Low High

Simple and robust rotor Yes No

Simple stator No Yes

PM temperature management Easy Difficult

Copper slot area Small Large

Copper loss High Low

Saturation Heavy Light

Overload capability Poor Good

Stator outer flux–linkage Yes No

PM volume High Low

10 _{Due to doubly salient structure and smaller overlapping areas between the stator and rotor teeth, as well as the position of field excitation (usually}

PM) sandwiched in–between ‘U’–core laminations in the stator of FSMs, the flux linking the rotor and the stator teeth tend to saturate, resulting in high airgap flux density. In Kim et al (2005) [135], it is called spoke configuration which is meant to mount magnets in PM machines such that magnets with poles of the same kind are placed facing each other, with alternate magnet poles in–between. This allows the flux from a large area of PM to be directed into the rotor pole, that is why it is known as flux focusing or flux concentration.

 PM–FSMs being stator–active machines, does present an easier option for cooling of the field sources in instances with high operating temperature concerns.

Considering these important qualities which have triggered enormous traffic in the study of PM–FSMs, the researcher became quizzical and proceeded to perform an assessment on the state–of– the–art on PM–FSMs in relation to wind generator drives. The next section is thus dedicated to this inquiry and the outcome thereof. The operating principle of the flux switching machine, which is yet to be discussed, should the reader be interested, is referred to in Appendix A1.