Reliability-Oriented Design of Electrical Machines: The Design Process for Machines' Insulation Systems MUST Evolve

As the world of transportation keeps moving toward greater electrification, the main design objectives in terms of power density and efficiency of system components are becoming increasingly important. Meanwhile, transportation applications are also very safety critical. The extra stresses being seen by important components, such as electrical machines (EMs) and their insulation systems, to achieve the required performances, are accelerating the degradation of components, making them less reliable and shorting their lives. In general, lifetime consumption and degradation of components, such as for insulation systems of EMs, is still assessed through statistical methods (i.e., recording the number of imposed cycles until failure of a built prototype). This is a very timeconsuming and expensive process. In the mobile applications, such assessments can be the major bottleneck in the development timeline of an electrical product, especially for certification.

statistical methods (i.e., recording the number of imposed cycles until failure of a built prototype). This is a very timeconsuming and expensive process. In the mobile applications, such assessments can be the major bottleneck in the development timeline of an electrical product, especially for certification.
This article argues for the development of new processes, based on comprehensive physics-of-failure (PoF) methodologies, that will enable the EM designer to make reliability considerations for insulation systems of electrical machines, a main design objective, right from the very start of the design process. We investigate and describe how this advanced philosophy will enable manufacturers to design insulation systems without relying on the outdated, traditional safety factors and overengineering philosophies.

EMs Today: Growing Markets, Growing Pressures
As efforts intensify to reduce emissions [1] and address climate change [2], a push is on into research and development related to the electrification of transportation. Areas of special interest include the more electric engine and general electrified traction and drives [2]. Various international transport agendas from all of the major authorities across the world have identified electrification of transport as a key technology area. Such agendas include those related to ongoing research strategies, such as the Advisory Council for Aviation Research and Innovation in Europe's FlightPath2050 [3], which falls under the European Union's umbrella, the U.K.'s Low Carbon Vehicle Partnership's transport roadmaps [2], [4], various NASA technology roadmaps [5], [6], and several Chinese agendas, including the Chinese government's 12th and 13th five-year plans [7].
In this context, EMs represent a key component [8]. Considering the automotive sector, which has been the precursor of most modern electrification programs, the role of the EM is already critical, where today various marketready examples already show how high power density, traction EMs are replacing internal-combustion engines.
With the more electric aircraft (MEA) initiative, an ever-increasing number of aircraft systems are now being operated through electrical systems [9], [10].
The magnitude of the opportunities related to higher-performance and more-reliable EMs for transport applications can be easily observed from the predicted market growth in areas related to transport electrification. It is expected that approximately 40% of all vehicles globally will be electrified by 2025 [1]. In China alone, the market growth of plug-in electric vehicles (EVs) increased by almost 2,000% between 2009 and 2015, while the Chinese government expects a nationwide EV presence ranging upwards of 5 million EVs by 2020.
From an aerospace perspective, there is a growing concern (and related actions and reactions) related to predictions about jet-fuel emissions. To address this, it is estimated from ongoing market and research studies that electrification activities in aerospace in the next 10 years will boost overall MEA activities by approximately 50% [11]. In aerospace applications, the future global market for electrification and electrical drives (including power electronics) in the civilian area is expected to grow by US$1.78 billion each year [12]. In all these applications, high power (and torque) density (kW/ kg) and efficient EMs are a primary requirement, as these requirements enable and underpin better system-level operation, such as less fuel consumption and therefore fewer emissions.
However, a main constraint of these EMs is reliability. Transport applications require systems that are fail-safe, robust, and reliable. For EMs, it can be argued that the two major bottlenecks of failure are 1) the bearings and 2) the insulation systems of the windings [13]. Assuming appropriate scheduled maintenance and replacement procedures [14], i.e., if the bearings are adequately lubricated, then the main source of failure in an EM becomes the insulation system for the windings [15].
The reality, however, is that the more these EMs are pushed in terms of power density, the greater is the probability of failure of their insulation systems. The most common way to improve their power density is by increasing the electrical stresses in these machines, i.e., either increasing the current density and/or using higher switching frequencies for higher speeds. These electrical stresses automatically evolve into thermal stresses. If the thermal capacity of these insulation systems becomes inadequate to cope with the required operational performances, then the machine is bound to fail very quickly.
Meanwhile, regardless of what its power density is, an EM still needs to be certified according to international standards and procedures. Such EM standards as the International Electrotechnical Commission (IEC) Standard 60505:2011 [16] and the IEC Standard 60034-18-41:2014 [17] set the context in terms of the evaluation and qualification of insulation systems. When developed for transport applications, these EMs also need to comply with such standards as the Military Standard (MIL-STD)-704F [18] and the requirements contained in D0-178C [19].
All this indicates that while it is true that power density and efficiency improvements are essential, even the best-performing EM (from an operational perspective, such as power density, torque-speed profile, and efficiency) will never be certified for a transport application without the appropriate reliability and robustness. However, the certification processes in use today, which rely on statistical methods based on accelerated lifetesting of finished specimens, are very expensive and time-consuming and therefore are fast becoming unsustainable. A new approach is required.
This article seeks to challenge the long-standing processes of how EM insulation systems are designed and implemented and also how EMs are evaluated and certified. We contend that a radical step-change in how reliability in EMs is assessed and certified is necessary. This paradigm change represents an absolute novelty in the EM panorama, whereas for power electronics [20], the transition toward better reliability estimation is already ongoing. We propose the use of advanced methodologies that address this much-needed direction change and develop a coherent argument as to why we believe that the proposed philosophy needs to become the norm across the whole field of EMs soon.

Reliability and Certification of EMs
For the development of insulation systems of EMs, there are currently two main challenges that demand urgent attention. These are 1) the reliance on overengineering to design robust systems and 2) the outdated methodologies by which EMs are assessed and certified.

Design of Insulation Systems of EMs Today
EMs are generally designed to comply with a preset minimum machine winding life. Most industrial EMs are forecasted to survive and operate for approximately 20-40 years. In transport applications, EMs are never expected to operate for more than five years. These life expectancies are commonly calculated based on the projected lifetime of the EMs' windings. Traditionally, the insulation system design, to accomplish a specified life target, was largely achieved by trial and error. Thus, if a winding design failed, then the EM designer simply added extra insulation in the next iteration or version of that design.
Today, to prevent premature failures, heavy safety margins are added immediately to the design, resulting in insulation systems that highly exceed their actual operational requirements. However, this also means that safety margins are excessive and can greatly increase the mass and the cost of the EM. This becomes very important when considering the goal of lighter and more cost-effective machines and systems as required for future aircraft and EVs.

Certification of Reliability of EMs Today
In terms of assessment and certification, EMs are still assessed through statistical methods (i.e., recording the number of imposed cycles until failure of a built prototype). These methodologies, which have been the same since the 1960s, require a sample of the finished product, which is subjected to various stress-inducing tests with failure times recorded. This is obviously a very time-consuming and expensive process. For conservative areas, such as the transport field, these requirements represent a major bottleneck in the development timeline of an electrical product. Figure 1 shows a typical development cycle of an aerospace EM today, based on information gathered by the authors from working with a leading tier-1 aerospace machine manufacturer. The imbalance between the time required (and therefore cost) for design and for reliability certification can be clearly observed.

Moving Forward
A different approach is necessary for the next generation of EMs. Reliability needs to become a design objective right from the start of the design process of EMs. This aim can only be achieved through the use of a new approach and methodology that combines accuracy with time efficiency.
We believe that the only feasible manner in which this can be achieved is by implementing comprehensive PoF methodologies that will enable the EM designer to consider reliability and lifetime factors from the very start of the design process. This methodology is based on the development of advanced aging models that can accurately estimate the lifetime of the winding insulation, assuming that the stresses applied can be quantified and reproduced for validation purposes. Such lifetime prediction models are a type of aging model that for particular inputs will inform the user of the number of cycles to failure of that insulation system.
We contend that the model must be based on a real understanding of the deterioration mechanisms of the insulation system. A real knowledge of the chemical, material, and physics phenomena happening in the insulation for particular operation cycles is crucial. Only then will these models be able to predict the end of life of the insulation system without having to rely on the traditional methods.
A conceptual representation of such a model is given in Figure 2. The model must be able to define the end of life of the product with a failure probability distribution curve, such as that shown in Figure 2. The goal of the model is to predict the moment of the failure with a reduced variability than commonly expected today. This is drastically different from traditional approaches that only define a mean time to failure without really considering any variability of the exact moment of failure. The shortcomings of the commonly applied reliability-handbook techniques are very well described in [21].
The model proposed in Figure 2 can then be seamlessly integrated into the design process of EMs. A conceptual  representation of a potential process (one of many hopefuls that can be envisioned) is shown in Figure 3, where the lifetime prediction model is providing direct input to the design of the machine. This contrasts with the traditional method of first designing the machine and then simply checking that the machine survives for some set conditions. Here, reliability is considered right from the start and is actually an input to the design process.
Currently, this type of philosophy is practically nonexistent in the standard design processes of EMs. It is perceived that if fully implemented, this methodology will result in 1) significant reductions (and ultimately the removal) of the need for overengineering and the use of the traditional safety factors, thus leading to improved power densities and cost-effectiveness 2) very important reductions of the time required for assessment and certification of EMs, thus shortening the time needed to develop products. The aging models described here are very different from the failure-prediction methods commonly used today across the world of science and engineering. The common fault-tree methods are based on techniques called the probability density function and the cumulative density function of the potential failure happening. These methods are very simplistic functions that go back to the 1950s and do not consider the development of new technologies.
These assumptions represent a very crude approximation of the reliability and cycles-to-failure problem. Thus, to be successful, any process based on the use of the concept given in Figure 3 requires modern and accurate aging models. As a result, we feel that such processes can only be beneficial if the proposed use of the model in Figure 2 is based on more advanced methodologies, such as detailed PoF techniques.

Understanding Today's Methods and Their Limitations
The statistical methodologies described in the section "Reliability and Certification of EMs" rely on the availability of a set of completed products that can then be subjected to accelerated lifetime testing, usually by implementing harsh testing regimes on these prototypes until they reach a set failure point. The time to reach this failure point is called the lifetime of the product. While this process is still the most common methodology in use today, it is also known to be very unreliable and inaccurate [21].

Calculating Failure Today: General Components
In more practical terms, i.e., when any particular product is working in its intended normal operating environment (not subjected to abnormal operating conditions, such as for accelerated lifetime cycling), the lifetime of that component is usually defined as the period of time from the moment when it comes off the production line to the exact time that it fails to produce its required performance. This time window is, of course, dependent on the particular system application and on the environmental conditions to which it is exposed. During this time, the degradation of the components can be said to depend only on the system and environmental loads.
As mentioned previously, most engineering products have traditionally been assessed for robustness through statistical methods, i.e., by recording the number of imposed cycles until failure of a built prototype. Over the last two or three decades, manufacturers have started to realize how costly and time-consuming these  old methodologies are and have tried to investigate more modern ways to understand the real degradation phenomena. Thus, much effort has gone toward improving methods that typically aim to correlate the statistical data from the traditional tests to slightly more advanced degradation models.
However, even the more advanced of these methods still rely heavily on statistical methods, where the finished products (or specimens of them) are put under multistress testing with accelerated lifecycles. The specimens are then tested and categorized according to their state of health after every test. A widely adopted statistical tool for the reliability assessment of electrical insulation (as well as for roller bearings, electronic components, ceramics, capacitors, and other components) is the Weibull distribution [22]. The popularity of the Weibull distribution is mainly due to the following: 1) its flexibility in modeling either increasing or decreasing failure rates 2) its ability to fit data coming from accelerated lifetime tests in a much simpler manner [22], [23] than with exponential, normal, and lognormal distributions 3) its requirement that only a small number of samples need to be subjected to accelerated aging tests, without compromising the final result's accuracy 4) its ability to allow data censoring. Equation (1) describes the Weibull cumulative distribution function, where a is the scale parameter, t is the time to failure, and b is the shape parameter. In particular, a is the 63.2% percentile of failure times, while b is inversely proportional to the data scatter, which is directly correlated to variance in failure times: Calculating Failure Today: Machine Insulation Systems EMs are found in various applications, including those related to transportation, home appliances, and industry [24]. Machines employed in transportation are particularly demanding in terms of requirements for insulation to withstand high stress levels, including electrical, thermal, mechanical, and environmental stresses. For these applications, EMs usually operate with a voltage level below 1,000 V (i.e., lowvoltage EMs) and generally adopt a random wound layout, using round enameled magnet wires [13]. The inter-turn insulation is provided by a thin enamel layer, e.g., a type-I insulation (organic chemical composition). A number of stresses are responsible for the main aging in electric transportation applications, where a significant level of power density is often required [25]. The machine's power-toweight ratio is normally limited by the (maximum) stresses at which the insulating materials can operate [26], [27]. At the same time, EMs for transportation applications need to be especially reliable throughout their whole life. The premature failure of low-voltage EMs is generally originated by excessive aging and degradation of the turnto-turn insulation [28]. Therefore, a precise knowledge of the turn-to-turn insulation deterioration level can enable predictions to be made about an EM's time to failure. Some of the common nondestructive tests are those involving insulation capacitance (IC), insulation resistance [23], and partial discharge (PD) detection [29].
For medium-voltage machines, comprising mixed organic-inorganic insulating materials (i.e., type II), several tools can be employed for predicting the insulation's aging/deterioration level [13], [30]. These techniques can be either offline or online. Among the offline strategies, the tan d (or dissipation factor) measurement is frequently implemented for machines and power transformers [31]. If properly implemented, the dissipation factor test can give precise information regarding the insulation status [32], as it can accurately indicate if the insulation is subject to the inception of PDs.
If suitable sensors are adopted (e.g., antennas), the online PD test can detect the occurrence of PDs in the interturn insulation [33]. Nonetheless, such measurement does not provide information useful for lifetime assessment/ modeling, as the PD inception generally represents the end-of-life point in type-I insulation [34], [35]. In [36], the aging level of a traction motor is assessed via a nonintrusive, online method. The analyzed specimens are twisted pairs aged at a constant temperature in a ventilated oven. A similar study is proposed in [37] for medium-voltage EMs. In this case, the aging (due to electrical and thermal stresses) of the main wall insulation is linked with variations of IC, dissipation factor, and PD patterns. The thermal aging influence on the IC is also verified in [38], where the samples are twisted pairs and the thermal degradation is associated with the decrement in partial discharge inception voltage.
All of these excellent advances were made in the name of understanding the degradation phenomena of EM insulation systems. In fact, the last two decades have resulted in much data and methodologies applicable for this area. However, the main challenges remain that these methodologies are still focused on only one particular stress at a time and that the information being achieved is usually not fed back into the development process to inform a lifetime prediction model.

Advancing the State of the Art
Novel multistress models that can estimate the lifetime of an EM's insulation when simultaneously exposed to several stress types can be proposed and investigated. With an adequate implementation of these models, an EM can be designed not only in terms of its performance requirements but also considering the associated reliability and lifetime aspects.

Example of a Lifetime Prediction Model, Developed for a Particular Scenario of Low-Voltage Machines
Low-voltage EMs employed for transportation applications are typically machines designed for high-power/torque performance operating under harsh, highly variable, duty cycles, where the main aging factors that need to be considered are those stemming from thermal and thermomechanical stresses. In fact, type-I-insulated EMs (i.e., organic) are designed to be exempt from PDs (i.e., PD-free) throughout their whole lifetime, and thus electrical stress does not contribute to the insulation aging [35]. For short-time duty applications, the winding temperature of the EM typically varies between ambient temperature and a maximum temperature reached at the end of the loading period, without ever reaching a steady-state value.
Since the aging factors considered in such applications are very temperature-dependent, the main input required by any developed lifetime model is the winding (hot-spot) temperature profile ( ) t i as a function of time .
t The lifetime of the insulating material under thermal aging at constant temperature i can be modeled by (2), where ( ) L i is the insulation lifetime under thermal stress at temperature , i L0 represents the lifetime at a reference temperature , 0 i and the parameter B is related to the activation energy of the degradation process: This model [39] relates the thermal aging of the winding to the rate of a temperature-dependent chemical reaction through the use of the Arrhenius equation. To consider the effects of thermal cycling (i.e., temperature variation), the Arrhenius-Dakin law, as given in (2), must be combined with Miner's cumulative damage law [40]. The insulation loss of life during a single activation cycle, i.e., temperature profile depending on the machine's mission, here called , LFcycle can be calculated as in (3), where tcycle T is the time duration of the temperature profile, and ( ) t i is the vector temperature versus time: Therefore, according to Miner's law, the predicted number of cycles to failure K can be calculated as in (4). Alternatively, the total lifetime , Ltot in a unit measure of time (i.e., hours, days, and so on), of the cyclically stressed insulation can be calculated as in (5) These expressions in (3)-(5) were experimentally proven in [41] and [42] and define a direct relationship between a number of cycles of a particular mission applied to the machine's insulation and the predicted remaining cycles to failure for that insulation.
These simple yet elegant expressions can, therefore, become the key to all of the reliability aspects that the designer needs to consider for a short duty cycle, PD-free EM. This emphasis on these conditions is important because, for the model example given here in (3)-(5), only the thermal stress is being considered, i.e., it is a single-stress model. This is acceptable because, for these conditions, the thermal stress is the predominant aging factor. However, the very nature of the models being discussed here allows for them to be easily extended into multistress models that will then include also electrical, mechanical, and environmental stresses.
Depending on the steepness of the temperature profile (i.e., the temperature gradient), the proposed model might also need to include an additional corrective factor. This is necessary to consider any potential, thermally induced mechanical stresses on the insulation, due to mechanical shear between the conductor surface and the insulation. Such a model has been also proposed by the authors in [28], but is omitted here for space constraints.
The Arrhenius-Miner model, as de scribed by (3)-(5) provides a conservative estimation of the insulation lifetime, as the model's input is the hot-spot temperature. In other words, it is assumed that the whole EM winding volume is subject to the (uniform) hot-spot temperature. A more precise (but less conservative) estimation can be carried out by evaluating the insulation lifetime relative to each winding portion (e.g., active/slot and endwinding regions), although this would require 1) the precise knowledge of the winding arrangement/layout and 2) the volumetric temperature distribution. In light of this, the adopted methodology can be said to represent a fair compromise between computational effort and lifetime prediction accuracy.

Using the Lifetime Prediction Model for Design
It is clear that the model described by (3)-(5) is basically the theoretical definition of the model's conceptual representation, illustrated in Figure 2. Such an Arrhenius-Miner model can then be used for extracting the Weibull probability density function of time to failure. This process is straightforward, as it just requires the model to be tuned for providing the lifetime relative to the 63.2% percentile of the Weibull distribution, i.e., a in (1), while the shape factor, i.e., b in (1), is essentially the slope of the insulation's thermal endurance curve (obtained by accelerated lifetime tests or provided by the insulation manufacturer) [42].
By applying this concept to the design/development process shown in Figure 3, very fast qualification of EMs' insulation systems will be possible. This will allow the machine designer to omit the traditional safety factors that were historically used to ensure a long life of the machine. The machine designer now will know the cycles to failure of his/her insulation with a muchincreased confidence and will design the EM without having to overengineer the product, allowing for savings in the EM's mass, cost, and certification time.

The Scientific Meaning of Such Methodologies
The proposed methodologies depend on accurate estimations of a machine's lifetime. Considering, once again, the example model described in the "Advancing the State of the Art" section, certain parameters need to be found through experimental testing of specimens to achieve the model described by (3)-(5) and conceptually shown in Figure 2. This method of combining knowledge of degradation mechanisms with data taken from experiments is the basis of the so-called PoF method.
In more practical terms, the lifetime model described by (3)-(5) is based on the block scheme illustrated in Figure 4. From the material characteristics and the mission profile of the drive system (i.e., working cycle, environmental conditions), an accurate analysis of the degradation mechanisms is made. Then, with appropriate lifetime prediction models and the data of the intensity of the degradation mechanism integrated with the information gathered from the sensors, an appropriate lifetime prediction can be made. This prediction can be used to validate existing designs or reach a predetermined requirement of lifetime. A perceived added advantage of the PoF method is that prognostic and health-monitoring procedures can then be used to assess the status of the system when in operation.
Thus, the use of an accurate tool, such as that shown in Figure 2, will result in a step change in the centuries-old development processes for EMs. The model addresses the aforementioned two main challenges and can result in significant advantages, as follows.

Reducing the Weight and Cost Of EMs
Today, it is clear that excessive safety margins in the design of the insulation systems of EMs can increase the mass and the cost of the machine. For example, it was indicated in [43] that a 20% reduction in the slot-liner insulation thickness in the stator of a large EM will allow the slot area to be reduced. Depending on the speed of the EM, the weight of the stator core can then be reduced from 23% to 13%, the copper is decreased from 64% to 50%, and the mass of the insulation is cut down from 57% to 12%. As the price of an EM is very dependent on the mass of these materials, then the cost is also significantly reduced. However, it is important to note that the example given in [43] is not so recent (from 1999) and is still loosely based on a trial-and-error system. Through the proposed use of the models discussed here, much more significant improvements can be achieved.
Recent literature [44] shows that if an aerospace EM was to be designed by implementing the proposed process of Figure 3, then the total volume and mass of the EM will be decreased, when compared to the same EM designed in the traditional manner. This is reflected by the main outcome of [44], which is illustrated in Figure 5, showing how much smaller a machine can be made (for the same power rating) when the introduced process is used [ Figure 5(a)] compared to a machine made by traditional processes ( Figure 5(b)].
Thus, smaller machines for the same power rating (improving the power density) without compromising their reliability values can be achieved. This is done by removing the need for safety factors in the design. In [44], a 20% power density (kW/kg) improvement has been achieved.
It can also be argued that such a decrease in mass and volume will result in a reduction in cost. For the case given in [44], a cost reduction of at least 15% when compared to the original machine was calculated. However, this needs to be taken as a relative value. It is very difficult to predict costs accurately, as the price of raw materials and of manufacturing processes fluctuates depending on availability and other factors.

Reducing the Development Lifecycle for EMs
The second main perceived benefit of these methodologies is the potential of reducing the whole development lifecycle time. To be compliant with standards,  such as the MIL-STD-704F [18] and the D0-178C, machine developers are usually required to plan for long periods of reliability assessments and certification. A typical development cycle for an aerospace EM is shown in Figure 1. As mentioned, these long periods are mainly due to the need for a finished prototype to be tested for long periods of time after it's been manufactured. The lifetime prediction models being discussed in this article will allow for reliability assessments to be done in parallel with the development of the product. In other words, the accelerated life testing required for reliability assessments is done in various stages as part of the PoF scheme shown in Figure 4, as opposed to the traditional manner of simply cycling a finished product. This allows for the development lifecycle of an EM to be reduced significantly. Working with a leading tier-1 aerospace EM manufacturer, which cannot be named for confidentiality reasons, we have performed an investigative study in an attempt to quantify this perceived potential gain for a particular EM range. The main outcome of this study is illustrated in Figure 6. This work was performed on low-voltage aerospace machines. However, the result shown in Figure 6 has been generalized to address a wider range of machines than those analyzed with the industrial partner.
Comparing Figure 6 with Figure 1, a considerable improvement in terms of the time required to launch a new product on the market can be achieved. Through the proposed use of the methodologies discussed in this article, it is predicted that the low-volume production period of a new product can start almost two years earlier than it would if using today's quasi-obsolete development procedures.

Conclusion
As the world shifts to greener and more electric solutions for transportation applications, the role of all electric systems, including that of the EMs, will become more important. This is reflected and can be observed in the huge market growth for the aerospace and automotive industries, as discussed in the section "Challenging How EMs Are Evaluated." All this growth depends, however, on the industry's capability to remain sustainable and commercially effective in all aspects of the development processes related to EMs and associated systems. The methodologies discussed in this article can propel the industry toward a much-needed step change in how EMs are designed, developed, certified, and brought to market.
The era of overengineering, trial and error, and specifying lifetime through accelerated destruction of completed prototypes is over. Today, every minute and penny counts. Thus, the EM manufacturer that invests in these advanced methods now, will, in a few years, be able to put EMs on the market that are lighter, more cost effective, and in a much shorter time. This will enable these manufacturers to gain an extremely important competitive advantage over the rest of the players in the field.   He is a member of the IEEE Industrial Electronics Society and a Student Member of the IEEE.

Biographies
Giampaolo Buticchi (giampaolo .buticchi@nottingham.edu.cn) earned his M.S. degree in electronic engineering in 2009 and his Ph.D. degree in information technologies in 2013 from the University of Parma, Italy. In 2012 he was a visiting researcher at the University of Nottingham, United Kingdom. He is currently an associate professor in electrical engineering at the University of Nottingham Ningbo China and the head of power electronics of the Nottingham Electrification Center. He is the author/coauthor of more than 180 scientific papers and an associate editor of IEEE Transactions on Industrial Electronics and IEEE Transactions on Transportation Electrification. He is the chair of the IEEE Industrial Electronics Society Technical Committee on Renewable Energy Systems and a Senior Member of the IEEE.