Power Density as the Key Enabler for Electrified Mobility

Electrification for mobile systems is a long-term trend that touches every mode of transport we currently use. As these systems proliferate, and increase in power levels, there is a strong desire to increase the power density of the electrified systems on board. A key barrier to increased electrical power density is the associated thermal management that is necessary to prevent failures. This article provides an overview of a long-term effort aimed at integrating the research and education of combined electrical and thermal systems for the sake of increased power density. An overall motivation for this effort is provided in order to gain context. Then, specific examples are provided of how research and education can be performed within graduate programs across the United States (U.S.) and the world.

We live in an increasingly electrified world. For stationary applications such as industry and manufacturing, this statement has been obvious since the start of the twentieth century as steam and belt drives in factories gradually gave way to electric motors for machining, conveyor lines, and all manner of other industrial applications (Atkeson & Kehoe, 2001). For domestic stationary applications, modern conveniences blossomed as electrification grew starting in the middle of the twentieth century (Gordon, 2016). Lighting, air-conditioning, cooking, and cleaning, as well as many types of in-home entertainment were fueled by growing abilities to provide relatively cheap electrical power over long distances. Now, a fifth of the way through the twenty-first century, we are seeing electrification rise in the mobile domain. The progress has been steady for several decades but it is really during the past several years that electrified mobility has seen an almost explosive growth at the level of individual consumer. What is most interesting is that this growth cuts across different modes of mobility. Using an aircraft example, electrification has been increasing exponentially both in the commercial and military domains as shown in Fig. 1 (Williams, 2017).
Whereas much of the auxiliary componentry on aircraft used to be performed by mechanical means, including pneumatic or hydraulic power, increasingly the auxiliary systems are becoming electrified. Bleed air systems to run air cycle machines are being replaced by high-speed electric motors. Hydraulic actuators for flight surface control are being replaced by electric drives, or at least electro-hydrostatic actuators. The insatiable consumer appetite for communications and infotainment are leading to increased loads for networking within the aircraft as well as communicating to ground or satellite stations. Some of the benefits of electrified auxiliaries are increased efficiency, reduced emissions, and potentially increased long-term reliability (Sarlioglu & Morris, 2015). All this led to the more electric Boeing 787, the most advanced commercial aircraft, which boasts a megawatt of power coursing through its infrastructure.
While aircraft have a long and steady growth rate, there is an event horizon that is upcoming. Manufacturers and government are now looking past auxiliaries and subsystems as the sole domain of electrification and are focusing on the prospect of electric propulsors taking the place of the venerable gas turbine turbo-fan engine (Henke et al., 2018). Efforts under investigation include both battery-driven systems without any combustion engine as well as hybridized power trains where the engine provides electrical power to run the propulsor fans. These will take the electrical power levels well above 1 MegaWatt for commercial transport.
Switching to an automotive example, the growth in the number of battery electric vehicles (BEVs) and hybrid electric vehicles (HEVs) have increased dramatically since the first Toyota Prius was released in 1997. Figure 2 below indicates a steady growth in electric vehicles in the United States (M. (. R. D. Ortiz), 2018a). This does not include hybrid electric vehicles such as the Prius which account for an even larger number of highly electrified vehicles on the road. The motivations for electrification in on-highway mobility are many of the same ones as with aircraft: lower overall cost of ownership after initial purchase, fewer parts and hence better reliability, and lower environmental impact. These have started to matter at the individual consumer level over the past 10 years which has accelerated adoption. To get the true global picture, one has to also consider the number of electrified vehicles in countries such as China which have a great incentive to reduce pollution. This is shown in Fig. 3 and the growth numbers are impressive as China has dwarfed the U.S. in electrified transport.
The results of the passenger vehicle and aerospace sector are not unique. Similar trends can be seen in delivery trucks, busses, off-highway equipment, and ships. Many of the same drivers apply here as well. For example, government regulations limit or prohibit idling of engines in delivery trucks while stopped to perform loading and unloading (D.C. Anti Idling Law, 2018). This is a natural incentive to electrify. An examination of Figs. 1, 2, and 3 makes at least two things clear. First, increased electrification in mobile transport is a long-term trend that shows no signs of reversing at any point in the future. In fact, it will likely accelerate as economies of scale drive unit prices down. Second, the level of electric power being used in mobility platforms is monotonically increasing, potentially with significant jumps as discussed with regard to electric propulsion.
There are clear and irreversible drivers governing these trends. A key question raised in this articles is BWhat are the challenges and barriers to progress faced by these systems?Î n Section 2 of this article, we make the case that one of the main barriers to progress is the challenge associated with increasing power density. Power density is arguably the most important metric when discussing mobile systems. Yet, as we will see, there are limits to increasing electrical power density. Section 3 advocates for a systems-level solution to these challenges and provides specific examples of how this systemslevel approach can bring tangible technological benefits. Since technology without people is just an abstract idea, Section 4 describes one approach to developing the right types of technologists to address these systems-level problems. Section 5 concludes the paper with some thoughts for our increasingly electrified future.

Power Density Challenges
When considering mobile systems, the power to weight ratio is the key metric that drives all other performance indicators. Increased power to weight allows for a given propulsion level to be accomplished in a smaller or lighter form factor. This results in greater cargo carrying capacity with similar acceleration performance or, conversely, greater acceleration with the same cargo carrying. For electrified systems, part of the cargo to be carried are the batteries that store energy in addition to the people and goods to be transported. Therefore, increased power density provides greater range, more passengers, and lower cost per passenger mile.
Power density can be measured either gravimetrically, as in kW/kg, or volumetrically, as in kW/l or kW/m 3 . Both have their relative merits depending on the application. However, increasing both result in processing larger amounts of electrical power in smaller volumes since weight is often tied closely with size. As electricity passes through a given volume, there is resistance which leads to resistive, or Joule heating. One way to minimize this is to utilize high temperature superconductors but this is outside the scope of consumer-ready transport. Joule heating leads to a buildup of thermal energy. Unfortunately, as a consequence of physics, it is far easier to put thermal energy into a confined volume than it is to take it out. A relatively thin wire embedded in a solid can create a significant amount of heat. It is simple to find heaters that can provide 100's of watts of heat in cubic centimeter packaging. Moving the heat away from its source in the solid requires either conduction or convection via fluid mass transport. Convection requires relatively sizeable passages to get the fluid mass close to the source which results in increased bulk and reduced power density. Conduction results in a relatively slow transport of heat away from the source with a speed governed by the thermal conductivity of material. The choices of big (convection) or slow (conduction) result in key design barriers to the increase of electrical power density with significant impact.
Heat moves from a region of high temperature to a region of low temperature in a manner related to the temperature difference between high and low, i.e., the thermal potential difference. For many materials systems, including electrical conductors such as copper, there is an inverse relationship between temperature and the conductivity of material. So, as temperature goes up, the resistance goes up, causing the temperature to increase further. One area of particular challenge is in electrical storage, or batteries. These operate on chemical reactions. If there is an exothermic reaction, it can increase the temperature of the system which can increase the reaction rate in a positive feedback. This self-heating is illustrated in Fig. 4 and results in what is known as thermal runaway. Thermal runaway was responsible for the fires in the battery boxes on the Boeing 787 Dreamliner and led to grounding of the worldwide fleet for 4 months (N. T. S. Board, 2018).
Another thermal issue with electronics results from their construction. Power electronics are comprised of multiple heterogeneous systems including metal conductors, semi-conductors, capacitors, inductors, and substrates. These all have different coefficients of thermal expansion (CTE). When two or more materials, with dissimilar CTE's, are bonded or joined, the heating and cooling of them causes stress due to a mismatch in displacement between the different materials. As these heterogeneous systems heat and cool due to the operation of the power systems, there is a mechanical stress that is cycled at the interface. Over time, and if sufficiently large, this stress leads to failure just like any other mechanical stress cycling phenomenon (Khasaka et al., 2015) (Durand et al., 2016). This reliability impact is one of the most important aspects of systems that process electrical power.
It is imperative for modern electrified systems to carefully match the thermal power management with the electrical power management. Unfortunately, too often the electrical system requirements are designed first and then the thermal management requirements are imposed afterwards. This serial approach to design is illustrated in Fig. 5 that shows two Bwaterfall^plots associated with subsystem design. The result of a serial approach is problems show up late in the design cycle where they become very difficult and expensive to fix. A result of this approach, if the design is successful, is a suboptimal design. Either the electrical system has to be de-rated or the thermal management system has to be over-designed to meet unanticipated loads.
The thermal limits for electrified systems act as a barrier or wall to increased power density. As a result, modern systems have been limited in power or energy density. Motors have been limited to approximately 5 kW/kg, excluding the cooling system. Power electronics have been limited to approximately 2-5 kW/kg for inverters depending on the level of refinement of sinusoidal signals being sought. State of the art lithium-ion batteries range between 150 and 200 W*h/kg; much of this limitation is electrochemistry but there are also significant thermal limits associated with charge and discharge rates for batteries.
What is needed is a systems-level approach to increasing power density. This means a co-design of the electrical and thermal elements from the beginning. It also means a multidisciplinary approach spanning electrical engineering, mechanical engineering, and materials science among other disciplines. Recently, a new effort was started to specifically address the systems-aspect to what is termed electro-thermal systems. The U.S. National Science Foundation has sponsored an Engineering Research Center entitled Power Optimization of Electro-Thermal Systems (POETS). This 10year federal investment, starting in 2015, encompasses 4 U.S. universities, 2 international universities, many commercial partners, as well as multiple government partners. The U.S. universities are The University of Illinois at Urbana-Champaign as the lead, The University of Arkansas, Howard University, and Stanford University. The international universities are the University of Sao Paolo in Brazil and The Royal Institute of Technology (KTH) in Stockholm, Sweden. The goal is to develop a different approach to developing electro-thermal technology and the engineers who will carry this forward into the future.

A Systems Approach to Electro-Thermal Design and Operation
Mobile power systems involve generation, conversion, transmission, and storage. It should be understood that there is no single technology that will improve power density across the entire chain of subsystems. Moreover, the introduction of improved components may simply expose a weak link elsewhere in the chain. Consider a relatively new semiconductor silicon carbide (SiC) that is being advanced in the application of transistors and switching devices which are at the heart of power electronics (Ozpineci & Tolbert, 2011). Included in many of its superior attributes to its replacement, silicon, is its ability to operate at high temperatures. While silicon can operate up to 150°C, SiC can safely operate to nearly twice that level. However, even if the active switching devices are made to operate at higher temperatures, there are still many other components within a power device (capacitors, control circuits, and drivers) that may not be able to withstand higher temperatures and may fail. Therefore, advancing the device development needs to go hand in hand with the overall component development and this component development needs to coordinate with the system development. The following presents two examples to illustrate this point.
Consider an inverter which converts a DC voltage, from a battery for example, to AC form in order to run a motor. One of the approaches taken by POETS researchers is the development of a flying capacitor multi-level inverter based on gallium nitride (GaN) semiconductor switching devices (Modeer et al., 2017). This approach has multiple benefits including higher efficiency and smoother waveforms for driving motors. Shown in Fig. 6 is a hardware prototype of the inverter. By having multiple levels, and using a very high switching frequency, the energy storage can be handled solely by capacitors without need for large and bulky inductors. This leads to a form factor with a very high surface area to volume ratio which is beneficial for stacking inverters in a modular fashion to scale up or down the power processing as shown in Fig. 6. The form factor is also favorable for heat transfer. Therefore, the power density, in terms of kw/l or kw/kg can be much higher than conventional efforts. In fact, early smaller prototypes of this system could get power densities 40 times those of the state of the art.
Having a large area over which active devices operate has potential for good thermal transport. However the large area distribution poses an interesting challenge for efficient thermal management. The active devices need to couple their thermal power they are creating with the heat sink to take that power away and prevent overheating of the active device. Typical designs would have a thermal interface material (TIM) that interfaced between the active devices and a heat sink collecting the thermal energy given off. Slight mismatches in manufacturing alignment over the entire area of the heat sink mean each active device may be at a slightly different height with respect to the heat sink. This leads to problems in getting good contact across all active devices with the TIM and between the TIM and the heat sink. Many devices may have good contact but there may be some where the TIM is not in good contact and these may lead to heat buildup and failures.
To overcome this challenge, the POETS team, consisting of researchers from the University of Illinois and Stanford University, created a modular approach to heat sinking (Pallo et al., 2018). They put an individual heat sink over each active device as shown in Fig. 7. Then, they created a custom manifold to draw air in over the heat sinks in a parallel flow manner with a common exhaust stream, also shown in Fig. 7.
The POETS approach taken is counter-intuitive from an electrical design perspective where power density focuses on Fig. 6 (left) Photograph of a 9-level, interleaved hardware prototype inverter showing two flying capacitor multi-level inverter legs. (right) CAD representation of multiple inverters stacked together to create higher power module. Note the aluminum heat sinks between each inverter the board and module design. The electrical designer would wish to pack the active devices as close as possible together, thereby increasing nominal electrical power density but giving up a lot in thermal management and reliability. The modular approach moved the active devices further away from each other to insert the modular heat sinks. This led to a larger electrical footprint than the non-modular approach. However, the modularity bought significant improvements in the thermal management which led to better and more uniform device cooling and a more efficient and reliable overall system. Overall, the power density of this system is 5-10× higher than comparable state of the art systems and maintains both active and passive device temperatures well within operational designs. Current designs are focused on optimizing the flow paths, reducing costs by having one fan instead of multiple distributed ones, and also the benefits to liquid modular cooling as an improvement over air. It is an excellent example of electro-thermal systems where considering both electrical and thermal functionality at the start results in an overall module improvement.
While the previous example illustrated co-design for a subsystem, the co-operation of overall electro-thermal systems is also important. This co-operation becomes more significant as the system under consideration becomes more complex. Fig. 7 (left) Inverter with modular heat sinks arranged over active devices. Inset shows front and side view of modular heat sink (Pallo et al., 2018). (right) Integrated fans and manifold with forced air cooling path. Intake air (blue) passes through the modular heat sinks, along the PCB and other components (purple), then exhausts across the inlet and outlet filter passives (orange) (Pallo et al., 2018)  Consider a candidate aircraft system of reasonably complexity shown in Fig. 8 and Fig. 9. The goal is to maintain the efficacy of the electrical systems while maintaining their thermal integrity. For newer more electric aircraft, such as the Boeing 787, the composite skin makes it harder to move heat out through large areas in the wings, for example. Therefore, the fuel system itself acts as a thermal capacitor to absorb much of the energy being generated by the electrical system.
In order to manage both the electrical and thermal systems, POETS researcher at the University of Illinois developed a hierarchical approach  combined with a graph representation (Williams et al., 2018) of the systems. This hierarchy is illustrated in Fig. 10. Each C i,j element in Fig. 10 represents a controller. For the purpose of the investigation in , the individual controllers were assumed to be receding horizon optimal controllers such as Model Predictive Controllers. Figure 11 shows the results of tests on hardware in the loop (HIL) systems over a candidate flight profile that compare the advanced hierarchical power management with a state of the art baseline . Since a smaller area within the radar chart is better, the results in Fig. 11 illustrate that the thermal and electrical systems are better coordinated. As with the component example above, the electrical systems give up some performance by shedding some electrical loads so as to enable the thermal system to meet its constraints. This is counter-intuitive to the electrical design goal which would be to maximize the performance of all electrical loads at all time. The electrical loads become thermal loads so the two branches of the tree in Fig. 10 are coupled. By coordinating and cooperating, there is a lower incidence of electrical system loads being shed to maintain temperature constraints than the baseline. In line with this, there are far fewer dangerous temperature constraint violations. The resulting impact on power density is that the power levels can be increased significantly above a baseline level without reducing overall vehicle performance. Actual values will vary depending on the number and types of subsystems involved. However, results from (Williams, 2017) indicated a factor of 2 increase in overall vehicle level power density without sacrificing baseline performance for the major figures of merit given in Fig. 11. Examining the two examples in this section, from a component and systems level, it can be seen that consideration of the electrical and thermal interaction leads to better design of physical components as well as better interaction in operation for the various subsystems.

A New Type of Engineering and Engineer Needed
The previous section indicated a new type of multidisciplinary engineering approach to increasing power density for electrified mobile systems. In addition to the two examples given, there are several more that could be used to illustrate the integration of electrical engineering with materials science and mechanical engineering. To accomplish this type of integration takes teams of engineers who are ready and willing to work across boundaries. It is important that academia actively prepares engineers for this setting.
One of the key goals of the POETS center is the development of graduate students that are both broad and deep, socalled T shaped engineers as shown in Fig. 12. These are engineers who understand the fundamentals of their given discipline, be it power electronics or micro-scale heat transfer. They are sufficiently proficient in their area of expertise to create and disseminate technical ideas of the highest quality and in the top venues. However, they also understand how the pieces come together to form systems and how they can lead innovative teams to take advantage of working across disciplines. They know how to communicate complex topics, both listening and presenting, including those that are outside their direct area of expertise.
In addition to the scientific and technical research being performed within POETS, there is also examination of how best to train students to become the ideal shown in Fig. 12. One active way of doing this is to create courses that are intentionally multi-disciplinary and require teams of students from different disciplines to address a complex project. This is currently being done across three different departments with cohorts of 20 graduate students per session. At a recent offering, the design project goal was to create the most power dense motors possible without the need for cryogenic cooling. The course was taught by 4 different faculty, each with a separate technical background. Early feedback from both students and faculty involved support the idea that this is a challenging, but very valuable pedagogical endeavor.
Another approach is to provide students with specific training in non-technical skills necessary to bridge disciplines. Short courses have been created within POETS targeted at training students in creating a business plan or structured brainstorming and ideation. These workshops are brief and focused but are recorded and archived to allow for asynchronous learning. Student feedback is positive. Even more positive is the feedback from industrial partners who are looking to have these students move into jobs at their companies after graduation. There is a significant appreciation for the students' well-roundedness in some of the softer skills necessary to work on large complex projects in addition to their technical depth.

Future Outlook and Conclusions
In industrialized society, the use of coal power pioneered our mechanized mobility needs in the form of ships and trains. The age of coal gave way to the age of oil which pushed modern society further, faster, and higher than ever before with automobiles and airplanes. Now, with the advent of renewable energy and the need to meet climate change challenges, there is an inexorable shift away from oil and towards electrified transport. History has taught us that these transitions happen slowly, over generations. However, the signs are there that this is a direction we will not be turning back Fig. 12 Engineers that are trained to have depth in their discipline but also breadth to reach across disciplines and think about other aspects of the technological problem from. The drivers for this change differ by transport mode and by individual nation, but they are real.
The question posed, in summary, is BWho will benefit and who will be left out?^This is the key question to be asked when understanding these technology transitions. This article argues that those who will benefit the most are those who are able to work across disciplines to increase the power and energy density of an electrified mobile infrastructure. In particular, the ability to incorporate both electrical and thermal aspects of power and co-design tightly integrated systems will afford significant benefits. These individuals and organizations will become the leaders in this emerging technological landscape. Those who do not will, unfortunately, be left behind.