Easing EV Range Anxiety Through Faster Charging

The automotive industry is developing new ways to boost the range of electric vehicles and the speed at which they are charged, overcoming buyer hesitation that has limited the total percentage of EVs to 18% of vehicles being sold.[1]

Work is underway to improve how batteries are engineered and manufactured, and how they are managed while they are in use or being charged. This extends well beyond just the battery itself. It includes maximizing usage to the wheel, and ways to more efficiently and effectively bring energy from the grid to the battery, and in some cases back the other way.

“Compared to 10 years ago, battery energy density has gone up three times,” noted Puneet Sinha, senior director and global head of battery industry at Siemens Digital Industries Software. “That means in the same amount of battery, there is three times more range. Cost has come down almost 90% in the last 10 or 15 years, which is good. But the range anxiety meter hasn’t changed much.”

This is largely due to the speed at which batteries charge, which is heavily dependent on the architecture of the batteries and the chemistry inside of them. Charging time depends on how fast lithium can be moved from the cathode to the anode side of the battery. If the lithium is not accepted fast enough, it can lead to dendrite creation, where lithium piles up. That, in turn, can short the battery, leading to potentially catastrophic issues like fires. The design of battery cells and packs also comes into play here.

Today’s state-of-the-art lithium battery technology predominantly leverages graphite as one of the anode materials, but this material has limitations on how fast lithium can be moved, and is very prone to having the lithium dendrite issue that can lead to negative events. This is why charging is done in a controlled manner.

Heat management is another challenge. “When you’re charging something and there’s heat created, how are you managing heat not only in the battery but also in the charger? Chargers also get hot,” Sinha said. “If the charger gets very hot, especially when you are trying to push higher power into it, controls will set off and the charging speed gets reduced. The biggest enemy of electronics is heat, and heat is not always a good friend of batteries. If the battery temperature goes high and it stays there, it may reduce the life of a battery. That means the chemistry limitations of certain cell designs and pack designs impact how fast the battery can accept the charge. And if it can accept the charge fast, you can potentially generate a lot of heat. How that heat is being managed is going to reduce the charge speed automatically.”

Many believe thermal management, as well as sub-system energy balance, will require model-based system engineering (MBSE) tools to couple various sub-system models to create a full network model representing the entire vehicle. “Control software development and subsequent software calibration are key steps to achieve the necessary levels of optimization,” explained Kumar Srinivasan, senior technical director for CFD solutions at Cadence. “This process involves concurrent evolution of hardware and software design in a fully integrated simulation framework to facilitate HIL/SIL/MIL development,”

Today, even the fastest chargers commercially available still take between 20 and 30 minutes to fully charge a vehicle. While that is much better than in the past, there are many efforts in the industry to reduce charging time further, and this is where many industry players are now looking to add their expertise.

“EV energy management involves optimization of various sub-systems, both individually as well as cross-functionally,” Srinivasan said. “Starting from the initial target for ‘energy budgets’ associated with all sub-systems that draw from the vehicle’s battery power, the balance between sub-system and full-system optimization involves multiple simulations that necessitate coordination and frequent exchange of technical specifications and operational assumptions. The strong coupling between battery state of charge (SOC), operating temperature, and instantaneous energy demand from various sub-systems necessitates a fully integrated simulation framework that represents a ‘digital twin’ of the vehicle with multiple layers of fidelity and physical modeling capabilities. During early development of an EV, aerodynamics, thermal management, weight optimization, battery and driveline controls and calibration, tire rolling resistance, and sub-system parasitic loss reduction are all key focus areas.”

Aerodynamics optimization, another key consideration in overall EV design, involves high-fidelity CFD tools that need to have reliable accuracy and precision to assess very small design changes. “Exploring a large design space is essential in aerodynamic optimization, and this means acceleration of CFD tools using the latest GPU architectures is a key enabler. The need for accuracy and precision of CFD simulation for aerodynamic optimization is further heightened for EV vehicles compared to conventional internal combustion engine (ICE) vehicles,” he said.

Also, EV weight optimization involves several structural design of experiment (DOE) studies to balance crash-worthiness, noise vibration and harshness (NVH), durability and ride and handling characteristics. “This requires several simulations to fine-tune and optimize the structural design of the vehicle from an early design phase. OEMs have adopted pre-CAD simulation methods to drive key architectural decisions from a very early phase with the focus on achieving the right balance between various attributes and maximizing vehicle range and performance,” Srinivasan said.

Further, power applications with rapidly varying energy demand require low-latency control to maximize power efficiency. “Asynchronous digital controls from embedded FPGA provide precision power management by reducing loop latency as well as flexibility to adapt to any application and power demand,” said Jayson Bethurem, vice president of marketing at Flex Logix.

Fig. 1: Asynchronous logic front-end modules For PMICs. Source: Flex Logix

Indeed, as EV battery efficiency is directly tied to the motor control, controller chip switching speeds are on the rise. Companies like AMD and others are looking beyond the traditional standard microcontrollers and DSPs that have been used for motor control for 20 or 30 years. Wayne Lyons, senior marketing director, automotive at AMD explained that with new battery technologies being introduced, along with new switching technologies, higher efficiencies can be realized in the EV. One such technology is silicon carbide. “There are two wins you get with silicon carbide. One is you get a quick win, just by moving to the technology, because the switching thresholds are much faster. Usually, the switching losses are also reduced significantly. A lot of car companies are using that quick win of changing to silicon carbide, but underlying it, they’ve still probably got the same control loops running.”

A bigger win comes from changing the software as well. “In a lot of the e-racing platforms where battery performance, and longevity are key and critical, they’re willing to invest in reengineering the software as well as the hardware in order to deliver the best solution,” Lyons said. “We’re seeing it today in the high end of e- racing, I think we will start to see it more and more in other applications. We’re developing additional solutions with FPGA technology because it’s flexible, because it allows you to create much tighter and higher resolution control loops, so you can change switching frequencies more quickly.”

Lyons expects FPGAs to gain a foothold here as these devices allow almost cycle by cycle control over the pins (IO). “You can map a software function or software loop directly into a hardware function, and that allows even more precise control over the IO, which allows to run cycles on IO, and respond to variations in the motor behavior immediately and change the control loop immediately. Control loops usually occur 10,000 times a second or so. You can move that up to 100,000 times a second, because you can shorten the control loop and respond much quicker. That allows you to drive additional performance out of the switching, and therefore improve the battery efficiency by a few more percent,” he explained.

This could help with charging speeds as well.

“Essentially it’s the transfer of energy, and because switching losses reduce or impair the transfer of energy, that equates to either lower battery life, or faster charging time if you’re in the charge point. FPGAs are being considered for the charge point, as well as the onboard charging as well, where you actually map the high current coming in, and direct it straight onto the battery and specifically onto the battery elements accordingly, as well.”

Also, since predicting battery behavior is critical to maximizing safety and performance, Lyons has also noted an uptick in work around applying its devices to battery prediction. “Especially in racing, you don’t just want to know what your battery behavior is right now, you want to know what it’s going to be at the end of the long straight that you’re driving down, and maybe in one or two lap’s time, so you’re predicting all the time what the battery’s going to look like. In academia, there are many different types of MATLAB models that you can run to predict the behavior of the battery on a certain day, in given environmental factors. What you can do is predict that, and then at the end of the lap, you’ll know how the battery is. Over time, you can use that information to perfect the model of the battery on that particular day. A battery will respond based upon many different factors such as wear and tear, overall lifespan, but also the environmental factors on that day, what the temperature is, the humidity — many different factors roll into that. Modeling is understood, but perfecting the model is really where the science comes into this, and we see high performance math algorithms being run. Again, not just in the processor side, but also in the FPGA, where you can accelerate the math functions, be it AI functions, or just general math models that are being accelerated.”

Collaboration is key
From a technology standpoint, EV energy management necessitates a strongly coupled design and development process with multiple layers of single discipline optimization (SDO) at the sub-system level as well as cross-functional multi-disciplinary optimization (MDO).

“This requires strong collaboration between the entire component supplier ecosystem and the EV OEM,” said Srinivasan. “Standard simulation tools and data exchange frameworks are needed to facilitate such a collaborative approach. The level of integration required needs to be fully understood and documented as part of commercial discussions at the sourcing phase of a new program. Detailed simulation timelines that capture when various sub-system models need to be made available to facilitate system integration need to be developed and communicated to the entire supply chain ecosystem. Standardization of simulation tools and methods as well as data exchange protocols are necessary to facilitate development of multi-layered digital twins.”

However, a mindset shift is required to optimize EV power management. “OEMs can no longer rely on changing ECUs one-by-one,” said Gary Martz, senior director of advanced technology standards at Intel Automotive. “Instead, they must adopt a whole-vehicle approach. For example, if an EV is charging, why is the ADAS ECU powered? It doesn’t need to be, but the power sub-system doesn’t know that. If instead the central compute system shared that the vehicle was currently charging, and the ADAS ECU is going to be turned off, then less energy can be drawn from the battery. Apply this concept across the entire vehicle, such that each ECU can be controlled from a power management standpoint, and there are infinite possibilities for conserving energy. It’s winter in Detroit, so turn off the A/C ECU. It’s summer in Phoenix, turn off the seat heater ECU. The reliability of EVs for long drives is a consumer concern because battery systems aren’t fool-proof. Current estimates don’t consider different situations that occur when you’re driving, such as changes in topography and weather, or the distance between charging stations. In order for optimal EV energy management, vehicle elements must be interoperable.”

Battery chemistry
A fundamental aspect of EV battery management involves the chemicals used in the battery, which is why battery companies are working to devise other chemistries that could replace graphite to enable fast charging, as well as working on improving cell and pack designs that allow fast charging speeds. “Those activities are happening in the battery world — how you design the cell for fast charging, the chemistries, solid-state batteries that replace the combustible liquid electrolyte, and with a solid electrolyte to minimize the risk of unsafe events,” Sinha noted.

Yet another challenge to overcome is how fast energy can be pumped into the battery. “Many believe one of the key technologies to unlock this capability is gallium nitride semiconductors,” Sinha said. “Many years back it was all IGBTs. Then came silicon carbide. Gallium nitride has the potential to improve the charging speed because it can pump a lot of power and still remain constant. For example, a lot of cellphone chargers, while maintaining the same size, are a lot faster because they’re switching to gallium nitride chips. That’s the direction these EV chargers are looking to take. There are challenges with cost, there are challenges with certain quality aspects of gallium nitride. But all of the big semiconductor players are investing in it, and many others are pivoting to that, so it is a key enabler for this to happen.”

Grid-level energy management
To spur higher adoption of EVs, energy must be managed at the grid level. “So what if you have the very high-powered charger? Those chargers still need to work with the grid, and that’s where having energy management at the grid level matters,” Sinha noted. “Depending on the time of the day, there will be extra demand on the grid, which starts increasing electricity costs, and raises the question of other renewable energy-related resources. What about the duck curve, and how all of these need to come together? That’s why there are many companies looking to participate in that part of the industry, and there are multiple ways the ecosystem is trying to attack that. One way is to modernize the grid. Renewables are also becoming part of it, and a key player enabling this is batteries. At the energy storage grid level, or even the industrial level, there are a lot of companies looking at the network of those batteries — like a Power Wall — and trying to determine if they can use the energy from them to support peak demand. How this is managed is an equally important aspect of EV energy management.”

This is directly related to vehicle-to-grid connectivity. Here, companies like Shell and others are actively participating from the utility side as to where to put their chargers, and bringing their utility and energy knowledge to bear since they know what it takes to put in large networks.

“Even with movement in all of these areas, there is still a way to go before the full potential of the combination of these technologies is maximized from materials for batteries to battery technology and semiconductor technology. Then, in a grid modernization, all parts can come together,” Sinha said.

Ecosystem collaboration
Enabling this requires extensive collaboration across the ecosystem, including EV companies, battery makers, and utilities. In addition, it needs local governance and transparency for consumers. “Consumers need to have a say in how they want to participate in this, and their participation is critical,” Sinha said. “There’s a lot of work happening on vehicle-to-grid, because most of us are driving cars only 5% of the time — 95% of the time the car is parked. Increasingly the world is looking at electric vehicles as batteries-on-wheels as they are growing in number, with essentially one in five cars being electric.”

On top of this, electric vehicles are rolling out in different categories, which means varying needs from the grid. Passenger cars and various types of trucks all have different mission profiles, which means the energy demand is different for each. “We are asking how the vehicle-to-grid concept, which batteries are at the center of, can be enabled. That includes the battery technology, the power electronics to support bi-directional energy movement, and the consumer who needs transparency in order to decide how to participate in that,” he noted.

No single company can solve all of the issues here on its own. Intel’s Martz notes that each automotive OEM already is working on its own advanced power management solution, but developing the solutions in proprietary silos is the exact what’s holding the industry back.

“The leading edge of the technology curve will only be enabled through collaboration and sharing the best ideas to build the common parts of automotive power management solutions,” he said. “This will enable OEMs to focus their R&D dollars on the true areas of differentiation, such as advanced management algorithms and AI. This is the focus of the ‘Vehicle Platform Power Management’ standard that is being developed by the industry in SAE J3311. The objective of SAE J3311 is to standardize the power management interfaces between automotive operating systems and ECU, to enable OEMs to focus on deploying their own competitive context-aware power generation and consumption solutions that sit on top of the common elements.”

Martz also serves as chair of the SAE Vehicle Platform Power Management J3311 committee, which is working to develop documents that will define ECU interfaces and functions necessary to enable OEMs to develop and deploy context-aware, vehicle-wide optimal power generation and consumption while allowing differentiation in implementation.

To really move things forward with EV adoption, a number of issues need to be solved on the technology front.

Sinha believes one step is to move solid-state battery technology toward commercialization. “Solid-state batteries have been in the making for many years. Progress is being made, but still hasn’t been commercialized at scale because with the chargers we have today, the infrastructure is not necessarily maximizing the power they can deliver because of the grid limitations. How can we unlock that? Second, on the charger side, we need to further gallium nitride and the work happening there. It is already commercialized for consumer electronics, and a lot of EV inverters are looking at how gallium nitride is starting to show up. Maybe the next breakthrough for that technology can help with EV adoption. Finally, grid modernization and vehicle-to-grid activities need to progress, which will take a bit longer because there are a lot more parties involved in that area.”

The future
Options are emerging as to how various flavors of AI can be used to improve battery management and predict battery behavior in electric vehicles. This may include the use of mathematical models and algorithms to predict battery behavior based on various factors like temperature, usage patterns, wear and tear, etc. Or, AI and machine learning techniques may be used to help improve these models over time by learning from real battery usage data. As with all things automotive, technologies are evolving at a fast clip, leaving the OEMs to contend with when, where, and how to add them into their already-evolving design and manufacturing processes as they move toward increasing levels of ADAS, AD, and infotainment features in their vehicles.

Reference

1. International Energy Agency report on trends in electric cars.