Agents of Change
Ajay Raghavan: Re-envisioning batteries
“Our goal back in 2012 was to challenge conventional wisdom. I was an outsider coming in with another idea.”
Making batteries work better
Ajay Raghavan is a scientist at PARC, a Xerox company, one of the world’s leading research centers. In 2012, Ajay began applying his expertise in sensors and smart structures to improve battery performance. The results exceeded his wildest expectations — and those of his colleagues at PARC.
Here’s a common modern-world problem. Your mobile phone or laptop tells you it has 10% of charge left, then suddenly shuts down when you open an intensive application. But then after waiting a few minutes, you’re miraculously able to power it up again.
This is a sign your device’s battery management system is not doing its job properly. It can be a tiresome occurrence. But the consequences of inaccurate battery management extend far beyond day-to-day inconveniences for laptop and mobile users. It can also lead to large-scale industrial accidents, and inefficiencies that cost the world billions of dollars each year.
Even more significantly, the inaccuracy of battery management systems has been holding back the adoption of grid storage, and the world’s transition from gas-powered to electric vehicles.
The race to improve battery performance has huge implications for the health of the planet.
That’s why Ajay Raghavan, an expert in the field of sensors and smart structures, felt a jolt of excitement in 2012 when he struck upon an idea to insert fiber-optic sensors into batteries to improve their management systems. The more Ajay thought about the idea, the more excited he became. It turned out to be the first step of a very long and challenging innovation journey.
This is a story about the ingenuity of a scientist and his team. But it’s as much a story about the courage, belief and perseverance needed to capitalise on moments of ingenuity — and turn them into world-changing innovations.
Ajay, how did you become interested in engineering?
Well, you could say it runs in the family. My dad is an engineer. He worked with Air India pretty much all his life. I have some uncles and other relatives who are engineers as well. That sowed the early seeds of interest. Then, in high school, I had a strong interest in science and math in general. I was very curious about how things work — taking things apart, fixing them back up. I also had interest in the creative side of designing things and making them as efficient and appealing as possible.
My bachelors was in mechanical engineering at the Indian Institute of Technology Bombay. That was a big force in shaping me as an engineer. It’s pretty much one of the best engineering schools in the world.
At what stage did you become interested in sensors?
I remember attending a seminar by a professor in civil engineering who talked about the idea of embedding sensors on bridges to make the bridge operators aware of its state and condition at any given time. As opposed to having maintenance personnel go out and check the bridge, say, every six months. He talked about how this is applicable to a broader range of structures, such as aircraft and spacecraft. Sensors monitor continuously — then algorithms can process those complex status streams to make sense of how the structure is at any given point of time.
This was my introduction to the idea of smart structures and systems. It really struck a chord and I decided to learn more.
How did you end up working at Xerox?
After my degree, I went to the University of Michigan, Ann Arbor where I did a five-year PhD research project funded by Nasa on how these ideas would apply to aircraft and aerospace structures in general. But in the aerospace industry the typical timeframe for new technology adoption is anywhere from 15 to 30 years. That, for me, was not fulfilling enough. I wanted to see rapid production and deployment of technologies I’d worked on. Then I came across a research opportunity at PARC.
During my interview, I got a short lab tour with a principal scientist, Dr. Peter Kiesel, who was working on some interesting early ideas around low-cost, compact optical sensing. I’d read about optical sensing as an option for smart systems and smart structures. So straight away I saw very interesting potential for a big project.
What kind of technologies did you have in mind?
Well, I teamed up with Dr. Kiesel and explored a few possible applications for low-cost optical sensing technologies. There were a few failed proposals along the way. Then we found out that the ARPA-E [the Advanced Research Projects Agency-Energy is a Department of Energy agency that funds high-risk projects with high rewards] was interested in condition assessment possibilities for battery management systems in electric vehicles, grid storage and other advanced applications.
I had already thought in passing about the possibility of putting sensors in battery cells, but, in that moment, I started seriously considering it. So I just dropped this gentle hint that maybe embedding a sensor inside would be a good idea.
Can you explain what a battery management system is?
A battery management system monitors the health and condition of batteries — using parameters such as voltage, current and temperature. The better the BMS, the more it can protect batteries, help them deliver full power and prolong their lives. When battery management systems don’t do their jobs, the consequences can be pretty catastrophic in certain cases. There’s the Boeing 787 battery problems that have cost the aviation industry over $1 billion recently. Then there are all the electric vehicle fires, and safety incidents. There have been some grid storage battery facility fires as well. There are literally billions of dollars spent every year trying to improve battery technology. It is a very important area of energy research.
What stops battery management systems from doing their jobs?
Well, voltage, current and temperature are all electrical parameters that are monitored from outside the cells, which really don’t tell you enough about the state and health of the batteries at any given time. Because of that, people tend to be really conservative with how they design the packs. And they tend to be really careful with how they use the packs — the full battery capacity is rarely used.
So it’s a problem for laptops and mobiles turning off when an intensive application is opened. But you see a much bigger version of this problem when you go to electric vehicles or grid storage because the loads and constraints are much more aggressive.
Everyone in the battery industry feels there should be a better way to manage batteries. But nobody really had a good handle on how to do it, because they were all so into that conventional mindset of, ‘We just have these three parameters that you can monitor outside the cells.’ Our goal back in 2012 was to challenge that conventional wisdom. I saw a big opportunity to use our fiber optic sensors inside battery cells.
How did people react to the idea?
I can’t say it was a slam-dunk. There was interest among colleagues and people in the battery industry but a lot of hesitation too. To be honest, in battery research there’s a long list of people who have come up with crazy ideas that have gone nowhere. A lot of people have tried different options for embedding a sensor inside a cell but they’ve gone nowhere. The inside of a battery is a hostile environment for sensors. Nobody had really tried fiber-optic sensors, and I was an outsider coming in with another idea.
I managed to convince everyone that at least we should put in a ten-page concept paper to ARPA-E to try and get funding. We proposed this very raw initial idea of embedding fiber optic sensors and using a version of a low-cost readout, tying that in with smart algorithms that can use that raw data to infer what’s going on inside battery cells.
We submitted that concept paper and got a hint of encouragement from ARPA-E as well as a gentle suggestion of partnering with a battery manufacturer. Then there was this window of a couple of months for submitting a full proposal. I think we talked to three different battery manufacturers in that time. The response was a very interesting mix of excitement and fear. One big manufacturer of batteries for electric cars had something like five conversations with us. Each time they said it was so intriguing and so exciting but they were scared of all the risks. Time went by and we were just ten days away from the proposal submission deadline. Internally at PARC we had decided that if we didn’t get a battery manufacturer’s partnership, we weren’t going ahead with the proposal.
It went down to the wire. Then one of the battery manufacturers we were talking to, LG Chem Power, finally decided to go for it. This happened ten days before the deadline. To cut a long story short, we essentially went without sleep for the next ten days to make sure we put in a decent enough proposal. We just about managed it!
Months later, when the ARPA-E funding announcement was finally made, it just stunned all of us. We got $4 million for a three-year project, which was at least $1 million more than any of the other awards. Pretty much everyone at PARC was blown away.
We knew there were a lot of technical risks associated with the proposal and so the next three years were going to be really interesting on a lot of different fronts. Keep in mind back in 2012, electric vehicles were a fledgling market.
The idea sounds an obvious winner, so why was it considered so risky?
Because it was risky. There are three things you have to ensure when you do this: One: you are not negatively affecting the battery’s performance, either in terms of how much charge capacity it can hold, or how fast or slow you can discharge a cycle with the sensor inside the cell. Two: that you’re not affecting the seal integrity. Three: after all that struggle, are you actually getting a signal that justifies all the effort you went through to get the sensor inside?
I can tell you that we must have gone through ten different iterations with LG where something or other failed along the way. Every time we went through with an iteration it was heartbreaking because the LG team that was helping us was located in South Korea. Each time we had to send them sensors, they had to put them into the cells and, because it’s a brand new cell, there’s a one-month activation period before the cell is ready to be cycled. Then they had to ship it back to us.
Did you ever feel you wouldn’t pull it off?
I won’t deny that there were down days when I thought, “Oh my God this project is going to end next month.” But ARPA-E, LG Chem, and the PARC management team kept supporting us throughout. They never wavered.
How successful has the SENSOR project been?
At the start of this project, everyone we talked to in the industry agreed that if we managed to achieve measurement accuracy within 2.5% that would be really valuable. To put that into perspective, the state of the art is 5% or worse. Nobody in the electric vehicle industry will openly admit how bad their systems are, but in cold temperatures the accuracy is probably no better than 10%. The initial algorithms that we’ve developed have shown we can get 2.5% or better cell-state accuracy across a broad range of conditions and case scenarios.
Our vision is for this technology to be adopted in electric vehicles — then who knows where else it might go? We’ve shown we can actually reduce the cost and size of optical read-outs. A lot of optical readout systems you get on the market today cost anywhere from $15,000 to $35,000, which is about the cost of an electric vehicle battery itself. They aren’t suitable for cars — they’re really just meant to be lab instruments. But with our technology for reading fiber-optic signals, you can shrink it down to smart phone size — and reduce the cost to a few hundred dollars. This gives me a lot of reason for optimism to believe that it can be deployed on electric vehicles.
So what’s next for the SENSOR project?
General Motors has taken the SENSOR technology in-house to their battery lab in Michigan. It is the best battery lab in the world — 85,000 square feet dedicated solely to testing. They are going through a much more comprehensive range of test scenarios we simply cannot do here. Time will tell whether we survive the acid test of commercialization. But I will say, even if somewhere along the way we fail, God forbid, there is a serious use for this technology as an evaluation tool in laboratories and facilities around the world. This in itself would be a huge success. We need to work with our commercial team to see how else this can be rolled out.
What has the reaction to this project been at PARC?
It’s certainly set the stage for a lot more ARPA-E projects at PARC. When we put in our proposal to ARPA-E there was no other ARPA-E projects despite several earlier attempts. Today there are eight at PARC. From ARPA-E’s perspective, we are their largest private sector awardee. Of course, this isn’t all down to my team’s work: several other brilliant PARC researchers, managers, support staff, and great partner organizations worked together to make this happen. But I feel that we had some role to play there.
Congratulations on the project.
Thank you. It’s been fun.
We’ve all changed the world. Every one of us. With every breath we take, our presence endlessly ripples outwards.
But few of us have the opportunity to change many lives for the better. And even fewer are challenged to do so every day. That’s the gauntlet thrown daily at Xerox research scientists — to try and effect change.
In return, we give them time and space to dream. And then the resources to turn dreams into reality — whether they’re inventing new materials with incredible functions, or using augmented reality to bolster the memory of Alzheimer’s patients.
We’re proud of our Agents of Change in Xerox research centers across the world.
Visit PARC to find more information about the SENSOR project.
The Perfect Industrial Internet of Things (IIoT) Platform
In recent years, we’ve seen an evolution of IIoT capabilities, with advances in data science and more accurate and cost-effective sensors leading to broader adoption of IIoT solutions.