So we’re in the Home Depot, which we’ve visited almost every day since Colin’s been staying with me. That’s because we’re building a cabin, and while we’ve got a good local hardware store, they don’t have lumber, and while there’s a good lumberyard a bit further away, their tool selection is lousy. And Colin’s complaining about my windows, which we bought at the ReStore in Pittsfield, giving $250 to Habitat for Humanity for five perfectly nice windows that fit tightly and are double-paned. But they’re not gas-filled, and that’s frustrating to Colin.
Colin comes by his frustration honestly. He’s a senior R&D advisor for the Department of Energy, which means he’s enormously knowledgeable about various bits of energy arcana. His knowledge of these topics runs pretty deep, because he’s not a political appointee – instead, he’s got a doctorate in physics, and extensive postdoctoral research focused on laser cooling, the process of trapping atoms in a lattice of light and cooling them to the point where they become probabilistic quantum blurs, sometimes superimposed over other atoms. (Heisenberg uncertainty means you can only know so much about an atom’s position and momentum. Slow the momentum to near-zero by trapping the atom, and the position gets uncertain. Get the position uncertain enough and you can smear atoms into each other, creating a new state of matter called Bose-Einstein condensate. I’ve long hoped this could be used to cool beer to truly frigid temperatures, but that’s usually when Colin ushers me out of the lab, apologizing to his colleagues.)
Double-paned glass works because heat has a hard time moving from one pane of glass to another, separated by a gap of air – heat has to transfer from one pane to the air, heating those molecules, which then slam into the other pane, transfering their heat. What you really want is two glass panes separated by vacuum so there’s no molecular transfer of heat, just radiation. But you’d have to massively engineer your windows to keep them from collapsing from atmospheric pressure. So instead, windows are filled with
gas molecules that are heavier than the nitrogen molecules that make up most of our air. Typically, manufacturers use noble gases like xenon or argon, but Colin is hoping they might get a little crazy and start using tungsten hexafluoride, a gas that’s 11 times as heavy as air, which means a balloon filled with the gas falls to the ground like a brick. (Of course, WF6 is highly toxic and forms nasty acids on contact with moisture, so Colin admits there might be some slight drawbacks to this approach.)
This gets us talking about the structure of hexafluorides and their ability to keep very heavy elements – like uranium – in gaseous states at room temperature. This gets Colin on the subject of sulfur hexafluoride, a largely unsexy gas that’s heavily studied by Iranian scientists. (Inhaling it is probably a bad idea, but breathing it will temporarily lower your voice.) That’s because it’s so similar to uranium hexafluoride that Iranian scientists involved with that nation’s enrichment efforts are able to publish peer-reviewed journal articles on their internationally-sanctioned research, presumably by experimenting with UF6 and publishing papers on SF6. (One can only imagine the conversations that led the Iranian nuclear industry to that compromise.)
This, of course, leads us to discussing centrifuges and their role in separating U235 from U238. And that triggers one of Colin’s other interests: arms control, specifically finding civilian research projects for former weapons researchers. (He spent a year as Representative Ed Markey’s science advisor, and Markey is a leader in thinking through nuclear disarmament, which involves, in part, finding jobs for US and former USSR nuclear weapons makers.) Turns out that if you’ve been working on building centrifuges, you’re good at spinning heavy objects at obscene speeds. And that means you’re well-positioned to design flywheels.
Ever used one of those flashlights that you power with a hand crank? Your cranking accelerates a heavy flywheel, which can remain spinning for a few minutes based on a few seconds of cranking. Put a magnet on that flywheel and you can generate electric power by induction – as the magnet passes a coil of wire, it generates a current, enough to power a radio and a small amplifier. Normally, we power a portable radio with chemical batteries – a flywheel is a mechanical battery.
And that’s when Colin proposes a short road trip, to Stephentown, NY. Stephentown is tiny town just over the border from the tiny town I live in, and is largely unremarkable, except for claiming to be the only town called “Stephentown” on Earth. And it’s evidently ground zero for the flywheel power revolution.
About half a mile from Stephentown’s main drag, surrounded by farmland and the town’s single pizza parlor, Beacon Power is easy to miss. Drive past the wooden fence to peer in through the chain-link gate and you’ll see 20 white shipping containers. Ten blue cylinders surround each shipping container – they’re about five feet in diameter and sunk deep into the ground. Inside each cylinder is a massive flywheel, a carbon fiber rim on a metal shaft, spinning at 16,000 revolutions per minute. The flywheels turn in vacuum to eliminate energy loss from friction with air, and they hover on magnetic bearings. Standing outside the gate, the 200 flywheels are spinning, but are totally silent.
Why keep 200 massive flywheels spinning 24 hours a day, 365 days a week? Turns out that electrical grids need lots of battery backup. Power use is extremely spiky – demand can surge very suddenly, and power demand doesn’t always occur at the same point as power generation. Think of an electric grid like Germany’s, where renewable power from solar and wind sometimes generates so much power that utilities bill customers at a negative rate to encourage them to use the excess electricity. The alternative is to find ways to store this power when winds are high and the sun is bright, and use it when it’s calm and nighttime. One simple solution is to pump a bunch of water up a hill when power is cheap and let it roll through a turbine when you need to retrive the power. This means grids can run a smaller number of plants at peak efficiency and bring on fewer “peaking” plants, which means less expense and fewer greenhouse gases.
Beacon’s system works a little like that. When power is cheap, the system uses energy to accelerate the flywheels. When power is expensive, the flywheels generate electricity and pump it into the grid. It’s a three-acre hedging strategy, buying power low and selling high. But that’s not why Beacon is so interesting. It’s interesting because it can store and release electricity really, really fast – in under four seconds, Beacon can go from storing power to discharging it.
Here’s why that matters: the electrical grid is all about stability. The US grid provides power at 60 alternating cycles per second. When demand for power balances the amount of power being consumed, the grid remains stable at 60Hz, but if there’s an increased demand, the frequency will tend to creep downward. That’s a bad thing, as many electrical systems will fail in unpredictable ways if the frequencies drop below 59Hz or so. (Same goes for high frequencies, caused by generating more power than there’s demand for.) Keeping the grid at 60Hz is the job of the Independent System Operators, non-profit organizations that manage a regional electric grid.
Beacon’s flywheels are one of the tools the New York ISO has to balance electrical load. If lots of people get up during the NFL playoffs and microwave a plate of nachos, the demand for power spikes, and the frequency drops. NYISO sees the frequency fluctuate and has a few options to stabilize the frequency. They can call up a gas plant operator and ask them to fire up their turbines, providing more power to the grid in a couple of minutes. They can ask a hydrostorage plant to release water and begin generating power in about a minute. Or they can call Beacon and start putting 20 Megawatts of power into the grid in four seconds. Beacon can’t sustain that output for very long – about 15 minutes – but the ability to store or deliver power that rapidly is a very valuable option for the ISOs.
Or, more to the point, it could be a very valuable tool for ISOs. Because Beacon delivers power in short bursts, it needs to charge a huge premium for power, perhaps 10x what “baseline” power costs. But the rates Beacon can charge are governed by regulations of the Federal Energy Regulatory Commission, and these rates were designed to compensate large power generators – gas and coal plants – for firing up their plants in times of need. The rates aren’t high enough for Beacon to be profitable, and in late 2011, the company went bankrupt. The bankruptcy got some press coverage, because Beacon had been funded, in part, with a loan from the Department of Energy. Because Solyndra had been funded through the same program and had also gone bankrupt, Beacon’s struggles were part of a narrative in which the Obama administration was throwing taxpayer dollars at crackpot energy schemes.
The reality is more complicated. Beacon Power lives and dies with FERC Order 755.
Order 755 changes the prices Independent System Operators pay to power generators for power used for frequency regulation, the short bursts of power that Beacon specializes in producing. The rules currently compensate operators primarily on the amount of power they inject into the grid; with Order 755, operators will be compensated based on their total power and their ramp speed – i.e., the time it takes operators to deliver their load of power. Because Beacon is much faster than anything else on the market, they are likely to get 2-3 times what they got before the order, making the company profitable.
At least, that’s the bet Rockland Capital is making. They bought Beacon out of bankruptcy and have paid back most of the DOE loans. Once Order 755 comes into force, the power station in Stephentown will be slightly noisier – we should hear a slight hum when the flywheels are shedding power or ramping up. That’s what Dave, dressed in six layers of his warmest camo clothing and wrestling with an ice-damaged cooling fan, tells us. He’s the only guy working at the 20 MW plant, which generally runs unmanned, and as he warms up, he reveals that he’s installed virtually every flywheel Beacon has built, both in the lab at the Stephentown facility. He’s also the guy who had to get the plant back online after one of the multi-ton flywheels came out of balance and slammed into the side of its concrete casing at Mach 2. Dave tells us it took only an afternoon to pull the dead flywheel out of the ground and put another in place, one of the many advantages of mechanical batteries over chemical battery plants, which can leak or explode.
Jason Pontin wrote an excellent piece in Technology Review titled “Why We Can’t Solve Big Problems“. Starting with a familiar lament – we got to the Moon in a decade, but now we can’t seem to accomplish anything big – it’s a thoughtful analysis of why progress seems to have slowed on some scientific and engineering problems. Pontin offers a few possible explanations: venture capital that’s more risk-averse than it likes to believe, a tendency of entrepreneurs to build cool toys rather than solving deep problems, a lack of visionary leadership to take on massive problems, a reduced capacity for tackling truly complex and multifaceted problems which manifests itself as the desire to find a quick fix. It’s a thoughtfully depressing piece and I’ve been thinking about his concerns for some months now.
There’s a good deal of hope that the folks who’ve built great internet businesses will turn their attentions to problems like energy independence and produce rapid innovation in that space. And while there are truly brilliant people like Elon Musk making that pivot, I’m concerned that few people will be able to make that shift.
I was a far less successful internet innovator than Musk. My company, Tripod, was part of an early wave of internet services that realized that users didn’t want to read professional content so much as they wanted to publish their own web pages. While we did some technically innovative work, we basically caught a lucky break – we read a market signal (users were more interested in their content than our content) and rapidly pivoted our business to meet their needs. While what we did helped presage MySpace and Facebook, no one believes that what we did was revolutionary or transformative. Clever, helpful, well-timed, maybe, but not world-changing.
Part of what was appealing about building Tripod was that the problems we were working on were small enough to be understandable. For most of the life of the company, everyone on the tech team understood, more or less, how the whole system worked. It was fairly easy to propose new features or steer the core product in different directions without discovering that what you were proposing was unfeasibly difficult. Personal homepages was a small problem, both in terms of total impact and in terms of the cognitive capacity it required. Ricardo Hausmann has a wonderful concept, the personbyte, which is the amount of information and skill a person possesses. Certain products require only a few personbytes to build – weaving cloth requires a shepherd, a spinner of thread, a loom maker and a weaver. (Add in dyers and spinning-wheel makers if you’re a completist.) Others – the laptop I’m typing on – require kiloperson bytes or more of skill.
Tripod required just a handful of personbytes to start it up. That’s the appeal of a tech startup – I get emails virtually every day from people who’ve got a great idea for a web service and just need one MIT intern to get it off the ground. And while they’re deluded in believing that I’m going to hand them one of my students to start their brilliant company, their assumptions about scale aren’t totally absurd: there are are companies like Twitter that got off the ground with just a few personbytes of skill and talent.
That’s not how companies like Beacon Power get started, I suspect. Building a 20MW plant required large scale collaboration between electrical engineers, materials scientists, sophisticated construction engineers, and the centrifuge designers that this post began with. Add in some smart businessfolks who know the intricacies of the power market and the needs of ISOs, and are capable of negotiating loan agreements with the DOE for tens of millions of dollars. I suspect Beacon required dozens of personbytes in the core team, leveraging thousands of personbytes one generation out, for bearings, concrete casings, vacuum pumps, etc. (You can argue that Tripod also had thousands of personbytes one hop out through server engineers, the authors of Apache, etc. But that supports my key point – because it’s so easy to leverage thousands of personbytes by using existing web frameworks, you might do something cool with one personbyte’s worth of innovation.)
Not only did Beacon need to marshall a big team of innovators and a great deal of capital – they are entirely dependent on external factors for their success or failure. With Order 755 in place, Beacon is probably a success. Without it, it goes broke. Most web products have had the great benefit that they require only end users to adopt (or reject) them. As Jonathan Zittrain has pointed out, if Skype required regulatory approval, it never would have come into being. Beacon doesn’t have this luxury – solving infrastructural problems requires working with regulators and with massive competitors and partners. Innovators in infrastructural spaces need to be brilliant not just about their technology, but about their niche, understanding complex systems well enough to see novel opportunities.
Because internet-based innovations are close to their users, they’ve got a lot of options for creating revenue. Skype was able to market VOIP calling to traditional phones directly to users. Tripod was able to charge for disk space and premium services. Beacon is very, very distant in revenue terms from its customers: they get paid a fee by the ISOs, which are paid by the utilities, which charge consumers based on rates established by local public service commissions, under rules established partly by FERC. To monetize a brilliant innovation in this infrastructure space, you can’t persuade a few thousand users that someone cool is going on and ask them to invite their friends to join – you need to convince extremely complex and powerful entities to make room for you and make it possible to generate revenue.
What does this mean for the future of innovation? I worry we’re often looking in the wrong place for new ideas. While it’s wonderful that people continue to create new software, I don’t expect those innovators to change how the electrical grid functions. And while I can hope for a wave of electricity hackers building self-configuring microgrids, I think the barriers to entry in that field are so massive that I don’t expect innovation from the garage, but from much larger startup firms.
The real opportunities to innovate around infrastructure probably aren’t pure technical solutions. They’re complex technical/regulatory/market solutions. If someone invents inexpensive roof shingles that function as PV solar cells, it will be a massive step towards reducing carbon emissions… if and only if we get better at connecting houses to the grid so they can produce as well as purchase power… only if we figure out how to better store and load balance the power created… and probably only if we subsidize PV roofing production sufficiently to make the tech affordable, which might require putting a meaningful price on CO2 emissions and making coal and natural gas power much more expensive. Oh, and you’ll want to train roofers how to install these new-fangled shingles, trying to convince Home Depot to begin stocking them, carry out UL tests to make sure the shingles won’t set your roof on fire, train fire departments how to modify their firefighting techniques for PV-shingled houses, work with insurance companies to see if installing PV shingles will affect your home insurance premiums, and on and on and on. I think I’d rather write software.
Brilliant work on energy technology will be done in labs, but the real hacking may be at the bureaucratic and policy layers. This matters because if we keep waiting for Elon Musk to save us, we may continue to feel Pontin’s frustration that we no longer walk on the moon. Technology entrepreneurship has been an incredibly powerful and positive force. But we may not be paying enough attention to problems that are too big, too multifaceted, too centralized to be solved by entrepreneurs in a garage. Perhaps there’s a way to make big innovation as sexy and appealing as small innovation, in the hopes that more people are willing to take the Beacon-scale risks we would need to tackle truly huge problems.
Tons more information on Beacon and their technology on their website.