Sambal

I’ve been playing with Google Maps recently. Driving not long ago, my wife wanted to check out Lake Wabamun for possible camping/picnic spots, so we took a detour. We didn’t find any, but did find something more interesting: a decent roadside tour of where our power comes from. Here’s a nice view of Lake Wabamun, and the Highvale mine, the largest coal mine in Canada, which provides coal for the Sundance and Keephills generators. The big blue thing is the cooling pond for the Sundance generator, with the Keephills pond being to the right a bit, kind of jellyfish shaped with the plant itself at the end of the one tentacle.

Keephills is the newest one, 766 MW net capacity, burning 3.2 Mt of coal/yr. That works out to 7.5 MJ/kg coal. Sundance is larger, 2020 MW, 6 units (boiler, turbine, and generator), three stacks, 9Mt/yr of coal (7 MJ/kg coal). A typical heating value for subbituminous coal, the kind mined at Highvale, is 20-21 MJ/kg, and the newest units at Keephills are supposed to be something like 38% efficient, so these numbers make sense. Since a kWh is 3.6 MJ, it takes about 0.5 kg C to make one, releasing 1.8 kg CO2. So that’s about how much CO2 my electric power releases.

I can now, at long last, work out the GHG emissions due to my ebike. A full charge contains 24 V * 12 A-h = 0.288 kWh, * 1.8 kg CO2/kWh = 0.52 kg CO2, and using a 20 km range, this works out to 26 g CO2 / km. This may be an optimistic number: I’ve only tried going a full round trip of 20 km a couple of times, and the second time the bike left me pedalling myself a km or so from home. Also, it doesn’t take into account the pedalling work I’m putting in (substantial), charger efficiency (pretty good in general, but my particularly old-fashioned charger could be as bad as 50%), transmission losses from the plant (probably just a few percent), or the fact that the average power I’m consuming may not be as efficiently generated as the newest generator installed at the local plant. Still, it’s workable as a rough best case. For comparison, a Smart fortwo emits 90 g CO2/ km (Chrysler spec), and my car emits about 180 g CO2/km (at 8.0 liters gas / 100 km and 2.2 kg CO2/liter gas, computed assuming gas is pure octane). So, best case, we’re reducing GHG, definitely by some, likely by a factor of 3 to 7, and certainly less than 10. In my understanding, ten-fold is the important target. Ten-fold reductions would allow India and China to develop to first world lifestyles while moderating anthropogenic climate impact enough to matter. The real story here is how hard 10x is to achieve, even in nearly ideal circumstances (short commute, beautiful weather, committed participant). In other words, get ready for a warmer planet.

The thing is, an ICE scooter would probably do as well, be more fun, and get me to work faster. One number I found on a 50 cc four-stroke says 1.3 l/100 km, which means 29 g CO2/km, in the ball-park of the coal-powered bicycle. Of course, any gasoline consumption still funds global terrorism, and pollution emissions per km from small ICEs like scooters tend to exceed (by a lot!) even large modern cars, so there are some tradeoffs. Not to mention my wife might classify them as a motorcycle.

Motorcycles have had shaft drive for years, nice to see it come to bikes too. I think it’s not quite as efficient as a chain, but more reliable and lower maintenance. If you’d like continously variable gears, that’s also possible, though not everyone thinks it’s necessary.

First day on the bike in awhile. Windy as heck this morning (flags straight out, ripping), so I tried to preserve the assist for playing with traffic. Just blasting through the wind is taxing on the battery, because the motor is geared high, so draws lots of current at low speed / high load. (Thus, the thoughts about transmissions.) Mostly I just pedalled, which though slow, worked fine. If I wanted a bit of assist, though, that’s not so easy. There are two problems with the throttle. One is ergonomic: it’s a thumb controlled pot, and bumps jitter it around. There’s no doubt a reason motorcycles have adopted the twist grip for throttle control. The other is electronic: I suspect the pot controls the effective battery voltage, using PWM. But until the effective voltage exceeds the back emf of the motor at the current speed, there’s no assist at all. So the entire range of available assist is wrapped into a short range between some partial throttle position, a position that shifts with increasing speed, and the top. It’s as if the accelerator pedal didn’t work until you’d pushed it down 3/4 of the way, then leaped forward. (Jerking you around, resulting in the pedal wiggling more, etc.) The interface is wrong: it should control current, which turns into torque, not voltage. This would make sense for a cruise control, where you wanted to set the speed, but for throttles we’re used to acceleration. The bottom line is, at speed, the Currie thumb throttle is only good for on/off, though it’s nice to have fine control for maneuvering at parking speeds.

I’m still chewing on the transportation efficiency thing. This isn’t necessarily productive, as it’s easy to go crazy trying to figure this stuff out in the abstract. (It’s also easy to get things badly wrong, like an early report that electric cars would emit 60x as much lead per mile as a conventional car burning leaded gasoline. Even were the numbers valid, are all “emissions” equal? Lead particles in the air, ready to rain down into the drinking water, seems different from the occasional fully encapsulated battery ending up in a well-maintained modern landfill.) It’s also unnecessary from a policy perspective. You don’t need to best answer to set policy, just a way to price the positive and negative externalities, so that those costs can get included in the prices paid by consumers. Let the invisble hand work out the optimal compromises.

Still, even if I want to do the calculation, I’m not sure how to do it. There are so many large uncertainties. Does it really make sense to charge the bicyclist for food energy? After all, perhaps they would normally have driven their SUV to the gym to work out for an hour a day had they not been biking, making their personal energy expenditure neutral. Do increases in exercise, at the margin, always lead to increased food consumption? In increased food purchasing? Human physiology is more complicated than mechanical engines, and it’s easy to imagine physiological states where increased exercise leads to decreased consumption. Does increased food purchasing lead to increases in production, or simply less waste as farmers produce all they can in any case? Are we interested in energy costs at the margin, under current conditions, or in some ideal Kantian centrally-planned state where everyone is making the same choice? You’ve got to answer all of these before generating an answer. Still, I bet the Lomborgians would love any calculation (no matter how valid) that showed bicycling to have a bigger environmental footprint than driving an SUV.

ObRide: Through rationing power usage to be kind to the battery, I made it all the way to work yesterday, and home again too. 25.3 final there, 25.5 final back. I felt like I was outpedalling the motor during the hard parts, though, resulting in no assist. Just can’t win.

I’m not the only one to think this. The NY Times recently printed a review of the new Panasonic Oxyride battery, an alkaline disposable which is claimed to last 1.5-2x longer than regular premiums. The journalist not only did some tests, but actually called the manufacturer when they didn’t live up to the hype. They laughed at him and his attempts at testing, but did explain some of the complications of trying to do it right. Keep reading, and you can laugh at mine…

(more…)

Justin Lemire-Elmore thinks so, and I finally got around to reading his paper. Bicycles in general are very good: wheels, minimal non-cargo weight, and relatively low speed minimize rolling and aero losses. One might think that adding a motor would only make things worse. However, a bicycle pedalled by a human still requires energy, but that energy comes from the food they eat, and food is not produced very efficiently. So the question is whether the energy costs of making and charging the battery etc. are more or less than the energy costs of making the food the bicyclist needs to eat.

The bottom line is that all battery types outperform food, because so much energy goes into food production in modern Western farming. Food production is so inefficient, producing 1 calorie of food per 7 calories of fossil fuels consumed, that I wonder whether conventional bicyclists even beat cars! Lithium, due to its light weight and high charge efficiency, is best. Lead acid is worst, despite having the lowest energy costs of production, due to relatively low cycle life and transportation costs. One can quibble here. He uses a “real-life” cycle life for lead acid, but a manufacturer’s spec for lithium ion. Worse, for transportation costs, he assumes air freight from the far east: reasonable for expensive lithium batteries, but implausible for heavy lead-acid batteries. In this analysis, initial transportation accounts for a substantial majority of the total lifecycle energy cost of lead-acid, so this isn’t trivial. Finally, and he does discuss this, nothing forces the bicyclist to eat conventional, high-energy input food. Bicyclists on a strict organic vegetarian diet can equal or beat the electric bike. But that’s what it takes.

Here’s a nice little EV conversion, John Wayland’s 1972 Datsun 1200. The range isn’t so good, only a quarter mile, but it gets there pretty quick. Under 13 s at over 100 mph makes it the fastest street-legal EV, and it beat a built 375hp modern Camaro V8 to get there. The story makes, um, interesting reading. Drag racing breaks stuff, and high power electrical equipment, unlike ICEs, has an annoying tendency to break into a shorted, full power condition.

Anyone interested in this stuff who’s near Las Vegas could go to Wicked Watts, the first event in the 2005 NEDRA season, this Saturday at the LV Motor Speedway.

Had a failure yesterday. It’s convenient if the battery box is easy to remove, so that you can charge it elsewhere, but also resistant to tampering by curious passers-by. The Currie solution is to attach it to the bike with a security bolt. This consists of a thin bolt with a funny rounded triangular hole which fits its special key (that’s the “security” part), and instead of threads, there’s a tee which rotates a quarter turn into a locked position with a bit of a soft click. That’s the “quick release” part. It’s not a great system, and in fact the battery fell off the bike the very first time I tried to install it, but it did seem to work okay.

Well yesterday, I headed out for actual transportation (instead of collecting data), and as soon as I hit the road, the battery box dropped out of the frame, skittering across the asphalt being dragged by the power wires. Right in front of neighbors, too, who helpfully offered me sympathy and a ride. What happened is that the rounded triangular head had progressively stripped, enough so that it could no longer torque the bolt tight, but not so much that the key slipping in the head didn’t feel like the soft click of the bolt engaging.

It’s a bad design, and I’ll have to do some surgery to replace it, but the bike won’t work without the battery (well it will, actually, but that kinda misses the point), and the battery needs to come with the bike, which requires the battery case actually remain attached to the frame. Details, details…

I tried the no-pedal ride it until it quits range test again. Temperature 1 C, pack started at 26.4 V, tires were pumped to 77 psi. It went 9.43 km in 23:17, 24.1 km/hr avg, lowest speed on the hill was 22.0 km/hr, 21.9 km/hr. This is 40% more than last time. I have no reasonable explanation for the increase. It was colder, and I was pedalling less. The battery pack has received a couple of deep discharge cycles with the lights, but that ought to (if anything) reduce its capabilities. There was less snow on the ground, however, and perhaps the tires were more inflated, and who knows, perhaps the wind was different.

It is clear that “range” is a fairly poor metric, because of its variability. The usual recommendation is that an EV should be speced have double the range that seems needed, not just to accomodate this variability, but also to accomodate changes in plans and to keep the average depth of discharge down.

Toshiba has a press release hyping a new lithium battery, which they promise to commercialize in 2006. The big news seems to be the recharging time, which seems nearly physically impossible given that EVs take all night or weekend to charge. But is it?

First of all, the slowness of EV fueling is mostly a myth. Even lead-acid batteries, which are the slowest of all chemistries used, can get topped up to 20% SOC in 5 minutes. The limitations are things like thermal control. That’s pretty fast, though, corresponding to a rate of about 2 C (double the battery’s capacity each hour). This is higher than most lead-acid battery manufacturers recommend, although evidently you can get away with it for awhile if you’re careful. Other fast charge companies are Posicharge (which seems to do 0.6 C) and Minit-Charger, both aimed at the industrial market. A typical manufacturer peak charge rate specification is 0.4 C, which is still enough to get you back on the road after a stop for coffee. Actually achieving these charge rates for a large battery requires a fairly powerful charger, however, which is expensive, so mostly EVs have made do with slower fueling. And finally, to have a truly long and happy life, lead acid batteries like to get fully charged every once in a while, which really does take eight plus hours.

Other chemistries do better. The nickel-cadmium reaction is endothermic, so charging the battery cools it. It can be charged at several C, up to 70% SOC. Nicad charging is more energy efficient when it’s faster, too. (Nicads also like to be fully discharged, making it the most abuse accepting chemistry. Too bad cadmium is so toxic.) NiMH also requires about 1C charging.

Lithium seems to normally be around 1C, though Kokam has some material talking about the behavior of their RC batteries up to 5C. So what rate is this fancy new battery, full of nano-magic? In one minute, it’s supposed to get 80% charged, which corresponds to about 48C. Very fast, about 10x the rate of anything else (except maybe ultracaps), but only 10x faster. So maybe it’s possible. Being able to do it at low cost and with a long lifetime may be the hard part, lots of things work in the lab.

It got warm enough (5 C) to try a no-pedal ride till the battery gave out. Starting voltage 26.4, decreasing performance over the ride (first time up the hill 21.6 km/hr, second time 21.0). It went for about 18 min total, 6.65 km, 21.1 km/hr average till it cut out going up a hill. 24.7V final. This is pretty much totally unacceptable, something like a third of what I would expect. I think it’s time to declare the battery pack toast.