The braking problem – energy analysis
I’ve been just talking about regenerative braking for too long. It’s time to define the requirements.
When trying to dimension ultracapacitors or batteries for regen braking, we have to take into account the energy and power of the braking in it’s worst case: from top speed down to zero, and in the shortest time possible. Let’s start with the energy (I will post again later about the power).
So, I already set the top speed at 120 km/h, and that makes it easy to calculate the total energy generated by a full-stop braking: it is equal to the total accumulated kinetic energy of the car (well, not really, the aerodynamic drag helps the braking a little – but I’ll go for the worst case for now).
So the total energy of my car at 120 km/h is:
- E = 1/2 . m . V^2 [J]
- E @120 = 0,5 x 1230 kg x 33^2 m/s
- E @120 = 669.735,00 J = 670 kJ = 670 kWs
- E @120 = 669.735,00 Ws / 3600 s = 186,04 Wh
186 Wh is easily managed into a small battery pack, whether for acceleration or braking.
Now, for the first phase of my project, I’m only considering rear-wheel motors; this means that the full braking energy will be divided by the two axles, and unevenly so, as usual. A “typical” stock car distribution of braking forces between front and rear axles is 65% / 35%, so:
- E br@120 (rear) = 0,35 x 186,04 = 65,11 Wh
However, we know that the electromechanical system is not perfect; in fact, there are people that say regenerative braking is almost useless because of all the losses. Assuming a generator + converter loss of 15%, we get:
- E br@120(rear)(effective) = 0,85 x 65,11 = 55,35 Wh
Hmm… we spent 186 Wh to accelerate the car up to 120km/h and all we get back from braking is 55 Wh. So, the regenerating efficiency is:
- Eff br@120 = 55,35 Wh / 186,04 Wh = 0,2975 = 30%
But we still have to factor in the losses of the battery itself, and consider that the energy spent on speed maintenance (air drag and rolling resistance at constant speed) is never recovered. So, I’m looking at a 30% maximum potential autonomy increase, but I know it will be less. It all depends on the driving circuit and habits.
Ok, 55 Wh is pretty ridiculous for an EV battery. But if we are talking about a “panic stop”, it is delivered at a very high rate. How high? I don’t know, so I’m taking a guess and estimating that my car could stop from 120 km/h in 10 seconds. Just for the sake of argument. This means the average braking power would be:
- P br@120(rear)(avg) = 55 Wh x 3600 s / 10 s = 19.926,00 W = 20 kW
Although 20 kW is a perfectly acceptable power level for an EV battery discharge, it may be excessive as a charging rate. And a “panic stop” involves a much higher power at the beginning of braking than at the end, so the real maximum braking power may be several times larger than 20 kW. This means I will probably need a few ultracapacitors to take the power hit.
So, how many ultracaps are we talking about? Here I have to extrapolate even further. Luckily, my new friend Guy gave me a hand here and picked out a 70 F / 2,1 V capacitor from Digikey as an example.
I have no idea what voltage the capacitor bank finishes its charge after the braking. But that is irrelevant; let’s assume we put all the caps in parallel, forcing the voltage down to a maximum of 2.1 V – then the corresponding capacitance would have to be:
- E = 55,35 Wh x 3600 s = 143.910,00 J
- E = 1/2 . C . sqr(U) <=> C = 2 . E / sqr(U)
- C total@2,1V = 2 x 143.910,00 J / sqr(2,1 V) = 65,2 kF
And the necessary number of capacitors would be:
- N @2,1V = 65.265,31 F / 70 F = 933 capacitors
Ordering 1000 caps, the price would be 6,912 USD per unit, or:
- Total Cap Cost @2,1V = 1000 x 6,912 USD = 6.912,00 USD
Yes, 7000 USD is expensive, but still realistically achievable.
Now, something tells me I shouldn’t be getting away with putting all those caps in parallel… 
The way I see it, I’ll always need some controller in between the motor and the capacitor bank to regulate the motor braking current and the battery charging current; why can’t it work at such a low voltage as 2,1V? The only disadvantage I see is having a bidirectional “buck/boost” converter there, which would be expensive to build and complicated to manage, and probably inefficient. Another disadvantage would be the very large currents that circulate between capacitor bank and motor, and even between capacitors – also not good for efficiency.
So let’s try another approach. Let’s put enough caps in series to make them withstand 100 V and call that a module.
- N caps / module @100V = 100 V / 2,1 V = 48 capacitors / module
- C / module @100V = 70 F / 48 = 1,458 F / module
- C total@100V = 2 x 143.910,00 J / sqr(100 V) = 28,782 F
- N modules @100V = 28,782 / 1,458 = 20 modules
- N @100V = 20 * 48 = 960 caps
- Total Cap Cost @100V = 1000 x 6,912 USD = 6.912,00 USD
So, here we are at the same place again. 6.912 USD or 4.713 EUR at today’s rate.
When it comes to Energy, it doesn’t really matter how you wire the capacitors, they store the same. The small difference between the two options is due to rounding imprecision. Parallel or series is only a question of how much Voltage and Current you need for your circuit.
The fundamental question is: is 15% ~ 25% increase in autonomy worth $7.000 (€5.000) to you?
I bet most DIY builders will answer NO.
So, another new friend (John) suggested a composite solution: tie an Ultracap bank to a Lithium Phosphate bank and then to a Lead Acid bank… like this:
- High energy / Low Power store: The Lead-Acid bank will provide a large and cheap(er) energy store, but will not be able to accept very high (dis)charging power;
- High Power / Low Energy store: The Ultracap bank will provide an expensive but very high power energy store, but it’s limited capacity will not handle the larger energy bursts;
- Medium Energy / Medium Power store: The Lithium-Phosphate bank will provide a compromise between the two, allowing the energy from the Ultracaps to flow to the Lead-Acid bank only after crossing the Lithium-Phosphate.
It’s much like a large electronic circuit: you have your big power supply feeding a group of circuits, then each bus has it’s own big capacitor, then each consumer device has it’s own small capacitor. Distributed energy system. Divide and conquer. 
This way we reduce the system cost by reducing the amount of High Power components, without lowering the requirements.
As with any composite system, the problem here is the correct dimensioning. Where is the sweet spot? How large must each of the banks be, in order to make the overall solution financially acceptable?
That will be the subject of another study….
——- Addendum ——-
At the request of John, here goes the math for the full 4-wheel-drive system with industrial-grade efficiency (95%).
- E k@120 = 669.735,00 J
- E br@120(4×4,effective) = 0,95 x 669.735,00 = 636,3 kJ
The example capacitor implementation would cost:
- C total@2,1V = 2 x 636.248,25 J / 2,1^2 V = 288,5 kF
- N @2,1V = 288.547,96 F / 70 F = 4.123 capacitors!!!
- Total Cap Cost @2,1V = 4.122 x 6,912 USD = 28.492,5 USD
- …which constitutes a mere fantasy at any level.
Now, replace the caps with the LiFePo4 batteries that John suggested (3,2V/40Ah/87,99USD):
- E k@120 = 669.735,00 Ws / 3600 s = 186,04 Wh
- E br@120(4×4,effective) = 0,95 x 186,04 = 176,74 Wh
- Q total@3,2V = 176,738 Wh / 3,2V = 55,23 Ah
- N @3,2V = 55,23 Ah / 40 Ah = 2 batteries (with 27,5% spare charge!)
- Total battery cost @ 3,2V = 2 * 87,99USD = 175,98 USD
- …which is 2 orders of magnitude below the cost of the caps.
But these calculations do not account for the charging current limit of the batteries. If we take the average power for a 10 second braking and the continuous charging current limit of these batteries (3C=120A), we get:
- P br@120(4×4,effective,avg) = 176,738 Wh x 3600 s / 10 s = 63.625,68 W = 64 kW
- U br@120(4×4,effective,avg) = 63.625,68 W / 120 A = 530,21 V
- N @530,21V = 530,21 V / 3,2 V = 166 batteries!!!
- Or
- I br@120(4×4,effective,avg) = 63.625,68 W / 3,2 V = 19,9 kA
- N @3,2V = 19.883,025 A / 120 A = 166 batteries all the same.
- Total LiFePo4 cost = 166 * 87,99 USD = 14.606,34 USD
- …which just means that the chosen batteries are too big: they support too little power and too much energy, and the usual (H)EV manufacturer approach of many small cells is the only way to support higher power.
It remains to do a recalculation with a smaller LiFePo4 cell size (like A123’s cells that can charge at 4,5C)… but I bet a composite solution with ultra caps and batteries must be devised…
Meanwhile, I’ve come across a new type of Lithium chemistry battery that is (almost) commercially available and promises even better performance than LiFePO4: Lithium-Titanate (LiTO). Toshiba’s “SCiB” 24V/4.2Ah pack apparently delivers the same performance as a capacitor; they state charging currents in the order of 12C, which is far better than the available LiFePO4’s 3~5C.
So, redoing the power math with these SCiBs (24V/4,2Ah/50A):
- P br@120(4×4,effective,avg) = 64kW
- U br@120(4×4,effective,avg) = 63.625,68W / 50A = 1,3 kV
- N @1272,51V = 1.272,51V / 24V= 53 batteries
- Since the price has not been published yet, I’m going to extrapolate the total cost. I’ll imagine these new batteries will cost 25% more than an equivalent 25,9V/4Ah Lithium-Polymer pack: 1,25 x 200 USD = 250 USD.
- Total LiTO batt cost = 53 * 250 USD = 13.250 USD
- And they would pack a nice energy capacity of 53 x 4,2Ah x 24V = 5,3kWh, which would be sufficient for 5.342,4Wh / 176,74 Wh = 30 full-stops/accelerations 0~120 km/h!
Well… 13.250 USD in batteries is better than 28.500 USD in caps, right? Especially if the batteries have a life expectancy of 3000 cycles…
The conclusion to this article is that it is trivial to store the Energy required for an acceleration from zero to top speed or resulting from a full-stop braking; however, even when contemplating very modest values of average braking power, it is quite challenging to harness the full regenerative braking Power.
I will post again exclusively on this subject. Please keep your power questions and suggestions in the box until then.
September 28, 2008 at 12:34 am
Regenerative breaking is indeed an interesting problem… because it seems to not to have yet an interesting solution (discussion started in the comment’s zone of http://myownhybrid.wordpress.com/2007/12/10/motor-design/ but it’s better to continue here, I guess).
There are no cost effective solution today, as far as I know, and since we want the system to reach as many persons as possible, it has to be cost-effective. Have you calculated how much more Kms of range can you get with $7000 worth of Lithium-based bateries instead of spending that money in ultracaps?…
I would do the following, which I can classify as both cheap and easily upgrade-able
: use a high dissipation resistance (can we get away without one?) to dissipate the heat far from the wheels, then wait for a more cost-effective solution. Meanwhile you can try other low-cost solutions, or starting from this zero-recovery method and progressively think about methods to recover growing parts of the energy lost as heat.
Dissipating the energy in a resistance has in fact many advantages, except the recover of the energy. For example, you can dissipate lot’s of it, and during the entire break period; remember that the voltage on the generators will fluctuate, it’s not constant (without regulation). It’s cheap. You can choose the place to heat. You can leave the resistance in the system even when there are energy-recovery methods implemented, as a security measure and to guarantee that you can always dissipate (break) the energy until the very last drop.
I have some other thoughts concerning this subject but they’re still too raw to share.
September 28, 2008 at 12:48 am
By the way, the ultracaps method (this is actually a method extensively used in electronics, tipically under the name “decoupling”; also applies when using “faster” battery technologies instead of ultracaps) can also be used to deliver to the motors a power peak that the battery pack cannot (another application of the same principle, very “fashion” this days, is the usage of big capacitors (a few F) at the power input in big car amplifiers, used by car “tuners”; the capacitor provides to the amp the peak power the alternator/battery cannot achieve).
When building a(n) (ultra)capacitor bank, you also need to take into consideration the maximum current the caps can withstand. Although infinite in theory, in practice it’s finite (for instance, the package lead’s thickness imposes a maximum current).
September 28, 2008 at 12:54 am
I guess Guy’s “electric egg cooker” in http://myownhybrid.wordpress.com/2007/12/10/motor-design/#comment-279 refers to the “energy dissipation in a resistance” method.
September 28, 2008 at 4:15 am
you also need to consider stopping on a grade, or braking down a hill.
September 28, 2008 at 4:40 am
you did the energy/cap calcs based on just the 35% from the rear … might be nice to consider storing the entire KE (less losses) back in the storage system … and raise the eff a bit.
September 28, 2008 at 4:42 am
because copper losses are R*I^2, there is an eff case to be made for not only starting slower, buck braking slower if you want to save energy losses from copper heating.
September 28, 2008 at 10:18 pm
NJay:
And “decoupling” was exactly the term I wanted to use, but I couldn’t remember it when I wrote this!!! :/
I agree with you and Guy – I think a high-power resistor will be inevitable (because of storage system failure or overcharge), and as you say, it is intelligent to put it in right from the start and add-on other technologies to recover the energy, leaving it there for the “fail-safe” mode. That allows us to eliminate the mechanical brakes right from the start, even while we wait to have ultracaps or whatever. Nice tip, you guys.
John:
I don’t see what we can do in terms of system design to cut down power losses, other than trading higher currents for higher voltages, and eventually lowering the maximum power limits…
Yes, I’ve been thinking about braking on negative inclination road. It certainly raises the bar for the maximum power requirements; I don’t know if the energy requirements will change too much, though. Needs some math to clarify (I don’t expect much linearity in a braking event).
Ok, I think I’ll do the math for a full 4×4 system and see where that leaves us – but I bet the cap cost will be prohibitive.
What efficiency would you consider more realistic?
As to the copper losses, you’re right, less power means less losses; but isn’t that strictly a question of usage? If you want less losses, don’t step on the pedals so hard…
September 28, 2008 at 10:20 pm
My next post will be (in principle) a reprise of this analysis, but from the point of view of Power instead of Energy. I think it is very interesting and absolutely critical for the correct dimensioning of the whole system – and it may give me precious hints for system design.
September 29, 2008 at 8:20 am
I live in an area where people drive out of the mountains every day to work, where one person had to retire his ev daily driver after moving 1,000 ft higher. Others drive up 4,000 ft to go to work each day, and return in the evening. Others have to climb over a 2,000 ft “pass” to go to town, and again on the return trip home. There are 300-600 ft rolling hills.
I don’t have very much of this between my home and town, although there is significant elevation changes in the area I drive for work each day … including all of the areas above each week. What this brings to light is trip planning can effect decisions about when, where, and how much charge is optimal in an ev at any point in the planned trip.
For instance, someone that lives high, needs to only recharge for the net losses each day, so there is reserve capacity in the batteries to absorb the regen power descending the mountain
at the beginning of the trip … to be “full” when reaching the bottom each day. Someone that lives low needs the opposite, and will need to be full when leaving. The net altitude change in a daily trip plan may well exceed the energy available in batteries, which for a hybrid changes when/where the system should be full/empty to maximize fuel savings. These altitude
changes in daily driving place a much higher emphasis on when/how ReGen is use, and what target battery capacities are as a function of location and trip planning.
I’ve been looking at integral GPS tracking with advance trip planning to provide the system clues about how/when to use ReGen. Looking only at KE, and ignoring PE, can for some folks co
mpletely miss the target, forcing energy to be wasted as heat, simply because the capacity isn’t there to store the KE+PE when it’s available to be recovered. Many cities here have time
d stop lights, synchronized inside the city, or independently predictable. These clues can augment decisions about speed, braking, battery levels and mix of storage technologies.
Lead acid (both flooded and SLA/Gell) technologies only have a few hundred cycles in their life, so they do not make a good choice for ReGen storage during a trip (stop/start or altitude changes). They do make a good choice for bulk energy storage to extend ev range past the daily routine as an exception, simply because they are dead cheap.
LiFePO4’s of certain constructions are rated as high as 25C (charge/discharge currents above normal rating – see RC Toy batteries). Some of these batteries have a life rating (to 70%) of 8000 cycles as long as they remain above 20% charge, or the currents remain below 1C. Some have about 1/3 their rated life when operated below 20% charge, or cycled above 5C extensively. So, it might make sense to mix and match different LiFePO4 technologies to balance the driving demands with the available battery technologies. Like, have a relatively small bank of
25C rated batteries to meet peak current demands, with long term average currents taken from a technology the has a high cycle life with lower currents, as a balancing tradeoff.
Maxwell Boost caps are relatively expensive, but for stop and go ReGen, are dirt cheap compared to any battery technology to absorb and release KE+PE repetitively each day, simply because they do not degrade rapidly with cycling that destroys other battery technologies.
Each of these technologies needs to be matched to the typical weekly trip planning, costed out for their life expectancy, and amortized back to a per trip cost for comparison.
September 29, 2008 at 8:24 am
Of course … YMMV
September 29, 2008 at 9:03 am
As for efficiency, we see solar racers running 1KW motors at 96-98% eff using poly phase iron less axial motors. They cheat the R*I^2 copper heating losses by using Litz wire stators to minimize (almost completely remove) eddy current losses and skin effect losses on switching transients. For a particular motor design the losses are probably not symmetrical between motor and ReGen modes.
The Back EMF will almost certainly not match battery/cap charging voltage, meaning there are likely to be some level shifting circuits and losses for ReGen. Ditto for motor modes, where back EMF impacts required voltages and currents. Excessively high currents, have excessively high R*I^2 losses, even when PWM ‘d to a lower average current. Losses can be controlled bymatching motor driving voltages to the minimum votage required to defeat back EMF while providing just enough additional voltage to produce the minimum required currents to match power demands. The actual motor PWM switching losses and R*I^2 coper losses may well exceed losses for a variable voltage drive system, depending on how the level shifting is done, if the PWM currents are too high. Ditto for ReGen design.
Interesting solutions to this, are variable “banking” of the storage array … operate in 12V mode (all batteries in parallel) at low speeds when back EMF is low, and switch to 24, 36,48, … 96V to match motor driver voltage to just above the back emf of the motor as speeds change. Using a few dozen 10mohm fets in parallel, have much lower power losses than PWM level shifting circuits or PWM motor drivers.
September 29, 2008 at 9:51 am
How about the flywheel method, would it be cheaper? Instead of trying to store the breaking energy in the batteries (which today we probably won’t be able to do), we store it in the flywheel and use it, let’s say, to re-accelerate. I didn’t mention the (well-known) flywheel before because my head was thinking “mechanical coupling to the wheels”, but it doesn’t have to be mechanical; the energy generated by the vehicle’s motor(s) would be directly used to drive a flywheel electric motor, and vice-versa. Extracting the energy from the flywheel could be to directly drive the wheels’ motors or to slowly charge the battery bank; either way, the energy should be used asap to reduce losses.
September 29, 2008 at 11:04 am
That sounds like a very good idea. It is certainly worth investigating.
And he is right, from the point of view of efficiency. But if we have to compromise, it is worth a look…
I give it 10 minutes before h0tR0d screams “AAAHHH! flywheels have to be mechanical! electromechanical conversion sucks!!”
Besides, electrical flywheels have an advantage over the mechanical ones: Transmission. You can place them anywhere in the car, because the wires make for a very flexible transmission line; the same cannot be said about fully mechanical flywheels somehow attached to the wheels.
September 29, 2008 at 11:32 am
John:
Wow…. slow down, I’m getting dizzy!
* GPS integration: yes, it had occurred to me that the terrain could be easily included into regen planning via GPS, just like you described (Lisbon has 7 great hills too); however, I had never thought about factoring in the traffic lights – it sounds logic, but won’t it be a bit of a mess? I doubt that the possible traffic light determinism will be comparable to the simplicity of a never changing terrain. On the other hand, there could be several possible routes over inclined terrain… So yeah, in the end we are looking at a routing graph with several possible “energy weights” caused by terrain and traffic light (a possibly other) options – a problem in the family of the traditional “traveling salesman”.
This is a good candidate for a “machine learning” system – something that can build the energy and routing graph and solve it for optimization, presenting the solution much in the same way a normal GPS does today, in a “turn-by-turn” basis, for example. The system would take into account not only shortest distance, but also elevation differences and stop-start accelerations, and if the driver accepts the proposed route, it would control the energy flow (via power controller hints) as you suggested.
I’m going out for lunch, I’ll get to the other points afterward…
September 29, 2008 at 12:45 pm
We could play a bit with other, probably cheaper but with less capacity, storage methods, such as (strong, thick
) springs. Same charge/discharge method: electrical.
September 29, 2008 at 12:54 pm
NJay: now you’re being a little “too innovative”…
I’ve got enough on my hands with the wheel motors and control system, I wouldn’t go into inventing another system…
But in the future, you never know…
September 29, 2008 at 2:57 pm
http://www.public.asu.edu/~mgomes1/Bungee_Bike/Web%20Page%20FINAL%20Group3.htm
September 29, 2008 at 4:28 pm
Sort of automating “Hypermiling” for ev’s
Traffic lights create two interesting decision points into the route specific energy management plan. First is avoiding unnecessary start stops, second is making sure that the stat/stop rates mesh with predicted light state. For instance, if we always start too fast/slow, it’s possible to ALWAYS meet the next light always at red, where a very small change will cause arrival always at green. Above a certain threshold, increased accelleration/deaccelleration rates have a predicatable R*I^2 copper heating loss. Likewise, certian grades have a predicable loss/gain in route choices.
In this area, the “sectional roads” (major roads on a 1 mile grid) have a very predictable, synchronized schedule, with very few “demand” lights. Once you get off the sectional roads, t
he pattern can shift to demand only, which is still synchornized with the nearest sectional road. The cycle time for these systems are fixed, and skewed (phased) to best manage “platoon
s” of cars that are created by the synchronized lights.
This is one public policy area where smart cars can beniffit from traffic management data, if designers start working with traffic departments in our cities.
September 30, 2008 at 7:04 pm
http://www.edn.com/article/CA6594105.html?industryid=47042&nid=2432&rid=808729273
Anybody know any more?
September 30, 2008 at 8:49 pm
never thought of electric flywheels but its a good idea. Although the flexibility comes with a price, less eff and more weight .
And the typical driver has to be “re-educated” to fully take advantage of the EV’s…
regen braking has to be think thoroughly at downhills. Eff wise, its best to let the car roll…
October 1, 2008 at 5:53 am
There are many difficult choices in automotive energy trade-offs, where many excellent technical choices have with them horribe social costs … like safety. Energy wise the ideal car weighs about 500lbs so very little wast energy is required to start and stop it. The reality is that cars under 2200lbs tend to kill people, especially kids, when they hit stationary objects (parked cars, trees, concrete, rocks) such that the car decelerates an 50G’s or better ripping internal organs apart even with air bags.
NiMH and LiIon batters were great energy storage devices in the last decade, but exploded when crushed. H2 fuel cells have similar problems, with H2 storage. So do high energy flywheels in autos.
At the end of the day fuels like diesel are near perfect … easy to ignite in an ICE, very difficult to light with a match. For EV’s and Hybrids to sell well, they need to have a very high perception of being safe … people don’t die when they drive them.
In the US, the CAFE laws forced a huge number of light cars on the market, and they killed a lot of people in single car accidents hitting stationary objects in the 1990’s. Despite CAFE laws, the public started buying affordable larger cards with relatively higher safety.
A huge mistake for alternative energy autos, would be to create a new round of un-safe cars/trucks that send the public the message they are all unsafe again.
The biggest single technology break thru that will enable EV’s and efficient hybrids, is LiFePO4 batteries, that do not go up in flames or explode with crushed …. and at the same time can have a 10 year life when managed properly in combination with supporting storage technologies.
As for coasting down hill, the M*V^2 problem with moving air wastes the energy just as good ash using brakes … I’d suggest that having reserve storage capacity to recover it will save that energy, and allow it to be used after the down hill section of road.
October 1, 2008 at 6:14 am
Very high density LiFePO4’s are making it to market … checkout http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=250294078368 that are optimized for bulk capacity, and others like http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=280271237139 that are optimized for cycling at high currents.
October 1, 2008 at 6:31 am
Long term policy will be shaped at the ITS conference next month:
http://www.itsworldcongress.org/images/ConferenceProgramPDFs/15th_World_Congress_Plenary_Sessions_IWC08.pdf
http://www.itsworldcongress.org/conference-program.html
be sure and follow the papers that are being presented in this area.
October 1, 2008 at 11:43 am
So, John, you’re saying we should _only_ look for technologies that can simultaneously cater for all of these requirements:
- are crash-safe (fire/explosion/corrosion wise);
- have low specific mass (also safer, as well as more efficient to transport);
- have high (reversible!) power density (to allow any accel/regen bursts);
- have high energy density (to allow full recovery of regen, even with big terrain differences);
And you suggest only a combination of LiFePO4 + ultracaps is today’s response to these criteria?
At least the weight criteria rules out lead-acid…
Personally, I believe that the “biggest single technology break thru” to enable EVs will be the “EESU” – EEStor’s ultracaps that have energy densities in the same order of Lithium batteries. They combine the capacitor’s power, weight, and life cycling with the battery’s energy capacity. A one-stop solution to all the problems. Unfortunately,
it looks like they are taking longer to deliver than expected…we will have to wait for the end of 2009 to see the ZENN cars and LEVC bikes coming out of the factory line with EESUs onboard, because EEStor will not sell it otherwise.October 1, 2008 at 4:22 pm
John:
Regarding the voltage management issue (PWM vs. “Banking”)…
I think I see what you mean. But without proper modeling, I’m really not sure you can pull that off with real advantages…
I see 3 problems there:
1 – the system you described is discrete by nature, i.e. it supplies voltage in discrete analog “steps” (of 12V in your example) to the motor. This means there is no control of current within each step (because you don’t want to use PWM) other than the difference between supplied voltage and EMF, and circuit impedance. So, since you’re not regulating voltage (within the step) and you’re regulating current (by switching), wouldn’t there be excess current in many cases?
2 – “banking” the batteries like that looks like a good idea, but if you want to have a fine-stepped granularity of the voltage, then most probably your battery combinations will not have much symmetry – and so each battery will have different life cycling and greatly heterogeneous charge state. If a BMS is already a tough cookie, in that context it is a nightmare.
3 – Many years ago, I rode on an EV built by a Professor of mine. It had the system you described: 6 x 12V batteries (Lead-Acid), whose connections where managed by contactor switches, and these where wired to an ingeniously hand-crafted accelerator pedal containing a few micro-switches. It worked well, but it was unacceptable according to today’s comfort and safety standards: the 12V steps where too large, and the powerful Samarian-Cobalt permanent magnet DC motors “jumped” ahead each time a switching step was done. Of course, we’re not talking about the same kind of motor here, but the principle is still valid – it generates enormous torque ripple…
October 1, 2008 at 5:40 pm
Hi Vasco,
If “we” are hobbiests building our own EV’s and Hybrids, there are no restrictions on the technologies “we” can use, experiement with, or even advocate for hobby use. If “we” are engineers building volume EV products, there are explict and implict goals which need to be met to mature the EV/Hybrid industry, or at least remain competitive when other suppliers are doing the “right” thing.
LiFePO4 batteries are shipping in high volume today from a number of suppliers, and have demonstrated that you can drive a nail thru them without explosive discharge. Part of this is because the energy is stored in a controlled chemical change, that is more difficult to extract when the battery is crushed or deformed. This is similar to the built in safety of lead acid flooded batteries, which may briefly explode from a small amount of H2 gas, but the bulk of the energy is lost when the container ruptures venting the electrolyte. Capacitors however, have a charge state that is always willing to become a high temp plasma once an arc is started, that will vaporize the electrodes and out gas a large volume of hot metal vapor in the process at these energy densities. The ability to control that charge energy goes away when the capacitor is crushed or deformed. That hot metal vapor gas can be contained with some engineering effort, and is certainly much safer than traditional LiIon batteries going up in flames, assuming it doesn’t contact other materials nearby which will explode/burn at plasma temps.
EESU’s might start shipping in low volume this year, and it’s still unclear when they will become high volume and sourced from multiple mfg lines. It is also unclear what the relative pricing will be long term, as their technology is much more expensive to manufacture than the relatively low tech LiFePO4 process. There are already production LiFePO4 processes with demonstrated lifes at 8,000 cycles, far enough out that battery management electronics become the primary failure mode. One can expect increases in cycle life to continue to improve for afew years as the technology matures.
Long term … 5, possibly 10 years, there could be an advantage of EESU’s over LiFePO4 … this year, and most likely next, LiFePO4’s are available in high volume to build products with, with multiple suppliers, while EESU’s are still being talked about.
As for banking, I agree it’s difficult to control with one or three phases, just as you discuss. The problem however largely goes away with 20-70 phases, where the number of active phas
es sums to the desired torque … one, or a few, when low torque is needed … or all when full torque is needed. So there are two torque control variables (current plus active phase count) which as a product, produce hundreds of fine grain discretely controllable torque points without PWM. PWM isn’t however precluded, and is always available for a third degree of freedom for selecting torque (with the implict higher switching losses).
October 1, 2008 at 5:47 pm
John, you are american right? Only an american would say that “The reality is that cars under 2200lbs tend to kill people, especially kids, when they hit stationary objects (parked cars, trees, concrete, rocks) such that the car decelerates an 50G’s or better ripping internal organs apart even with air bags.”
Don’t take it personally what I’m saying but the truth is that’s a lie that american auto manufactures use to tell people just that… I saw a Smart survive a crash with a 20′000lb truck! about decelerations (50G’s or whatever) has to do crash box’s or zones inside car structure (which in fact the Smart has very little but many european car’s weight less than 2200lbs and are very safe). In the 90’s that could be a problem but not anymore… But don’t worry, the battery packs will add a lot more weight!
“At the end of the day fuels like diesel are near perfect … easy to ignite in an ICE,” not really that easy, in fact that’s one off the major problems in diesel engines…
October 1, 2008 at 6:29 pm
I am an American, and I have also had direct experience analyzing with the real on the street nation wide death statistics … the numbers are getting better with improved safety design, but they are real, and far from FUD from the auto industry. I will be happy to post the US statistics supporting these numbers if I can find the last paper I wrote on this topic.
October 1, 2008 at 6:35 pm
as for diesel engines, I’ve been driving them for 20 years, for over a million miles. I do my own engine rebuilds, both gas and diesel, and have for 40 years … more than a 100 engines. Current drivers in the family are ALL diesels … GM 6.5’s in the heavy trucks, and VW TDI’s for commuter cars … 48mpg in a VW bug that is peppy and fun to drive is my preference right now …. nice zippy turbo performance.
Diesels are a whole lot less picky than trying to get the fuel air right on a gas car.
October 2, 2008 at 2:05 am
Still looking for the report I wrote 10 years ago … didn’t find it online yet, but I did find a good recient report based on the NHTSA current data. If you need the raw data behind th
is, it’s normally online (or at least used to be when I was working with this data 10 years ago).
http://www.iihs.org/sr/pdfs/sr4003.pdf
Notice on page 7 the summary data regarding weight … it’s not really about the raw weight, but about the integrity of the frame in the front and rear, and the fenders which provide the crumple zone … lighter cards stop faster, higher impact G’s on occupants. When you reach into the raw data by make and type of accident, what is more striking is the death rate by material and construction, making a number of very popular cars death traps with death rates several times the average of their class. If you can, track down the raw data, then spend a few months doing research in the junk yards about fatal crashes … see what failed. Doing this first hand quickly gets past the BS and finger pointing. You will have a really good idea what car your wife and kids should be driving, and what the risks are should some drunk hit them.
October 2, 2008 at 3:52 am
Public US raw data is available here in lots of detail … which is not filtered by any auto maker:
http://www-fars.nhtsa.dot.gov/QueryTool/QuerySection/SelectYear.aspx
I’ve had to mine this data for clients on and off since the data base was created … it’s very useful.
Is there an equivalent data base for Europe?
October 2, 2008 at 12:50 pm
So, should we all be driving cars with more than 4000 lbs (1814 Kg)?
How does the “death per million” relates to the number of cars per “weight class” on the roads?
October 2, 2008 at 12:52 pm
Ok, I got the 2nd part, is “per million *registered vehicles*”.
October 2, 2008 at 2:51 pm
NJay: the unit is “registered vehicle years” (vehicles X years they are registered) – otherwise it would be difficult to make a fair comparison.
Although I find this discussion extremely interesting and even eye-opening, I regret the fact that the topic has been hijacked into “The crashing problem – energy analysis”.
Well, if you can’t fight them, join them….
Here’s a link about it.
Smart ForTwo almost gets the “Top Safety Pick” award from USA’s IIHS
I can see there are two basic rules to conclude from all the data:
1 – the larger the vehicle, the more space there is for “programmed deformation materials” to absorb crash energy; the smaller the car, the more difficult it is to “spend” the crash energy in ways that are harmless to the occupants – there just isn’t enough material there. The same reasoning can be applied on the inside of the vehicle: the bigger it is, the more free space there is for slower deceleration of the occupants (after the vehicle has stopped absorbing energy) – seat belts and airbags will not help you if they confine your deceleration movement to a mere couple of inches (your brains turn into soup).
2 – the heavier the vehicle, the more relative energy is stored in the vehicle’s mass (as opposed to being stored in the occupant’s mass), which means the vehicle will absorb most of the energy; now if that actually protects the occupants I don’t know – I suspect not, because at any given constant speed the energy of the occupants is always the same, and the fact that the car weighs more just means there will be more damage (to itself and to others on the outside) – it should not have influence on the occupants. The only way I see it tip the balance is in the case of collision into moving objects: heavier cars will push lighter cars along with them, making the deceleration of their occupants a little lower (but making the occupants of the lighter car pay for it).
So, I personally rule out weight as the main factor of self-safety (although it is very important when evaluating safety towards others); I’d expect that the key indicator of safety would be a ratio between the car’s size and weight, not just size or just weight. Given the same engineering quality, I’d expect that vehicles with the same “density” would perform the same way in terms of safety. Heavier cars need more (crumple+inside) space to decelerate safely. Of course, the occupant’s mass must be figured into the math as well.
The “Smart ForTwo” (the quintessential example of what the “american way of life” is NOT) is quite revealing as a safety design exercise: it shows that the inner volume of a car can be more important than it’s outer volume when it comes to protecting the occupants. The Smart “replaced” the large outside crumple zone typical of auto engineering with a generous inside volume that allows for safe deceleration of the passengers trough internal energy absorption mechanisms: safety belts and airbags. The fact that it didn’t do so well in the whiplash test is probably related to this, but I guess it could also be engineered to comply.
The future common car should be like the Smart: lightness promotes all-around safety (lower energies are involved). Heaviness promotes more damage, and an uneven (and socially unfair) distribution of it.
John:
Now I understand your motor/controller design intention… playing with those 3 degrees of voltage switching freedom certainly would give tighter torque and speed control, possibly equivalent to “simple” PWM; as to efficiency, I really can’t decide without looking at a model or working prototype – “the devil is in the details”, or so they say.
October 2, 2008 at 4:15 pm
Njay writes “So, should we all be driving cars with more than 4000 lbs (1814 Kg)?”
Not at all, especially if that is 2,500 lbs of SLA’s in a 1,500 lb small car that doesn’t have enough materials to protect the occupants.
There are plenty of mid-sized cars that have safety factors nearly as good as heavy ones. There are very few cars under 2,200 lbs that demonstrate similar low rates of death and injury. There is a high parallel between serious injury and death rates.
The original point is still what is important, that the industry shouldn’t become obsessed with trimming weight just to make the M*V^2 part of the KE better, especially when non-structural elements like batteries are providing the weight. We need SAFE energy efficient EV’s and Hybrids.
Since Hotrod wasn’t aware of the real statistics, we took a little side trip to correct a few points. Partly from the ignorant American bashing, and partly from already having spent hundreds of hours doing this safety research before.
I’ve shared privately with Vasco some ideas around my own EV/Hybrid conversion project, and the motors I will be building this month for it. My project car is a 1965 Corvair Corsa that I used to race. I have several of these project cars stored, and plan to make an EV/Hybrid daily driver out of one, and a “fun” sports car out of another. My current plan is a wheel motor on all four axles, using a low cost axial design that is both efficient and easy to build. The first round design will be just getting something to work well, then I plan later to scale up the power so that the “fun” car is as “sporty” in power as it used to be. It’s old power plant was a turbo charged large valve 140 engine that had around 250HP, and at one point in the late 1970’s I did a 240 mile road race in about 2hrs and 15min, pushing better than 150mph much of the way. It will probably be reworked to autocross in the EV conversion, rather than sustained high speed racing.
I’ve been highly focused on efficiency, where iron and copper losses are a big part of the heating. The design I’ve been working with is an ironless stator with gap flux around 1T to minimize the amount of copper required in the stator, partly to get the gap narrow, and partly to manage the stator resistance. Stator heating (copper losses) is R*I^2, where the optimal efficiency points are very low R and very very low I. Using four motors divides the current at each, for the same net total power, by four to significantly lower the I^2 component of the copper losses.
The design is also based on a large number of phases to lower the current per phase, to also reduce the I^2 component of the copper losses. So the balancing act is to keep resistance low while increasing the number of phases with a low current. this allows rewriting the copper losses to N*R*((I/N)^2) where a lower amount of heat is distributed over a larger area of the stator to aid cooling. Like the Litz wire solar racers, the design is optimized to reduce the skin effect and eddy current losses as well. This works because the resistance is a linear function with cross sectional area, while the power lost is the square of the current. There are some additional gains, in that smaller cross section wires dissipate heat better than fat wires, so you can put higher relative currents thru smaller wires for their size without fusing, to better handle short peak loading for start/stop.
The initial prototype design is 32 or 64 phases, limited to 6A per phase, with a peak power of about 10KW. The current design operating point is about 1.5KW at 90% efficiency including electronics losses, with some hope to improve that to about 96% using better MOSFETs in the H-Bridge’s.
It had started out as a 13″ design, which I have decided to expand to 16″ for more torque. Current plan is to build the initial prototypes around the wheel bearings, and later to integrate the wheel, bearing, motor and disc brake as a unit.
Initially it will be EV only, and I will remove the air cooled Corvair motor and use the space for batteries. Later I plan to use a VW TDI diesel engine/trans for a hybrid configuration.
Once the prototypes are built and tested, I will post DIY details on my own website for the project.
October 3, 2008 at 7:13 am
The Smart cars are interesting, but not selling well here, as their fuel economy is marginal at best — the 33 mpg city is middle of the pack, not even close to the fuel economy leaders. Our VW TDI’s easily get 46 city and about 50 hwy at 65 mph … even got 47 mpg average on a 1,800 road trip at 80mph. It will be interesting to see how they stack up in practice when the stats are in for injury/deaths after two years (assuming they sell enough to get some valid stats).
I will see if there are any salvaged yet in this area, checkout the damage and police reports. A trip to see a few wrecks is always more valuable than the dealer ads.
October 3, 2008 at 12:29 pm
John:
Thank you for sharing your design strategy with us. If you don’t mind, I’ll be collecting these (and other) good pearls of knowledge and compiling a new page of “design tips” here in the blog.
Just a comment on your choice of names: I believe what you call “phases” does not really map into the common usage of the term. I say this because you speak of switching more or less “phases” simultaneously to control torque, and as far as I know, the definition of phase implies it has distinct timing from all the other phases. So, if I understand your design correctly, you have 32 “feeds” or “coils” in the stator, but since they are controlled in redundant sets, the number of time-distinct phases is actually a little lower.
Question: how lower?
October 3, 2008 at 2:16 pm
John, Smart MHD (not available in the USA) can easily pull 71mpg
In Smart UK you can read:
Fuel consumption (urban cycle) in mpg [1]
57.6
Fuel consumption (extra-urban cycle) in mpg [1]
72.4
Fuel consumption (combined cycle) in mpg [1]
65.7
In Portugal (and other countries) we have Smart CDI (turbo diesel engine) without MHD and it can pull 74mpg (advertised). I’ve got friends with this car, and they frequently pull 90-100(!!!)mpg. Imagine with MHD!!!
http://www.smart.com/-snm-0135207752-1221795829-0000011376-0000000000-1223043295-enm-view/engines/mpc-en_en_EN_EUR_urn:uuid:e0a1fb03-d93b-5af7-80ab-7c81f0ff63f2
October 3, 2008 at 4:25 pm
Hi Vasco,
They are real phases … normally each is different in angular time, and must be commutated independently for each pole crossing. Depending on the choice of magnet in the rotor, there may be as many as 95% of the phases in one, or two, magnets flux gap at the same time, with the remainder in the inter pole gap . This is the case for tightly packed angular wedge shaped magnets. For bar shapes there will also be some phases that have partial flux at the edge transitions.
Each phase has an independently controlled MOSFET H-Bridge switch. You can choose to wire and drive them independently, or in a more typical Delta, Wye/Star configuration, where the latter two may require independent isolated supplies for each phase.
The considerations for clustering them into two or more banks, which are dynamically configured in a ladder arrangement to match back EMF to supply voltage is an optimization to allow nearly full regen without having a high current voltage converter when the back EMF voltage per phase drops below the battery voltage.
With them banked, the stacked back EMF is Vemf*BANKS in the ladder array. At high speeds, the ladder is retracted so that the stack is one high, and at low speeds the ladder is extended to BANKS high. The battery array can also be be banked in the same way, which can be very useful where LiFePO4’s are available at the unit cell level, so the lowest Vbat is 3.2V.
The only other alternative is to use a buck/boost converter to boost voltages up/down depending on speed, which would have efficiencies of between 60-80% … banking avoids that.
Where the speed causes Vemf*BANKS to falls below the smallest Vbat value, regen ceases in this model, and EBraking requires dumping a relatively small amount of heat somewhere. Given that M*V^2 rapidly approaches zero, very little energy is wasted when Vemf*BANKS falls below the smallest Vbat value.
It’s the buck/boost converter losses that make ReGen a poor performer in other designs. I believe the banking solution is original, novel, and highly practical over other level shifting choices.
There are similar efficiency gains when powering the motor, by setting the Vemf*BANKS height such that coil currents are limited to (Vbat-(Vemf*BANKS))/Rcoil. As Vemf increases with speed, Vbat is increased as well, to control coil currents without PWM or lossy resistive strategies (transistors in a servo amp configuration). The H-Bridge switching frequency (and related losses) is then only the AC frequency of the motor … typically under a few hundred Hz at the fastest speeds, and insignificant at low speeds.
October 4, 2008 at 12:06 am
John:
If I got it straight, you’re using the “variable banking” scheme in both sides of the problem: power supply AND power consumption!… You put the battery cells in parallel for the higher start/stop currents, and stack them in series for the higher speed voltages; and on the motor side you do the reverse: put the phase banks in series to lower the start/stop currents, and parallelize them to lower the higher speed voltage! By managing both sides simultaneously you can always keep the current AND the voltage within comfortable limits…
I only have one word: WOW… this has to be the most elegant traction power design I’ve seen (although I am a newbie in this game). It should cater for efficiency AND power at the same time. Very well thought out!…
Question:
- Aren’t the batteries going to get a raw deal? I mean, their usage is going to be biased, and not very symmetrical. Unless, of course, you are also using the same kind of H-bridge strategy on the batteries, which would allow you to balance them during charge and discharge at will…
October 4, 2008 at 2:50 am
The H-Bridge ladders are used on both storage and load sides, and has some optimal points on both sides to balance current draw per bank, those are where there are the same number of banks on each ladder level. The unbalanced points are used only for transition states, because they are current constrained by the narrowest bank width. The design of the ladder management software should include an e-meter for each bank to track charge state, and a balancing algorithm in the folding operation.
On the load side, the current per bank is controlled by the number of phases activated in that bank, so that the same number of phases are active at all ladder levels. It’s not quite that easy on the storage side.
As I said before … the goal is to operate the motors as close to 96%, or better, efficiency as possible, except where demand requires short term currents that have higher losses during start/stop with an urgency that justifies the losses.
When you carefully examine the efficiency of motors, there is a 3 or 4 degree optiminal solution space offered by this strategy. You simply can not get there with single valued supply and load voltages, where current is modulated to control torque/power delivery. The mismatch is always the motors K function which linearly decays with rpm due to back emf.
See the motor sim at http://www.ebikes.ca/simulator/
October 8, 2008 at 1:29 pm
John et alia:
I’ve written an addendum at the end of the original article, with considerations on a full 4×4 vehicle with 95% conversion efficiency. I also factor in the latest Lithium ion battery technology (Lithium Titanate) because it is the one with the highest power density and will become commercially available soon.
I’ll be writing a new article dedicated to “regenerating power” soon, so save your best shots about Power for later.
October 10, 2008 at 8:26 am
Most start/stop cycles for our family’s driving are under 35mph, about 30-60 per day depending on driver and route. This is the speed range most people brake in. Higher speeds typically coast down to this braking speed, with some notable exceptions, including poor driving habits. So this is the typical high current charge/discharge energy storage needed to be reserved for regen.
Doing regen with a battery technology that has a nominal 3,000 cycle life means that they will show wear/damage at around 50-100 days for this capacity. Over a 10 year life that would be 30-60 battery sets. You can provide excess capacity, with a slower degradation rate, at about the same annual cost. This portion of storage needs to be cycle cost effective, at high currents for both charge and discharge. This is where caps, at this storage level, start to show better reliability and cost at 10 years. As do a small bank of LiFePO4’s optimized for high discharge/charge … like http://www.lifebatt.com/cellspecs.pdf
LiTO’s are not new .. the patent filings go back more than 11 years. Volume production and cost are a problem, or they would have dominated by now.
In 1982 we all wanted to skip the stop and go driving, and invest in Moller’s skycar that was demonstrated in ground effect near where I worked at the time. 25 years later, and still in ground effect. http://www.moller.com/skycar.htm
October 11, 2008 at 8:53 pm
John, an alternative to need to replace batteries frequently is to add sufficient energy storage and build a suitable discharge/recharge cycle so as not to deep drain the batteries. The cycle life of a battery only comes into play if you drain the battery over 80%. Or at least that is what I have read. By planning a recharge cycle before deep draining, you can dramatically increase the battery life. I know of public use EVs in Canada that have been running for more than ten years on the same set of batteries.
October 11, 2008 at 9:29 pm
Yes, that’s why both the Prius and the Civic use a very small charge window (around 40% I think) – otherwise they would never guarantee a battery life of 8 years!…
I suppose I will have to do the same, but I may sub-dimension the batteries a little in order to spread the cost along time.
LiTOs are not new… 11 year patents… John, by that measure, there is very little stuff that is new. The very same thing can be said about EESUs…
The relevant question is: is it on the market? and LiTOs really _are_ coming to market in the next few months.
October 12, 2008 at 1:45 am
There are batteries that have better than a 10X cycle life when used lightly (not more than 70% of total charge), and there are others that only have 3x their life with the same discharge levels … depends greatly on what the specific battery modelwas designed to do. High capacity batteries are optimized for storage density, and high current batteries for cycling. To maintain the 70% level of use, requires 50% more batteries for the extra safe-life capacity, plus extra space and wieght to carry the extra 50% more batteries, which has additional costs besides just the battery cost, as it degrades performance too. YMMV
Part of having this discussion is to help people consider that there isn’t a one-size-shoe-fits-all solution. Also to raise the suggestion that picking up a logger for your car, and packing it around for a few months, is probably necessary to understand what your usage and design points will be. My rural letter carrier makes about 500 low speed stops per day, an application where high efficiency with solid ReGen would make a real difference. Fortunately, I don’t have to design for her level ofdaily use.
The figures I’ve used, are different than Vasco’s, as are the cost expectations over a 10 year life.
Using the design 120km/h figure above, and the 2KF Maxwell 1500 boost caps available on ebay http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=370076930494 would require 130 caps to store this panic stop energy. At $52USD each off ebay, that is about $6,760USD. These should not degrade significantly over 10 years, so amortized over 10 years would be about $676USD/yr cost. About 25% of the $28,500USD cost Vasco quoted above. For high start/stop low speed city driving routes, the fuel/battery savings over 10 years can be much more.
The reality is that between 15-45% of the panic stop energy can be pushed back into your battery pools, so the cap costs should be 15-25% less. So this is one advantage to having 50% excess battery capacity to avoid cycling costs, as it also increases peak charge current for worst case panic stops. I will be using a combination of LifeBatt’s for high current and probably high density Yesa cells for the low current/high density battery pools.
This is the best I can do today. Next year, there will be different choices and costs, but I don’t plan to wait a year or two, with oil prices likely to rebound once the economy picks back up.
Fuel costs, for the TDI Beetle are $3,000USD per year, so I feel I have a 4-5 year fuel budget (about $12,000USD to $15,000USD) budget at todays prices to break even on the conversion project, or do better and save, over the next 10 years. If there are better solutions when I do the second car next year, I will make optimimal solutions for that project come purchase time.
October 12, 2008 at 2:17 am
Ops, those caps are 1.5KF, not 2KF … need to increase cost by 33%.
October 12, 2008 at 9:17 pm
John, that’s exactly what I’m doing now: carrying a GPS logger (my phone) in my car recording my day-to-day driving. I hope that one position sample per second will be enough for me to draw some nice velocity, acceleration, and power curves. If that fails, I can increase the resolution to 4 samples per second. GPS is a little indirect and has relatively low precision, but so is everything else. At least it is cheap and easy and will get me in the ball park.
With this data in hand, I intend to write my next post (about power), and define some real requirements for the energy storage system – which cannot be dimensioned from the point of view of energy alone, as we all have seen.
Nice to know that the Maxwell caps are up to the task… I had ruled them out on the basis of cost, but I see I have to consider them again. Also, my fuel costs are way higher than yours… Gasoline instead of Diesel, Europe instead of USA…
so I’ll have a larger margin for storage tech investment.
It had occurred to me what you say about the batteries being able to take part of the regen current – but I’m waiting for my power usage data to come up so I can craft a realistic power distribution strategy between Caps, Lithium, and Lead-Acid.
October 13, 2008 at 6:22 am
That’s almost what I did two years ago when I was doing the basic physics research and just starting to refine the Axial design. There was too much jitter in the gps data to get good numbers for start/stop ramps, but lots of good data about the nature of trips and elevation changes.
This summer I ordered two Sure Electronics DC-SS009 3-Axis low-g Accelerometer prototype PCB’s to build better instrumentation for my project. Currently doing an AVR ATMEGA8 logger with USB to get better data. See http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=250304782994
They are easy to wire up to any of the Free Arduino ATMEGA168 project boards with a FTDI USB chip so you can get the data back to a notebook easily. I will probably end up building a half dozen of them .. one for each family car, my bike, and the project car.
October 13, 2008 at 3:52 pm
Nice.
My very hackable phone also has a couple of accelerometers on it. They have 2g / 8g selectable sensitivity, but being inside a mobile phone may make them very jittery, I don’t know yet.
I can see the usefulness of making a separate logger for each vehicle…
I haven’t started messing with the accelerometers yet, and frankly I haven’t thought much about how to do it… when I do, I may decide to get some ready-made boards from you!
How do you analyze that data?
December 12, 2008 at 5:38 pm
I made the same calculations to make them clearer to myself…and some questions bounce around in my head now: 1)The motors have enough torque for such a panic stop? The total torque must be around 1300Nm considering a constant torque braking action. If the constant power rpm range is to be considered the calculation is a bit more complex and depending on the motor specific torque/rpm curve. 2)Are the tires capable of transmitting all the torque needed to the ground? Usually yes but the motors must be controlled independently to achieve an ABS-like behaviour. 3)Is it really necessary to store back all the energy from a PANIC stop? How many of these stops you expect in one year of driving? I hope no more than a few or your life is really at risk in everyday driving…in the end it would be a cheaper and still very efficient car if, in case of an emergency stop, you use power resistors to take the extra current (the current exceeding the battery maximum charge rate).
December 13, 2008 at 10:38 pm
Torque is proportional to current. So, in the end, it all comes down to how much current you can have in the motor, and for how long. And this is equivalent to asking how much current your circuits can take, and also how hot will the system get.
Notice that I always say “model X has Y Nm at Z Amps per phase”… so any motor can take a higher braking torque than its rated motoring torque, provided you can keep it cool and your electrical system (batteries included) can take the power. Fortunately, braking is a transient action.
You’re right, it makes more sense to dimension the system for “average everyday driving” and putting in resistors for panic stops. I was just entertaining a wild thought.
I have started collecting GPS logs with my car in everyday driving, and I have a script that turns these logs into power graphs. I intend to publish this some day now, when it is ready. Anyway, when this method is perfected, I will be able to identify exactly which power levels I have to design for, and have the answer to “what is a normal stop, and a panic stop?”.
As to ABS-like, it is known that an electric motor is capable of superior reaction speed than ABS mechanical brakes. This because there is inertia (=delay) in the hydraulic circuit that the electromechanical systems don’t have. So, in the end, electric wheel motors will yield a better ABS (and other functionalities) than the present systems… assuming you can design and program a high quality controller!!