OK, I understand the basics of how most CVTs work: you have a pair of pulleys with continuously variable diameters, connected via a belt. The pulleys vary their diameter by moving two halves (with oppositely slanted sides) closer together or farther apart. What I haven’t been able to figure out is how the belt works.
With a normal set of automotive pulleys, you have a V-belt or a flat serpentine belt that grips the edge of the pulley. While a V-belt might get some grip from the sides, the majority is from the inside edge of the belt.
With a CVT’s pulleys, there is no pulley edge, so the belt has to get its grip from the sides. How can a belt get enough grip to handle the torgue of an automobile engine without binding?
I also notice that some CVTs (such as Nissan’s Xtronic) use a metal link belt. How can a metal belt get a grip on a metal pulley?
Lastly, it seems to me that CVTs should give great improvements in gas mileage - the engine controller can be programmed to run the engine at its peak efficiency point (or peak output when acceleration is needed) and just change the transmission ratios; however, specifications that I’ve read indicate a relatively modest improvement in MPG compared with a manual transmission on the same car. From this I infer that an engine’s efficiency doesn’t vary too much over a fairly wide range of RPMs. Is that correct? If not, is there some other explanation?
>While a V-belt might get some grip from the sides, the majority is from the inside edge of the belt.
A V-belt gets ALL its grip from the sides. Note that typical V-belts have a flat inner edge that is much wider than the radiused inner edge on the pulley.
My vehicle’s mileage in gear 4 at 55 MPH is 15% higher than its mileage in gear 3 at 55 MPH. We’re talking the difference between 2000 RPM and 3000 RPM, give or take.
As Napier notes, a V-belt makes all contact on the angled sides and none on the bottom. (If you think about this a minute, it’s obvious that it has to, otherwise if the belt were able to bottom out it wouldn’t be able to reliably contact both sides.) Here is a graphic of a V-belt type CVT. There are flexible metal mesh belts that make contact via cogs or friction, but the Nissan Extroid is actually a torodial CVT that replaces the belt with rollers; see here.
Gasoline IC engines are pretty effecient over a wide range of engine speeds, and with modern 4- and 5-speed automatics and 5- and 6-speed standard shift transmissions there isn’t a lot of efficiency to be had between the ratios when traveling at constant speed. Most ineffeciency comes from excessive fuel delivery during acceleration, where a lead foot on the throttle results in energy wasted fighting inertia. By analogy, think of trying to move a heavy cart by repeatedly ramming into it with your shoulder versus a slow, steady pushing motion that gets the cart moving. The ostensible benefit of the CVT is that it is able to adjust for these accelerations to quickly give you an optimum gear ratio for both your vehicle speed and throttle level; the reality, though, is that they take some time to respond and are at best only marginally more efficient that a modern planetary gear “automatic” transmission, and that a skilled driver in a standard shift with a light touch on the throttle can still extract as good or better efficiency than a CVT.
CVTs are really best when paired with a small engine with limited throttle response; when driven by a more powerful engine with a quick response current CVTs just don’t respond quickly enough, and it’s only been in the last five years or so that they’re robust enough to take large power throughput reliably. As the technology improves CVTs will probably become more common because they are (in concept, at least) mechanically less complex than planetary gear transmissions or even standard spiderboxes. However, it is still a somewhat nascent technology, whereas other types of transmissions have been evolving since the early days of the automobile.
Don’t forget that modern cars have already had their efficiency tweaked pretty high by previous improvements.
(gas mileage has doubled since the 50’s)
So most improvements are only going to give moderate benifits rather than earth-shattering ones.
For “holy crap” level improvements you need to do things like switch the frame to carbon fiber.
(still too expensive for mass-production but they’re working on it)
And probably never will. Laminar fiber composites are great for high performance cars where every pound counts, but for general purpose passenger cars intended to be in use for over 200,000km of service it just isn’t a great type of chassis structural material. Stamped/hydroformed welded monocoque steel and aluminum are significantly better materials for this on the basis of robustness, ease of manufacture, NVH, et cetera. You may see an increasing use of fiber composites in body work and cabin structure, but not in the chassis for general consumption.
My understanding is that the Tesla Roadster’s body is 100% carbon-fiber. As for the chassis, I dunno. True, this is hardly a car for “general consumption”. Tesla also plans a general purpose car, but I don’t believe they’ve released any information on what the car will be made of. And, in any case, I doubt they’ll get around to building it before they go under.
Only if you play with the numbers. True, a modern Prius does get vastly better mileage than a '57 Chevy, but hybrids like the Prius are not the dominant cars on the road. Actual fuel economy has been pretty flat or even declined over the years (an 80s model diesel will get the same kind of mileage as a Prius, and, IIRC, there’s a few gas powered cars from that era which match or come close to the Prius), what has increased, fleetwide, is the amount of horsepower on tap.
CVTs are actually one of the oldest automotive transmission designs out there. The problem they ran into way back when was that the leather belts weren’t up to the job. Subaru brought them back in the late 1980s (in the Justy), but they were not well recieved.
In some ways, we are at the limit of automotive technology in terms of fuel economy, at least in terms of what automakers are able to do. Switching over to a new design, even if it’s just built using off-the-shelf technology, can cost around $1 billion. To ditch a technology in favor of something that will have to be abandoned in a couple of years just isn’t worth the expense.
My understanding from a Nissan CVT perspective is that they are using them move because the are smaller, lighter and cheaper to supply than conventional manual and automatic transmissions. Not because they are more fuel effienct. I have one in my Altima coupe and have gotten use to not shifting anymore. Also, the Maxima I traded in had a problem with torque steer when changing gears. I don’t have that problem anymore, even though I have the same sized 3.5L engine. It’s pretty zippy in fact.
Point of order: the Prius, while it does get pretty good mileage, is by no means setting the bar for modern cars. For comparison, look at the mileage you can get out of an Insight - mine regularly gets between 55 and 60 mpg, even in the city. Admittedly, hybrids are still not the norm, but what I think the Insight does particularly well is show how modern materials and streamlined design can lower the fuel consumption on an otherwise fairly normal car. (Most of the mileage improvement comes from better body design and a smaller engine, and only a small percentage comes from the actual hybrid technology.)
I can’t speak to every CVT iteration out there, obviously, but I did a fair amount of research on the effects and differences you get out of one in Insights particularly. The mileage change is really quite minimal - as Stranger pointed out, you can still get better mileage with a light foot in a standard and gentle shifting techniques. There simply isn’t a fast enough response from throttle demand to ratio change for it to be as efficient as one would hope. There’s also the fact that, under normal driving conditions, one would only be varying between 2000 and 3000 rpm, as Mr. Slant said upthread. It’s hard to get a lot of efficiency variance in such a small range, when standard technologies are already so carefully calibrated.
I would disagree with the statement that you have to play with the numbers to get better fuel mileage numbers for modern cars.
Back when I started driving in the mid 1960s the average car was built in Detroit. It got about 10-15 mpg in town and maybe 15-21 on the road. My father’s 1963 Chrysler Newport had a 361 ci V-8 about 21-22 MPG on the open road. It had seats for 6, a fair sized trunk, and no air conditioning.
Compare that to my 2005 Volvo. Seats 5 with a larger luggage space (station wagon), and electronic climate control. I get 25-26 mpg in town, and I got 36.5 MPG on a trip back from my sister’s house on the 26th. I will note that my Volvo is not a particularly stellar car for gas mileage. It is a heavy battle cruiser, and there are plenty of cars out there that are similar in size and seating that get noticeably better fuel mileage.
Or if your prefer let’s talk about trucks. My 1961 Chevy 1/2 ton PU got 10 MPG. In town or on the road, it didn’t make a difference. My daughter’s 2006 F-150 gets 18 MPG on the road.
Let’s talk about cars from the past getting similar fuel mileages to a Prius. Sure there were a few diesel cars from the 1980s that approached the numbers a Prius achieves easily, but they were few and far between. A diesel Rabbit is about all that leaps to mind as a car from the 1980s that achieved Prius like fuel mileage numbers.
I cannot think of any gasoline powered cars from the 1980s or earlier that seat 5 and get 50 mpg in stop and go traffic. Maybe there are some, but I will be damned if I can think of them.
The Prius does not use the type of CVT described by the OP. It uses a type of planetary transmission in which all three elements move (a conventional planetary transmission has two moving elements and one fixed element). A computerized control system determines how the three elements should move relative to each other to optimize performance. This system allows energy from both the ICE and electric motor to be combined in different ratios and sent to the drive wheels (or, alternatively, for energy to be taken from the wheels or the ICE and sent through the motor/generator to recharge the battery pack).
There’s a good description of the Prius transmission here. As you can see, it’s a little misleading to call it a CVT. It’s really a power-routing system, not a transmission.
Will you settle for the fuel economy ratings for a 1980s gas powered Honda? I’m a bit too busy to go wading through the zillions of pages which come up when I do a google search on vw+diesel+mpg+golf (which IIRC, that’s the VW diesel that was able to do it).
Note that the Honda’s doing it without the sophisticated electronics and computer controls of the Prius. Of course, the Prius is slightly larger, and has more safety features than the Honda, so the comparisons are somewhat unfair. However, the Honda is a mass production vehicle, and not some kind of weird concept car, or one-off, unlike say the Peel P50, which was a niche vehicle, and didn’t have the capabilities of a normal car.
Those are a push-type belt, in compression rather than tension. The stack of links is effectively a single solid link from the driving pulley to the driven one. There are metal bands fitting in slots that hold the links aligned in the slack (return) side. That approach can transmit a much higher torque (and therefore horsepower as well) than a rubber tension belt, and without ratio variations due to stretching and resultant thinning of the belt. Friction coefficients are less than with rubber, sure, but still around 0.3 or so.
Depends on the “peakiness” of the engine’s power curve. If engine RPM can vary a bit without loss of efficiency, a gearbox can be made with enough ratios to keep RPM in the happy range. Power losses in a CVT are *higher * than with a direct-drive gearbox due to slippage (there must be a friction interface somewhere), and to unrecovered elastic strain in the pulleys and belt (whether rubber or steel).
1985 Volkswagen Golf/GTI 4 cyl, 1.6 L, Manual 5-spd, (DIESEL), Diesel
Estimate MPG under CURRENT system:
31/41 City/highway
2008 Toyota Prius
Estimate MPG under CURRENT system:
48/46 City/highway
Remember, you can’t compare '80s EPA ratings with current EPA ratings. They changed the scale…
So they’re close, with the Prius coming out ahead, but when you factor in the price savings (since an 80s era car will cost less than a modern one), I’d say that the VW’s a clear winner. (Note, I don’t like VWs and more than I like Priuses, so I’m not biased.) Also, a 2003 TDI VW Beetle, with minor modifications is reported to get 76 MPG. It, like the Prius, uses low rolling-resistance tires, so they’re somewhat evenly matched. He’s also running on biodiesel, which a Prius can’t (as of yet) do. Put a diesel in a hybrid, and you’ll really boost the MPG.
2 years ago, IIRC, so going back to the '04 models would be fair, I’d think. Toyota hopes to have a 100 MPG Prius on sale next year, and AFAIK, there’s no way a diesel could match that in a conventional car, without serious modifications.