jniemeier Posted: Sep 25 2005, 06:51 PM

Someone asked about schools. I have no special advice about that. Find a school that has a solid Mechanical Engineering program. Most state universities do. Specialization comes in your masters program, should you go that route, and most people advise going to a different school for your masters anyway. Getting a degree is to teach you to think, and what get's you in a company's door. Actual useful knowledge starts when you get to work. That said, it's a good idea to pay attention when they teach IC Engines by Obert. (See above) I took it from a German, ex-VW engine guy whom we were legitimately afraid of.

Electric Valve Trains: Now there's a subject we could all debate for weeks. Basic benefit: Changing the duration of the lift event on the fly. Think of the motion of today's cam and spring driven valves. Cam design programs calculate four derivatives: (you paid attention in math class, right?) displacement, velocity, acceleration, and jerk. Even the jerk has to be smooth, if that makes sense. There's a whole lot more to it than lift and duration. Ok, fine. Chances of replicating this motion with an electrical solenoid? Zero. But do we need this gradual start - fast lift - gradual stop type motion? Well, if you want to keep from fracturing valve stems, you sure need to gradually close the valve. The solenoid guys call this "soft landing", and I understand there are patents for that via reversing the current at just the right moment. How do you know when the right moment is? You now need a valve position sensor. And it better be a darn fast one considering how fast the valve is moving much less the extremely nasty environment (see earlier posts). If you can do the soft landing, how much juice will it take to run the solenoids? Last I read, a couple research firms said they could do it with 1.5 kilowatts for a 16v four cylinder. Not terrible, but still too much, and I don't think it's been demonstrated either, although EMVT cars have certainly been driven around. (I'm sorry I don't know right off how much power it takes to drive a 4 cylinder's cam- but keep in mind that when the follower is on the back side of the profile, the spring is returning energy to the system not consuming it.) After we figure out lift profile, soft landing, position feedback, and power consumption, we still must face reliability of elec connections, packaging it to fit, noise (very important), lubrication, cost of the solenoids, cost of the control electronics, packaging of the control electronics, manufacturability and of course durability. It's a huge challenge, but what's new? There are a lot of bright guys in research companies who thought they could do it. Notice I've switched to past tense. I think full electric valve train is a pretty cold trail. One caveat though: HCCI combustion (Homogeneous Charge Compression Ignition) could be a huge efficiency improvement, but may require VT control like electric valves. If you want to know about HCCI, you'll have to look it up. I can't explain it very well. Think of controlling a gas engine like a diesel. Extremely tough, although I think this one will ultimately succede.

The fully variable valvetrain that's in production of course is BMW's Valvetronic. It changes lift and duration at the same time. We designed a system that did this also, but we could never get it cheap enough to sell it. I guess BMW has a different view of the value equation for new technologies. Interesting question: how much valve lift does it take to idle a 4 cyl engine? 0.3mm or even a bit less. That's amazing. And you can do it at 500 rpm or so, which saves a bunch of fuel. Turns out cars spend a lot of time idling.

There are a bunch of systems that change lift in two and even three steps. Most famous is Honda's VTEC. They have at least 6 variations of it in production, including on the new 3 liter V6 that does high and low lift and cylinder deac in the same engine. That is fantastic, and no doubt comes close to the benefit of Valvetronic for a lot less money. The Mitsubishi MIVEC system can do three steps and has been around for a while. Porsche have a two step system on the Turbo which is made by INA of Germany. Then of course there's cam phasing. That's old news, and on most cars these days.

The system that I've just read that IS going to enter production is Fiat's Electro Hydraulic VT. They sometimes call it UniAir. Electric solenoids open a high pressure oil circuit which has an actuator to open the valve. This sounds like a good idea, but the hydraulic fluid was always a challenge. The viscosity change due to temperature was a killer. Evidently FIAT have solved that problem. They will use it on a diesel (lower redline) starting next year or so. UniAir can do variable duration and lift by basically truncating the lift profile when it's had enough by dumping the oil through a relief valve. Should be great if it works. I actually met Dr Petronio, the father of UniAir once. (I was selling, but he wasn't buying!) He had been working on it for years at that time, and you could just tell that by-God he was not going to give up. Sometimes, even in a corporation the size of FIAT-GM Powertrain (they've since split up) it boils down to one very tenacious guy. In fact, I think it almost always does. That's interesting, eh?

Oh, and don't bother trying to dream up other "gates" or types of valves besides poppet valves, like the spherical rotating ones you read about. Talk about patents- in valve types, everything has been tried before. Nothing will seal combustion pressure like an inward opening poppet valve. No matter what system creates the opening and closing motion, today's poppet valves will always be used in piston engines. And don't even get me started on ported two strokes.

Pop quiz: How does a Ferrari valve get a larger lift profile than the what's ground into their cam?

Now, if I write any more you guys are going to have to start passing the hat. I'm going to bed.
Jim N.
'73 2.0 with a nicely lashed, single return spring, pushrod and rocker arm mechanical valve train that I can fix and adjust in my garage.

23SE05
We call them ETC's: Electronic Throttle Control, and no, that won't work. For one, you can't put an ETC into a 500deg C exhaust stream. Well, you could, but it wouldn't live long. Why do you think this would work better? Doing it in the valvetrain is the closest to the combustion chamber we're trying to control and by far the most mechanically simple (engineers would say, "elegant") solution. There's no need to think up anything else.
Individual throttles implies you're using individual intake runners. Very expensive and won't fit for a V type engine. V engines like plenum's and a single throttle cuz it helps even out the flow balance to the cylinders, fits better, and of course is lots cheaper. Racing engines may have seperate runners, but keep in mind they are optimized around a very narrow rpm band.

Cylinder filling: I don't think you caught what I wrote before. You need to seperate the idea of quality of combustion from the amount of mixture in there. To have acceptable emissions, you need to keep the mixture at 14.7 to 1 air/fuel ratio. Period. After that, the AMOUNT of 14.7:1 mixture we get in there is the thing that directly controls the torque that cylinder makes. At Wide Open Throttle (WOT), less restriction to air flowing in, more mixture in the cylinder, max torque. At idle, tons of restriction past the throttle, little mixture inside, low torque. In both cases, the mixture is always 14.7 to 1. It takes a while to digest that fact. So, things like the ease of initiating the burn, and the combustion 'efficiency' are the same in both cases. The power generated is drastically different of course, so the heat loss to the wall is different, sure, but the efficiency is the same. Engines are most efficient at WOT not because they burn better, but because they don't waste work pulling air past the nearly closed throttle at low loads.
Answer to paragraph #3: Number 1 and 2 are both good, although I think number 1 is the easiest and likely the larger improvement, at least for big V8's. With a 914 engine pushing the car at 70 mph, you can't be lowering the revs too much can you? But, when Chevy was trying to avoid the gas guzzler tax on the Vette inspite of increasing hp to 400 (and now 500!), they went to gearing that left it turning what, 1600 at 70? Something like that. IF you have the torque, that's the easiest way to go. By now, I hope you agree that Number 3, increased filling, is simply a method to increase torque, not improve efficiency, BUT, you can use your increased torque to drive taller gearing, and THAT is option Number 1, the best option. So, 3 is just an enabler for 1. Got it?

Now, since you're such an inquisitive guy, here's my advice I hope you take on board: Go to a used book store (or probably Amazon) and buy a copy of "Internal Combustion Engines and Air Pollution" by Obert. That's the one used in most engineering schools and is the all time classic. Every engine designer in the country has a copy in his office. Read the whole thing. It's actually not that hard to read. I'm sure you will get a ton out of it.
Jim N.

21SE05
First, Eric- I agree completely. We could cut the oil consumption of cars and light trucks in half with today's technology. i.e. Double fuel economy. No sweat. But we can't do it while driving Tahoe's, Durango, Expeditions, and Land Bruisers... For me, step one is diesel. Instant 30% improvement with no sacrifice or downside for Joe American. Europe is now 50%+, US is <1%. I can tell you diesels are finally coming to the States. The fuel will be 30ppm sulfer I think in '06 or was it '07. That allows the diesel catalysts to live. Get ready for a gradual 20 year roll-in to reach 40 or 50% penetration. I don't know anything special about fuel cells, but I'm optimistic they will eventually make them economical. They've only scratched the surface compared to the millions and millions of man-hours invested in getting IC engines to work well.

Andrew- good guestion. You're making think, which has it's pro's and con's most days. Wish I could draw you a P-V Diagram. (Pressure vs Volume as a four stroke completes one cycle) Pulling the connectors on four injectors would certainly make things worse. The throttle blade would be open a bit more, as on a deac engine, but the non-firing cylinders would be wasting more energy pulling air in against the manifold vacuum and then pushing it out against the exhaust back pressure. It's a loser in both directions. These losses would be worse than the slight gain in the firing cylinders.

If you've heard it said that an engine is just a big air pump- well, that may be true but it's sure an inefficient one cuz the first thing you come to is a nearly closed off throttle valve. At highway cruise it's only open 10%. 90% blocked off, which is proven by the huge vacuum in the manifold. My 914 pulled 14.5psi vacuum at idle when I tested it this fall. That's a measure of how inefficient a motor is, not the opposite. One of the reasons diesels are more efficient is because they don't need a throttle valve, so there's no vacuum in the manifold. i.e., it's very easy for the pistions to suck in the next gulp of air. So, controlling a gas engine by throttling the air (both at the throttle valve and at the intake valves- don't forget) is inherently worse than controlling a diesel by the amount of fuel injected.

Charge density helps but it's not more efficient, it just has more air and fuel cuz it's denser. That's why intercoolers help. Cool air is denser (more) air, to which it's easy to add a smidge of extra fuel, and so get more power. Remember, you need to stay very close to the optimal 14.7:1 ratio. Burn rate and burn completion is mostly due to fuel atomization (droplet size), and local air/fuel ratio (how homogenious is it in there really?) and the degree of turbulence created by the intake velocity and direction and the shape of the upwardly rushing piston against the chamber shape.

Time to put the cookie-munchers to bed.
Jim N.

20SE05
This followup will be short, I promise.
Reduction in "pumping losses" is why Deac improves economy. By pumping losses I mean the work required to pull down on intake against the restriction of the throttle blade. When in 4 cyl mode, a deac engine has it's throttle blade further open than it would in 8 cyl mode at the same speed-load point. Think: working harder on the remaining four cylinders. Further open means air flows in easier, i.e. less losses.
If you mean the frictional losses of the piston going up and down with both valves closed all the time, yes of course that is an unavoidable loss of efficiency, but it's not that bad, cuz the work required to compress the air is recovered when it expands. You only lose the friction, which isn't terrible with a warm engine and the low tension rings they use now-a-days.
Saab's Variable Compression engine is no doubt still around, and still in development. (Did you hear about the one that sparked from the plug to the top of the piston? Those Swedes!) These things take years. I've also seen one from FEV, a Euro engine design company in their Detroit office. It's a steep uphill challenge: added complexity means lower reliability, and more variables means tougher to meet emissions reg's, which is the hardest aspect of designing an engine. And when a new engine line costs many hundreds of millions, they tend to be a conservative bunch. Believe me. Very conservative. On the other hand, engines are now extremely reliable when you think about it.
OK, not short, but let's call this one medium. See you around. Good questions.
Jim N
'73 2.0

19SE05
I'm new to the group, but you're on a topic I know something about.
Warning: long boring post to follow: Read only if you want to know more than absoutely necessary about cylinder deactivation.

(I was the Engineering Supervisor for the Delphi Valve Train group when we developed the cylinder deactivation system used by GM.)
The old Cadillac system used an electric rotary actuator on top of the rocker arms. We found a '82 Seville with a V8-6-4 when we started working on the new system in '99, and it was still working fine. Must have been the only one left. Putting electrical connections under the valve covers is generally a bad idea. Someone mentioned the Mercedes. Their V8 and V12 were the first of the 'modern' systems (I think in '99), and was very complicated. Big surprise. It's very tough to switch off the valves in an overhead cam engine. It has a very expensive, high pressure valve asm. You've only got the rockers to do it in, and it's hard to get the right oil circuits to your hydraulic actuators (inside the rockers!). Honda have also done it this way. You need a pivot shaft running down the length of the head so the rockers stay in perfect alignment and for the oil to run in. ($)
The new systems from GM (Displacement on Demand) and Chrysler (Multiple Displacement System) are almost identical. They (we) are doing the switching in the Roller Hydraulic Valve Lifters of pushrod engines. Much cheaper. There's enough space there to add locking pins and when hydraulically unlocked to absorb the cam lift without moving the pushrod. Ford would be doing the same, but they seem to have forgotten how to make pushrod engines, so they are out of the party for now. Chrysler's are made by INA (Germany); GM's by Delphi and Eaton. (Yes, legal wrangling has ensued, but let's not talk about that)
Operationally, it switches off the exhaust first, then the intake of every other cylinder as you go throught the firing order. The lifter has to be on the cam's "base circle" to allow the locking pins to move. That means inside cylinders on one bank and outsides on the other. Hence, the motor stays even firing. The engine mounts are bi-state (!), and are electrically switched to a different natural frequency in sync with the motor switching. It is all done within two engine cycles, and no, you can't feel it. GM's calibrators can't even feel it. They wire up an LED on the dash for reference. There is an electrically actuated hydraulic control valve for each cylinder that fire in sequence driven by the ECM and the cam position sensor. Think high speed: elec signal, solenoid movement, oil pressure buildup, locking pin movement... all in 10 milliseconds, and repeatable over the full range of oil temps. Hint: changing your oil regularly is a good idea.
As to economy, as always, it depends. I'm talking full size trucks and TrailBlazers here. The overall real world average they figure to be 8%. Might not sound too huge, but to a car company, that's a big number. In suburban driving, ~45ish and lot's of light loads, it could be over 20%. On the other hand, I heard the calibration guys say once that about 75mph is where the "road load" power required prevents it from going into deac. So, I doubt the heavier versions of the trucks are deac'g much on the highway. I guess running a 6 or 7 thousand pound truck on 4 cylinders isn't too easy. Our job was to just get them turned off and on, not make it more powerful in 4cyl mode. As mentioned, not it's also in the Impala SS. I didn't think they were doing the Corvette, but maybe so. It's also possible to do a V6, although the useable range of loads is a bit narrower. To even-fire a V6 in 3 cylinder mode, you need to deac one whole bank. And, yes, it's been done on a 4 cylinder by a research firm, but... well, let's see how the V8's are accepted.
Concerns? Oil puddling on top of the valve guides, and then getting sucked in when the cylinder suddenly "re-acts". GM's trucks periodically switch back to 8 cyl even if not required to prevent this. The fuel injectors are off of course, but the spark is still on. No need to add the complexity of switching it on and off.
Sorry for the long post, but engineers never know when to shut up.
Jim N.
'73 2.0