Intake Sealing

[Steve P.]

I'm attempting to to a variable intake for our 2012 car here at Virginia Tech. I have the design mostly figured out, but am having trouble with the internal dynamic sealing. If anyone has any information that they think would help, I would greatly appreciate it.

[RobbyObby]

I believe RIT ran variable length runners on their intake last year, at least at West. I may be completely wrong though. Anyone care to confirm? You might try contacting someone on their team from last year.

[jd74914]

IIRC, RIT's are not sealed per se as the runners lengthen inside the plenum.

Steven, why not seal with an o-ring? With the correct tolerancing they can seal well in reciprocating applications.

[ST]

All I can say is good luck, the idea of variable intake manifolds is awesome but it's the sealing that kills them.

UWA did two, in 04 and 05. One had awful sealing and good flow paths, the other had horrible flow paths and good sealing.

The net result was an awesomely flat torque, unfortunately it was a collection of all the local minimums.

The guy who did the motor in 06 ripped off all that junk and chucked on an ally welded box for a dyno run just to check things out. The dyno printout was just hilarious when laid over the top of of variable intake curve.

[exFSAE]

The net result was an awesomely flat torque, unfortunately it was a collection of all the local minimums.

I'll admit, that got me to laugh.

[Gruntguru]

OE systems generally comprise 2 discrete runner lengths with a set of butterflies to switch between them.

If you want continuously variable length, you could try Teflon piston ring seals or lip seals of the type used on pneumatic rams. I think O-rings might be a bit draggy. The linear bearing system might be a bigger challenge than the seals.

[Steve P.]

Originally posted by Gruntguru:
If you want continuously variable length, you could try Teflon piston ring seals or lip seals of the type used on pneumatic rams. I think O-rings might be a bit draggy. The linear bearing system might be a bigger challenge than the seals.

Thanks, that was exactly what I was looking for. I also thought that O-rings would probably have too much resistance. I'm going to look more into both options you mentioned.

[Wesley]

Ever made a slide whistle?

Resonance won't be killed by imperfect sealing.

I considered making one for a while back in '08, and my "cocktail napkin" design basically included short runners with a nested bellmouth inside it (think like a slide trombone, but slide the bellmouth instead of the main tubes.

The bellmouth portion would be machined from lubricated nylon and the runners that connected to the head would be Al - surface prep is important for smooth operation.

So essentially, on the exterior of the narrow section of your nylon bellmouth you have a double sealing ridge (the size of these ridges will depend on how well you can achieve sealing while still maintaining smooth operation, going to take some experimentation) and then all 4 horns are joined by a common rail that has a solenoid controlling it.

If your fitment is good enough, you'll be able to maintain your harmonic effects, then when it comes time to tune, just run the dyno and a steady RPM, bump your runner lengths throughout the range, tune, and get enough torque data at set RPM and runner length points to map out the trend and decide how you want to program your controller.

You'll still have to deal with transients, but that will at least give you a place to start.

The problem with this setup is that, to achieve a decent range of runner lengths, you're going to have to have a pretty obscenely sized plenum, which is why I never implemented that particular idea, but perhaps better minds than mine can work that out.

Another idea I've seen (on a bike of some kind, I don't remember) is a set of nested bellmouths - one sits inside the other, and once you reach your target RPM, the first bellmouth lifts out of the second, effectively shortening it. That would be tough to implement in a plenum-throttled design though.

[ST]

Also,

Toyota developed a telescopic runner for their F1 engine that ran up and down their sort of "Air Bucket" (they had individual throttle bodies but the plenum was still kinda there, just with one face taken off). It's in "race engine tech" October 2010 as well as plenty of fantastic things about that engine.

Would be easy to create and sealing wouldn't be a impossibly hard task, getting a linear drive that can follow the engine through its rpm would be a pretty tough act. Keep in mind when you change gears your going to have a near instant drop in rpm and your runners (if set to resonate at the pre-gear change rpm) are going to be not the right length.

Still love to see it done well, don't think there is one team out there that was impressed with it enough to keep in on their car throughout the years.

[Drew Price]

Originally posted by Wesley:
Ever made a slide whistle?

Resonance won't be killed by imperfect sealing.




This is a good point. Most OE street cars with variable intakes (like current BMW and older Cadillac engines) use some kind of flapper valve that blocks or exposes portions of the plenum volume for variable manifold volume.

Like the variable volume intake that ETS had on their 2008 car.


For telescoping runners I'd echo Gruntguru, I had the same idea, lip seals or even shaft seals, or o-rings with a light fit would probably work fine, you'll have to take friction (and the runner block's tendancy to get 'cockeyed' and uneven through it's travel) into account when specc'ing your actuator and linkage.

Based on gear ratios in many FSAE cars, you will need pretty fast travel speeds to justify it for transients at least.

[Gruntguru]

Originally posted by Wesley:
Resonance won't be killed by imperfect sealing.
But your chances will be killed, if the motor won't be killed, when the Tech inspector puts his hand over the intake.

[Drew Price]

...Unless your telescoping happens inside the plenum....

[Charlie]

Sealing a dynamic runner length is a difficult task.

Getting dynamic runner response time to match a FSAE car's RPM acceleration range is near impossible. I have never seen anyone do it. Make sure that is your main concern before implementing. Too many people have looked at static dyno tests and assumed their controller was going to be able to match them.

[RollingCamel]

Originally posted by Charlie:
Sealing a dynamic runner length is a difficult task.

Getting dynamic runner response time to match a FSAE car's RPM acceleration range is near impossible. I have never seen anyone do it. Make sure that is your main concern before implementing. Too many people have looked at static dyno tests and assumed their controller was going to be able to match them.

What if throttle response is taken into consideration? A predictive system still can't be fast enough?

[Charlie]

Originally posted by RollingCamel:
What if throttle response is taken into consideration? A predictive system still can't be fast enough?

I'm sure it CAN be, it's just exceedingly difficult.

[Adam Vaughan]

What response time would you consider acceptable?

Without optimizing the servo motor, I had less than 1 second full range with a lightweight, cable based runner assembly (see SAE Paper 2011-01-0420 for the design + experimental data). The servo system wasn't the focus of my project, or I definitely would have improved this metric. I also ran some CarSim simulations which suggested that gear shifts could be avoided through the variable intake manifold's flatter torque curve. Avoiding gear shifts saves dead time during shifts and simultaneously avoids full range transitions of the runners. Alas, I never fully trusted that CarSim model / I needed greater gains from variable action to truly support the claim.

As for sealing, Wesley is right, it doesn't matter nearly as much as you might think. If you need to prove it to yourself (like I did), take an RP mockup of your design to an acoustics lab and listen for the difference with your ear / mic + FFT analyzer. You can also listen to the engine change note on the dyno to let you know it's working too ;-).

Hopefully the current FSAE team at Cooper Union get's that engine + manifold + cam setup fully optimized for the 2012 competition, because it would be great to see the final set of torque curves along with some actual track data.

[Adambomb]

IMO anything close to 1 sec. would be immediately noticeable and undesirable. Truly good throttle response can be heard/felt when you can physically hear the difference between one "soft fire" at low throttle followed immediately by one "hard fire" at larger throttle. This is easily distinguishable on singles and v-twins with a lot of "dead space" between fires. I don't think my ears have a high enough sample rate to hear it as distinctly on an even-fire 4 at high rpm, but on a well-tuned 4 it does happen as fast as I can hear it!

So let's set a time limit of my preferred throttle response as the "dead space" between fires on a 72 deg twin at 10,000 rpm. Seems like a good enough ballpark estimate. That would be ((360-72)+360)=648 deg., and (648 deg)/(360 deg/rev)= 1.8 rev, and (1.8 rev)/(10,000 rev/min) = 0.00018 min * 60 = 0.0108 s, or about 100 Hz. Yikes. OK, 100 Hz sounds a bit extreme, but I'd guess that something on the order of 15-30 Hz would probably begin to become perceptible (if I remember rate the sample rate for your eyes is on the order of 15 Hz).

As for internal intake sealing, no, it won't affect the harmonics. The resonant tube length is the same either way, and the tone will be the same. However, the leaks will reduce the pressure of the pulses, which is likely how you get a "collection of all local minima." http://fsae.com/groupee_common/emoticons/icon_biggrin.gif External leaks, of course, are simply the evil, infamous vacuum leaks that we all know and hate...

I remember seeing the variable intake system on my friend's '97 Cadillac Catera like Drew mentioned. It had a 3.0L Opel V6 (was a re-badged Opel Omega), and was lacking the trademark Cadillac low-end torque, so in Cadillac style they simply threw a bunch of doo-dads at it. At low rpm it acted like there were effectively two log intake manifolds, one for each bank, that went through two throttle bodies, through two intake tracts that were divided to a point about 16" or so upstream. Going up to mid rpm, a flapper near the throttle bodies opened, joining the two tracts near the throttle bodies, and shortening the effective tract length to just the log manifolds, like a dual-plenum manifold. Going to high rpm, a flapper in the rear of the intake plenum (opposite of the throttle body) opened, joining the two log manifolds together.

I might add I did an engine swap on that car, and I nearly had to seek counseling afterwards...the wiring was a nightmare, and I've blocked out most of the rest of the experience. I don't want to talk about it anymore. http://fsae.com/groupee_common/emoticons/icon_eek.gif It turned out fine, but what for me is normally a relatively simple one day job disappeared to like 3 or 4 days in hell.

Although, as has also been mentioned, I recall Toyota and Ford/Yamaha have done dual runner setups (with the short runners blocked by another set of butterflies) that didn't look so bad.

[Adam Vaughan]

Oh, throttle response is a different beast... I'm talking about the time it takes to go from 7,000 to 10,000 rpm with the engine loaded down on a typical endurance track drive cycle, and then matching the engine speed with an optimized runner length using a continuously variable intake manifold. Something in the neighborhood of 0.1 seconds would definitely be needed to track gear shift transitions through that range, but if you're shifting a gear every 0.1 seconds on the endurance track then you're probably doing something wrong. The design I mentioned is able to track engine speed as long as engine speed transitions were less than about 3,000 rpm/s (1 second for the full range). This rate is on par with straight line acceleration for typical track speeds when simulated in CarSim using our car's gearing, and can definitely be improved through a higher torque servo. I would hazard a guess that 6,000 rpm/s tracking can be achieved without much trouble.

[Kirk Feldkamp]

There's not exactly a simple answer to the RPM rate question. It depends on all sorts of factors. Power output, gear selection, what gear the driver is actually in, vehicle weight, drivetrain inertia, traction limits, etc, etc, etc. The better question is WHY?

Making a variable intake manifold that moves and has controls and such is an interesting project and all that, but it seems like an incredible waste of resources (time/money/manpower) for almost any FSAE team. I assume you're asking what RPM rates other teams are seeing because your team doesn't have a data system already. If you did, you would already have looked at your drivers' throttle traces, and probably even generated a histogram of throttle positions during a FSAE comp. You would have also taken a step back to think about whether or not the answer is indeed more cowbell, or better/more experienced drivers. For the *vast* majority of FSAE teams, the BIG gains to be found are in the development of drivers that can better use the power and gears that are already there. Unless your torque curve is a peaky, ugly mammajamma that tries to kill a driver every time they get the engine into its happy zone, then you probably don't need to do anything. If it is a beast, then maybe some relatively simple changes to your intake/exhaust lengths to smooth things out a bit would be a better path forward.

I am a firm believer that improvements in driver ergonomics and driver/vehicle interaction in FSAE tend to offer greater gains per unit resources than anything having to do with intake/exhaust tuning. There are all sorts of results over the years where better drivers and better sorted/developed cars whomped on cars with fancier bits on them that didn't have enough time for testing and development. Don't be that team!

[Adam Vaughan]

I agree... you don't take on such a project if you're just barely getting a car out the door without enough testing and development.

That said, at the end of the day FSAE is an engineering competition, and I feel that if you produce a car that's completely dependent on driver selection, then you haven't _engineered_ anything. Further, if you produce more or less the same car year-to-year because all you care about is getting the most driver training time possible, then you're not really gaining much more than proficiency with angle grinder while building cars (which is good skill to have, but also not engineering). In fact, many people outside FSAE (e.g school administrators) often wonder out loud why the same go kart is being built each year. I honestly wish I could say that generalization was completely false. The truth is that a lot of designs do get carried over as a matter of practicality, but I think the spirit of FSAE is pushing the envelope / exploring what's possible while still in school (like Western Washington's kick-ass ~550cc V8). Clearly, this is all after you've satisfied the reliable car objective / recruited decent drivers.

In terms of points, yes, producing a reliable car that crosses the endurance finish line is the most important thing to do, and it should drive all design decisions. In fact, we finished endurance without a variable intake manifold because the manifold wasn't done on time (this deliverable was at the bottom of the priority list and we had an alternate intake contingency plan anyway). I continued on with the project for my master's because I was curious what a practical design could achieve for FSAE after seeing various results in the literature.

As for why such a manifold is a good idea, well, most drivers in FSAE are not F1 class. I see smoothing out the torque curve as making the car more accessible to typical drivers without much experience. I think this is similar to your ergonomics objective, but definitely not 1:1. At the same time, my other goal was to improve overall engine performance in a way that didn't have as many destroy engine failure modes when compared to boosted designs / variable valve timing add-ons. I have seen so many teams go after the turbo goal, only to have the engine blow up. If you're curious what an awesome boosted design can get you, see William Attard's Ph. D. thesis: http://bit.ly/11TGtA0 . Besides great results, you should also note that there are pictures of broken pistons in there.

Finally, it has historically been hard for my team to get actual track data (it's not easy to find a parking lot to test in NYC). This is why I asked. It would be nice if my engine results weren't preliminary and had some VE numbers decoupled from the engine calibration, but here are some first pass results if you're curious: http://bit.ly/congressPrez . Whether the ~5% improvement in peak power (using a basic engine calibration and unoptimized cams) / the wider torque band is something your team should go after, well, that's a decision dependent on your team's objectives.