Turbo Boost AND Flow Control

[Gollum]

So from what I can remember and also find via google, we've danced around the subject time and time again, with Tony D often the instigator, but I'm gonna bring it up yet again because there aren't really threads DEDICATED to talk about it. It just comes up in related topics...

 

What's "it"?

 

Compressor bleeding.

 

Bleed excess air at lower RPM when running high boost to prevent the turbo from surging. As boost increases and the engine can't ingest enough air to keep the compressor happy, you vent the excess air to push the turbo further into the flow axis of it's map, thus moving away from the surge. In theory you could run a turbo literally RIGHT on the limits of surge without actually surging if you had enough control points, data, a controller capable of doing it all.

 

Also as the theory goes, a simple PID controller will do the job, which makes sense. I've now used PID controllers in the temperature realm for industrial application, and I can see how it would be easy to rig. The problem I'm having is that I can't find ANYTHING in the realm of AIRFLOW controlling PIDs, especially for automotive applications....

 

Tony "claims"  to know a guy who was controlling his Z32TT with a product on the market that seems to work for this application. I haven't seen any info on what this product is, if it's really plug and play, what it uses for sensor inputs, what type of programmable control you have over it etc....

 

 

Any why bring this all back up again? Because with MS3 and the MS3X expansion card, stuff like this should be old hat to set up, yet I'm not seeing it done. I plan to go MS3 this year finally, and I have big plans on how I'm going to use the grunt of the MS3X expansion down the road. I plan to run a staged injection with different fuel types, running sequential on the smaller pump gas injectors. I also plan to control fuel temps, which I already have a PID controller picked out in case I can't get MS3 to do it how I'd like. I'll also be running COP with individual spark trimming. All this fancy stuff, and I'll still have I/O available to use. The question is... is there code to support something like this?

 

 

I think it was also mentioned at one point, that this type of bleed system could also double as your throttle lift BOV as well. If it's input for vacuum/boost is post throttle plate then as soon as you slam the throttle shut and there's vacuum in the intake manifold again, the PID controller will see a severe pressure/RPM offset and will start venting pretty much everything. If you could hook your valve to an output from MS and actually build a table for it, you could literally map out just about any possible scenario.

 

 

The beauty in this type of a setup, and what still draws me to is, is that it's a form of VARIABLE BOOST CONTROL, without all the mucking about with crappy electronic boost controllers that can just about never seem to do their job right. You can have a simple single setting boost control, be manual or electronic, and set it for the required pressure for your power goals, and the bleeder takes care of the rest. You'll always get as much air as your engine will allow that compressor to give it at that given point.

 

 

And on that note, this also means you can run an absurdly offset compressor/turbine combo as long as you don't care about pressure ratios (which you shouldn't really, you should just care about how the dyno looks regarding how power is developed and how much of it). With a setup like this you can run a really small turbine, or a large turbine with a tiny AR, with a huge, MASSIVE compressor because WHO CARES if it's gonna surge like hell. You'll be avoiding that completely and there's a good chance that even with the bleeding you'll still be developing good flow/boost at lower RPMs still, but you'll have the benefit of that giant compressor generating much less heat at your HP goals. Will top end be hurt? Not extremely likely. Make sure you have a GOOD wastegate setup and overcome any flow restriction with more boost. More pumping losses? Probably, but who cares when the goal is raw power?

 

Anyway, my time is up for tonight. Hopefully I'll get some input for some of you people out there and we can work on drawing up some system diagrams.

[Tony D]

This is standard or industrial compressors.

Some people used turbocharger tachometers on their car....wonder why....

"What is Surge" in the FAQ.

 

DEADARSE sent me the instruction manuals for his system, and my laptop promptly DIED within a month without a backup.

 

During that month me and another engineer did 65 gearbox overhauls in Shanghai...so I never even looked at what he sent to be able to give you a name.

 

There was a Puerto Rican company claiming to make a controller that did this.

 

It is always easier to,dump boost than to control boost via controlling turbine speed. Do what you can to optimise turbine speed (reduce variability) and tune the PID for boost control. You get a MUCH better control of pressure ( and anti-surge)

[TimZ]

This is standard or industrial compressors.

Some people used turbocharger tachometers on their car....wonder why....

"What is Surge" in the FAQ.

 

DEADARSE sent me the instruction manuals for his system, and my laptop promptly DIED within a month without a backup.

 

During that month me and another engineer did 65 gearbox overhauls in Shanghai...so I never even looked at what he sent to be able to give you a name.

 

There was a Puerto Rican company claiming to make a controller that did this.

 

It is always easier to,dump boost than to control boost via controlling turbine speed. Do what you can to optimise turbine speed (reduce variability) and tune the PID for boost control. You get a MUCH better control of pressure ( and anti-surge)

 

On this note, I've had the idea for a while that perhaps an OEM solenoid operated diverter valve (like this, with an adapter like this), used in parallel with a conventional BOV might work reasonably well for bleeding off some airflow to prevent surge, and then the conventional BOV could handle the higher flow required for dropping throttle at full boost.  I don't know whether these are PWM-able, but since they are relatively small to begin with, a simple on-off map for RPM and MAP might suffice to get the turbo past the surge range without bleeding off too much.  If you could make it work, this would be much simpler than implementing a PID controller...

 

 

 

Thoughts?

Edited March 7, 2013 by TimZ

[Tony D]

Turbine speed moves surge line significantly... Best results would not come from pressure-only, but a mapped pressure -vs-speed using valve Cv to determine percentage open (bias on TPS signal and maybe a negative/positive rate-of-change slope to preemptive lay get it moving to bleed or build boost faster would be a way to do it...

[Gollum]

Turbine speed moves surge line significantly... Best results would not come from pressure-only, but a mapped pressure -vs-speed using valve Cv to determine percentage open (bias on TPS signal and maybe a negative/positive rate-of-change slope to preemptive lay get it moving to bleed or build boost faster would be a way to do it...

 

EG: NAWWWWWZ

 

As far as part application, I think the diverter valve Tim linked should work, wouldn't it Tony? We're not talked about bleeding off HUGE amounts of air since most setups will only surge for a small portion of operation, and it's during building turbine speed, like you noted.

Edited March 8, 2013 by Gollum

[Tony D]

I was on super slow connection yesterday...that's why I concentrated on control scheme rather than wether a valve was suitable.

 

Yes, the valve would do that electrically (BEGI Compressor Bypass Valve? Pneumatic equivalent...)

 

THe difference here would be how you control it... Pressure alone does NOT tell you approach to,surge. ENGINE RPM vs SURGE may get that closer, but Engine Speed/Turbine Speed/Pressure is something that could be plotted and calculated beforehand and a preliminary anti-surge venting scheme / wastegate control scenario could be implemented.

 

Basically bias turbine wastegate towards closed and optimum turbine speed based on engine rpm - flow plotted against surge line at that point to control PWM signal to valve to dump overboard excess FLOW (not pressure) allowing higher boost at lower rpms without the "Big Phil Issue" go surging at 4500 rpms and 20 psi, but not 5500 MP's and up...

 

This allows for MUCH HIGHER torque output lower in the rpm range than possible by simply turning down the boost and upping it once the rpms rise and engine flow is sufficient to prevent mi flow surge on he turbo.

[TimZ]

I was on super slow connection yesterday...that's why I concentrated on control scheme rather than wether a valve was suitable.

 

Yes, the valve would do that electrically (BEGI Compressor Bypass Valve? Pneumatic equivalent...)

 

THe difference here would be how you control it... Pressure alone does NOT tell you approach to,surge. ENGINE RPM vs SURGE may get that closer, but Engine Speed/Turbine Speed/Pressure is something that could be plotted and calculated beforehand and a preliminary anti-surge venting scheme / wastegate control scenario could be implemented.

 

Basically bias turbine wastegate towards closed and optimum turbine speed based on engine rpm - flow plotted against surge line at that point to control PWM signal to valve to dump overboard excess FLOW (not pressure) allowing higher boost at lower rpms without the "Big Phil Issue" go surging at 4500 rpms and 20 psi, but not 5500 MP's and up...

 

This allows for MUCH HIGHER torque output lower in the rpm range than possible by simply turning down the boost and upping it once the rpms rise and engine flow is sufficient to prevent mi flow surge on he turbo.

 

I was thinking of the electrically operated diverter valve mainly because I can't visualize how to plumb a "normal" pneumatically operated valve to open during the boost build region, where you are typically at WOT and don't get a pressure differential for it to work with.  I was thinking perhaps using the electrically operated valve in this region to bleed some flow and move to the right a bit on the compressor map.

 

If I'm reading your post correctly I think it might be possible to achieve something workable (probably not ideal but maybe pretty good) for the "boost build" region using a PWM duty cycle vs RPM and MAP table.  I don't think this would work as well for things like going to part throttle under boost but I was hoping to use the standard BOV for this part.  I'm mainly trying to figure out how to use a function that my engine controller already has in order to avoid having yet another auxiliary controller.  Whaddya think, Tony - am I full of sh1t, or am I on to something?

 

One question mark for the electric valve is how much boost it will hold without popping open - I've read through a bunch of forum posts (which was painful at times ) and a couple of people claim that it's working for them at 24psi or so, but I haven't seen anything higher than that.

[Tony D]

Boost is merely resistance to flow... Get proper flow, and boost is not necessary. Up the RPM's and flowing boost goes a long way toward making HP!

 

Making 650+ below 25psi now.... And not hitting the power breakpoint on the cam yet...

 

600hp at 15psi and 8,500 rpms? Sounds like a hell a fun ride to me compared to 647 at 5,500.... If I wanted that I'd be driving a 72 Big Block Corvette Roadster!

[Tony D]

You are not fecally engorged...

 

Sounds workable to me...

[Gollum]

You are not fecally engorged...

 

Sounds workable to me...

 

I really don't understand half the crap you say sometimes Tony. But it's entertaining nonetheless.

 

BTW: I got it after about the third time reading it, very funny.

 

The thing about bleeding excess pressure to get away from the surge area of the compressor map, is that it can be tuned, like anything else controlled by a computer, PLC, etc. There's really no need for 3+ sensors to make it "work" unless you don't ever want to have the need to tune the sucker.

 

Take any NA engine and dyno it. It's going to produce a reasonably consistent amount of power, and the torque curve will be for the most part similar every time. Yes there's variances due to outside temp. Yes there's variances for other reasons. But, it's relatively consistent. It's these consistencies that makes the world of carburetors still a viable option today. Run the engine a bit rich and a multitude of sins are forgiven.

 

Same principle could apply here if it had to. You don't need to be on the bleeding edge of surge, just close enough to be safe.

 

I wouldn't measure against MAP. I'd make a simple tuning map that works by engine RPM and turbo inlet MAS. If we know how much air the turbo is sucking, and what engine RPM we're at, then we can bleed off the required amount of that inlet air in order to prevent surge. Will it bleed off excess during part throttle? I'd think not since the inlet to the turbo will be lower as well.

 

Does that sound like it might work Tony? For me, I'd like to avoid having to run a turbo RPM sensor since if I do this I want to make it fairly simple just to show how easy it could be done.

 

With megasquirt you can have both a MAP and a MAS, and you don't necessarily have to use both for fuel calculations. What I'm unsure of, is if you're running a MAP like the "MAP Daddy" which has two MAP sensors, one for manifold and one for atmospheric correction, if the software will still recognize another input for MAP (I simply haven't touched the software interface myself yet so I don't know)

Edited March 9, 2013 by Gollum

[Tony D]

For a MAF based system it would be set up like airsep units. The thing is the surge point of the compressor wheel doesn't go by a set flow point and engine RPM, it goes by turbocharger rpm-vs-flow. For any given flow point on the turbo map, there will be a corresponding surge point based on flow. Doesn't MATTER what he ENGINE rpm is, really.... If you know turbo rpm, and the MAF Point going through the turbo, you know where the turbo will surge. Period. It goes back to me mentioning people who (in the eyes of some) are inveterate liars and who purposely mislead others in their quest for power.

 

These individuals have a turbocharger tachometer and used it quite effectively back in the early 80's...

 

But I digress....

 

The engine rpm might be used if turbo speed was held reasonably constant...meaning a controller on the wastegate to do that.

 

There is a method I saw on a unit in New Caledonia which utilised absorbed set of holes in the inlet pipe to the turbo, and a bleed orifice at the torus. The delta P on this sensor was compared to a square root output from that sensor. As we drove the machine into surge, we noted the data log at the point where we saw both flow reversal on the DP Raw input, and the square root output. We did this for various points (6 surge points) and then put it into Excel to plot the points, and give us the slope of the line on the surge curve. No matter what pressure we ran at, we'd accurately predict surge using those numbers. The slope was entered into the control system with a 2% offset, and that was used to control the venting of he machine.

 

That may work...

[Gollum]

Wait, aren't we over complicating this? A surge point has to do with a FLOW and PRESSURE point on the compressors map....

 

MAF on inlet, MAP on output. Done! Who cares what the turbo RPM is, as that only relates to efficiency? A simple setup like this you could combine with any wastegate control scheme, weather you wanted to control boost at certain RPM in order to make the engine more streetable, or a basic manual boost controller. If you had a graphical map on your tuning points then you could literally just copy the compressor map provided by your manufacture, assuming they make them public. From there you can fine tune it in case your system causes variations.

 

Can you tell I'm avoiding the idea of a $1500+ turbo tach install?

[Tony D]

Look at the maps, the surge point moves with rpm drastically. It does NOT "only relate to efficiency"!

"We discount the effect of ambient temperature for the purposes of further discussion..."

 

IF you hold speed constant, then pressure and flow are the surge variables.

IF you hold pressure constant, then speed and flow are the surge variables.

IF you hold flow constant, then speed and pressure are the surge variables.

 

This is basic centrifugal theory. The greatest contributors are flow and pressure, speed variation of as little as 5% wheel speed can change surge pressure 1bar.

 

This unfortunately is a complex subject which automotive manufacturers try to oversimplify. By complexitizing it just a smidgen, the anti surge, and plenum pressure control you get is worlds apart better!

[Tony D]

Example: steam turbine drives...

As refineries (always refineries...) reach surge problems or capacity (flow problems), their engineers ALWAYS call asking if its permissible to up turbine speeds "to get more air"?

 

As an OEM Tech Services Engineer, the answer I have to give is "no, there is interior aerodynamic parts that have worn and must be attended to..."

 

They never listen.... Six years later, they call. Over the years, as the machine encountered surging, they simply increase the turbine speed, and it goes away. But as pressure in the plant drops, they get to a point, eventually, that the turbine is now tripping on over speed, the machine can't make design pressure, and surges if an operator sneezes.

 

Then you go and replace the parts and amazingly it works again. And you show natural surge points: 1000kpa at 2970 input speed, 1100 Kia at 2980, 1200kpa at 2990, 1300kpa at 3000... Each step in input speed changed wheel speed 14X or 10X, depending on pinion attachment. So 140rpm change resulted in 100kpa change in natural surge point. Suction (throttled) surge point mirrored the change. These two points mirror upper and lower points on the surge line, hey give you that surge line slope.

 

Trust me, that surge line they give you is variable by compression ratio. And it does the same thing with speed as it does with ambient temperature, moves that surge point (fixing pressure) left or right along the horizontal axis. Operate lower on the vertical axis, your two points on the horizontal axis generally become further apart...

 

There are literally six-day seminars you can attend on the PID Control schemes to combat surge.

 

Current technology in FIXED SPEED centrifugal compressors is to use calculated poly tropic head to determine the surge point and keep it away from it. For industrial stuff, you want to run as low as you can, tight (within 1%) of the surge line at any given point. This is or electrical costs.

 

For us, we want to only impart enough excess flow to keep within 1% of targeted maximum boost possible for that control point to maximise torque and horsepower.

 

Simply put, if you simplify the control, you don't achieve that goal.

[Tony D]

Like I mentioned previously, the SIMPLEST and likely most versatile controller would be to read a differential pressure across the turbo, and chart the point where surge occurs. This seemed to be fairly reliable predictor as as it approached the crossover point, you could predict surge. It happened at Natural Surge, and at throttled surge. The number on the square root output from the DP sensor was almost identical at a 13 bar natural event as at a 10% IGV control point running around 4.5 bar.

 

It seemed to distill all the variables into a control point that seemed a consistent output regardless of pressure or flow. But you had to surge it once to find that point, then after that tune your system up to respond appropriately. In the industrial setting, they just open the dump valve and vent... Keeping pressure going downstream as constant as possible.