Interesting dyno test (cats and boost)

[engineermike]

I did some dyno testing to determine for myself, once and for all, whether it’s more powerful to run less boost and more timing or vice versa, and how much cats hurt power.

This is a stage 2 Whipple 2018 10r80, xdi pump, and full 3” exhaust. Figures are STD. Another note is that I was running about 1.5 deg less timing than I normally can due to a low local dew point today. This is on pure 93 octane shell gasoline with no boosters or additives whatsoever, so power is knock-limited.

1st test was using the 3.5” pulley which is 13-14 psi. Lambda was .78 and timing was knock-limited to a max of 17 deg. This was with 3” “test pipes” (no cats). It made 724/609, backed up by a very similar pull. My bet was that this would be the best numbers it would do.

2nd test was identical to the first, but using the Whipple supplied stage 2 3.875 pulley. Still .78 lambda, still no cats. But, at the 10-11 psi boost, the knock-limited timing rose to 20 deg. Peak numbers were 679/561. It made 6% less power with 10% less air, so the efficiency was improved due to the higher timing.

3rd test went back to the 3.5” pulley but we installed stock gt500 cats and richened up lambda to .73 to protect them. To everyone’s surprise, it made 737/617. Upon further investigation, I found that the timing was a little higher at 18 deg. This was still knock-limited but made possible by the enriched lambda.

I’m not super surprised by the low/high boost results but it is counter to what some major tuners have posted. I am very surprised by the fact that it gained power when cats were added and richer lambda. The 1 deg of timing helps but I didn’t expect it to offset all the losses plus some.

Sponsored

 

[Scootsmcgreggor]

Very interesting Mike, thanks again for taking the time to share the results of your testing with the community.

I find the result also quite interesting because I would expect a drop in power with the drop to .73 lambda from .78, not the other way around. .73 is quite rich, cat protection aside.

 

[cbrtrx]

What kind of dyno was used? You put the cats back on then ran it again the same day? I'm not that shocked by the results, like you mentioned it had more timing and the gt500 cats are not very restrictive at that power level.

 

[engineermike]

What kind of dyno was used?

Dynojet

You put the cats back on then ran it again the same day?

Yes, all runs were within a 2 hour span without even unstrapping the car.

 

[engineermike]

I'm not that shocked by the results, like you mentioned it had more timing and the gt500 cats are not very restrictive at that power level.

We know that you can run more timing if it’s richer, but usually best power is around .85 lambda and adding timing can only partially offset the losses. I guess that’s too much of a generalization.

 

[markmurfie]

We know that you can run more timing if it’s richer, but usually best power is around .85 lambda and adding timing can only partially offset the losses. I guess that’s too much of a generalization.

I know people say rich/ lean best torque is in a plateau between .8-.88 lambda, some say its a bit wider .78-.91 because of extra timing advance you can get in. Its not only an effect when detonation limited but on increasing MBT vs when at stoich as well. This is all great for WOT "power" enrichment, but not effeciency.
Best power though will always come from the engines volumetric effeciency and be at stoich AFR and MBT timing at stoich AFR which are the most effecient fuel ratio and timing. Absolute Best power is heavily dependant on maintaining engine load when going up to high RPMs. Heat becomes a major factor so its also not sustainable for extended periods unless its well designed to mitigate and manage that heat. Trucks, with engines designed for producing torque, will run stoich at WOT for short periods of time for better power and therefore fuel mileage. Just to make sure you are clear on the differences of best torque and best power when reviewing these things.

I'll assume a reasonably flat lambda through out these pulls. I'll also try to ignore the wierd graph truncations. I do not care about 3.875" vs 3.5".
Its really hard for me to draw the same conclusion(cats= more power) when I see the pull with cats not only has a little blip it goes to about 200-250 RPM higher, which can easily explain 10 HP. The curves look almost laid over one another. I wouldn't have guessed there was much lambda and timing difference at all.
The torque curves, which is what you should be looking at when modifying AFR for best torque, look identical. Pretty flat from 4500-6000, with little blips where the peak numbers come from. Again I wouldn't have guessed much lambda and timing difference at all.
Suprising to me is the FFE starts losing torque above 6K compared to the catted which maintains it better. This is a RPM region where, with a static timing value, VE usually begins to matter, and I don't see how cats had anything to do with increasing the engines VE, just not effecting it as negativly as expected. I don't see how cats allowed for more detonation resistance either. If cat protection rich allowed for another degree of timing with minimal effect on torque, this should have benefited the FFE as well. Maybe the best rich torque plateau is very wide for coyotes. Many tuners say they work better richer and more timing, which disagrees with emperical evidence on ICE's, especially thermally effecient ones. Why would it be different and make more power running richer and be more, not less, effecient? Some property of the combustion chamber... some distribution effect on the fuel in the charge... No one seems to be able to explain what they find about it. This is margin of error between pulls, numbers we ae discussing, so it can't be that serious.

Looks like more power will come with more RPM,it was still climbing.

 

[Angrey]

It goes without saying that adding more air is the dominant/compelling factor.

I think you'll find the disparity between the two gets greater as one reaches the limits of the fuel.

On a static 19 psi, my car made over 800 wheel on zero (TDC) timing. As we added in timing, the effects are muted at first, then grow to their peak, then wane.

In the prime/meat of the increments, adding each degree (at those general power/flow levels) resulted in over 30hp of rw results. Then, just like most competent tuning sessions, we push it until the additional timing starts to taper and result in less and less of an improvement, at which point more timing is just more risk without meaningful results.

Adding BOOST however is going to make significant jumps from ambient up to the edge of the fuel and charge cooling. The timing is simply the amount one can wring out of it.

It's obviously a complex combination when dealing with "safety" but I'd say that if you can keep the IAT's reasonable, the preference would almost always be more boost and less timing. But again, that assumes IAT's don't become problematic. As Mike pointed out above, fuel mixture gets a vote as well.

My philosophy has changed over the years to "overengineer and underutilize" and that's achievable for most amateurs and anyone not trying to wring out the absolute highest levels. If you want it to make 1000 hp, build it to make 1200 or 1300 and back it off. Then all these combinations are an option. When you start optimizing fuel (lambda) and then you max out boost (for IAT's and cooling) before you know it you're down on options to manage knock risk.

Start with fat(er) fuel, keep timing a few degrees back from optimal, back off the max redline a bit and push as much boost as the cooling (IAT's) will allow. Whatever that makes, that's your ticket. to a healthy longevity with big power.

 

[engineermike]

I know people say rich/ lean best torque is in a plateau between .8-.88 lambda, some say its a bit wider .78-.91 because of extra timing advance you can get in. Its not only an effect when detonation limited but on increasing MBT vs when at stoich as well.

At first, I assumed that the extra degree of timing was moving me closer to MBT, but (at least according to Ford) both the GT and GT500 MBT modifier as a function of lambda show MBT going higher as you go from .8 to .7 lambda. It just so happens that my borderline timing and MBT both increased by 1 deg when I added fuel, so the "distance from MBT" was nearly identical. Again, this is assuming the Ford curves are accurate in my situation.

Best power though will always come from the engines volumetric effeciency and be at stoich AFR and MBT timing at stoich AFR which are the most effecient fuel ratio and timing...Trucks, with engines designed for producing torque, will run stoich at WOT for short periods of time for better power and therefore fuel mileage. Just to make sure you are clear on the differences of best torque and best power when reviewing these things.

Are you saying that best power is at 1.0 lambda? I think what you are saying that best efficiency, best power per unit fuel, is best at 1.0. This is why trucks like some Ecoboost F150's run 1.0 at full load right up until the cats overheat. I believe the lambda efficiency correction in the calibration is actually an "air" efficiency not a "fuel" efficiency, as it's used to calculate actual torque as a function of air flow and shows peak efficiency in the .85-.87 range. Undoubtedly best fuel efficiency isn't at .85...

I'll assume a reasonably flat lambda through out these pulls. I'll also try to ignore the wierd graph truncations.

This brings up a good point, because the w/cat curve was not flat. It starts lean then richens up top just like Whipple, Roush, and Ford GT500 do. Below I graphed measured lambda and removed the 3.875 curve. Where lambda was nearly equal, timing was also nearly equal, and torque output was nearly equal as well. This is somewhat of a "control". But as rpm rose past 6000, lambda and timing started diverging and w/cats starts to show an advantage.

Its really hard for me to draw the same conclusion(cats= more power) when I see the pull with cats not only has a little blip it goes to about 200-250 RPM higher, which can easily explain 10 HP.

I would have thought the same, but we did do 2 pulls in each scenario and they were repeatable, within a couple of hp. In fact, the last 2 pulls w/cats were within 1 hp of each other and they were back-to-back pulls. Had I not seen this repeatability, I would have put less stock in all of this. Technically, I know 2 pulls in each trim aren't enough to establish scientific levels of certainty, but we are also fighting dynamic temperature effects (ECT, EOT, MCT, and IAT) so dragging it out for an entire day probably invokes even more uncertainty.

Suprising to me is the FFE starts losing torque above 6K compared to the catted which maintains it better. This is a RPM region where, with a static timing value, VE usually begins to matter, and I don't see how cats had anything to do with increasing the engines VE, just not effecting it as negativly as expected. I don't see how cats allowed for more detonation resistance either.

I agree and don't think the cats did anything to improve power ouput. I guess the 2 surprise learnings here are that the cats didn't hurt the power, and that best power lambda was a good bit richer than published data, as you pointed out.

If cat protection rich allowed for another degree of timing with minimal effect on torque, this should have benefited the FFE as well.

This is also a good point. I started the test under the assumption that .78 was on the rich side of optimum, so it never occurred to me to try a pull at .73 and no cats. In hindsight, the data leads me to believe that max power could have been achieved with no cats and probably around .74 lambda.

Maybe the best rich torque plateau is very wide for coyotes. Many tuners say they work better richer and more timing, which disagrees with emperical evidence on ICE's, especially thermally effecient ones. Why would it be different and make more power running richer and be more, not less, effecient? Some property of the combustion chamber... some distribution effect on the fuel in the charge... No one seems to be able to explain what they find about it.

Perhaps there is some uniqueness of the knock-limited 12/1 compression GDI engine running high-ish boost that falls outside of the empirical historical ICE data set in this regard. My math based on the data in the calibration and linear interpolation would say that .73 should make about 1% less power than .78 and that being 5 deg below MBT would have the same torque deficit in either case. Perhaps linear interpolation isn't accurate, the MBT calculation is inaccurate, MBT lambda correction is wrong, the timing S-curve changes when not at stoich, or the S-curve table scale is too low to extrapolate anything useful.

Logically, if the lambda efficiency curve is accurate (it appears to agree with published data), I lost 1% (7 hp) from lambda. It was able to run 1 deg more timing. If you assume MBT increased by a degree then I would have lost the 7. But if MBT stayed the same then the 1 deg timing would be worth about 20 hp, and the net effect makes sense.

 

[engineermike]

This got me thinking...everything is in the factory calibration to figure all of this out. However, since Ford never calibrated a 12/1 coyote on boost, I had to resort back to the stock calibration.

The graph below shows how MBT, borderline, and torque ratio (multiplier) move with lambda:

This graph assumes that at stoichiometry, borderline timing is 0 and MBT is 10. You can see that the peak power lambda is, in fact, around .86-.87 when running MBT. But at 10 deg below MBT, the distance from MBT starts to get a vote, resulting in peak power occurring closer to .84. The further from MBT you are, the richer peak power lambda will be. I just spot checked one point in the GT calibration, and at that point Ford expects MBT to be 8 deg above borderline, and they command .83 lambda not .86.

I need to automate that spreadsheet so I can easily substitute different distances from MBT and perhaps even different engine calibration data...

 

[engineermike]

I need to automate that spreadsheet so I can easily substitute different distances from MBT and perhaps even different engine calibration data...

Curiosity got the best of me so I did this. My spreadsheet is now automated and I have 3 pages pre-populated with calibration data from Gen3, GT500, and gen2 3.5 ecoboost. I did the ecoboost because it’s semi-high compression and boosted, usually running far from MBT. You just plug in how far you are below mbt at stoichiometry and it plots the torque ratio as a function of lambda.

The GT and GT500 both are set up from the factory about 8 deg below mbt before any corrections or knock advance. The ecoboost is more like 17 deg below. When it starts out 17 deg below mbt at stoich, best power lambda is actually 0.70 because it closes up the distance to mbt which helps a lot more than the efficiency reduction hurts. Guess what the ford PE is for that engine….yup it’s .70.

In fact, of the 3, ford commands richer than .86 for 2 of them, and it matches up to peak torque at borderline timing as opposed to peak torque at mbt. The gt500 commands richer than this optimum but my theory is they are using PE lambda to keep the cats cooler to delay COT.

Anyone have a guess as to how many deg of timing below MBT I would have to be to run 1.0 lambda at 14 psi boost?

 

[markmurfie]

This got me thinking...everything is in the factory calibration to figure all of this out. However, since Ford never calibrated a 12/1 coyote on boost, I had to resort back to the stock calibration.

The graph below shows how MBT, borderline, and torque ratio (multiplier) move with lambda:

This graph assumes that at stoichiometry, borderline timing is 0 and MBT is 10. You can see that the peak power lambda is, in fact, around .86-.87 when running MBT. But at 10 deg below MBT, the distance from MBT starts to get a vote, resulting in peak power occurring closer to .84. The further from MBT you are, the richer peak power lambda will be. I just spot checked one point in the GT calibration, and at that point Ford expects MBT to be 8 deg above borderline, and they command .83 lambda not .86.

I need to automate that spreadsheet so I can easily substitute different distances from MBT and perhaps even different engine calibration data...

I think the lambda correction for borderline timing is an artifact of the MBTorque lambda relationship. Same goes for the lambda effeciency correction on the torque tables.
You keep saying power lambda, Im sure you mean torque lambda. Unless you start factoring in RPM and VE, we are looking at properties of a single combustion event or atleast the combustion events of one engine rotation, which is torque not power. Tuning you should be lookig at the torque curve and totally ignoring the HP curve (it tells you nothing about the tuning). Power is just the product of RPM and the best torque curve.

Stoich is the most complete reaction and will output the most power. Its only by spark ignition ICE design, that timing becomes a factor and richer lambdas which have a slightly faster reaction speed making the MBT vs lambda relationship more favorable to rich of stoich. In HCCI engines its stoich lambda and TDC where you want the charge to ignite. In SCCI(diesels) they actually are on the lean side of stoich claiming they are finding power. As charge densities go up, you get closer and closer to what an HCCI engine does. I don't see how this timing plateau can be extended towards rich lambdas, but I do see how it extends moving towards stoich for more power. I also don't see how the rich side of this MBT plateau with increasing MBT is benefiting an engine thats detonation limited by gaining an extra degree of timing. The MBT is increasing there because of worsening combustion so the same "detonation limited" S curve rules don't apply.

All engines running rich bog, NA coyotes do it, boosted coyotes do it. Very unlikely this bog is of lesser effect than the increase in detonation resistance the little extra fuel is giving it. Lambda effect on detonation is not that great, its not like you gain a bunch more evaporation from .85 lambda to .75 lambda. Its also very hard to control every dynamic effect on detonation and isolate just different lambda's effect. Pedicting anything out of a calibration assuming you will be detonation limited is not something you should be tring to do. I think the models and corrections around it are too general. If they ween't, there would be no need for the knock sensors and computer advance at this point with how many corections there are. Thats basically what the borderline corrections ae doing is putting the advance on a better starting point to reach actual borderline or MBT. You certainly shouldn't take the borderline lambda correction 6250rpm column and apply it to a 10* detonation limited scenario that would probably only occur at low much RPMs. Hopefully you see and agree the borderline correction is mostly just an artifact of the MBT lambda correction applied over the RPM range. Not an actual representation of lambda effect on borderline at those RPMs like MBT lambda correction is.

 

[Stymee]

U do a lot of tinkering, tuning and testing and I love your posts but imho it would be more interesting to see how it would've reacted to the stock GT cats vs no cats. The GT500 cats may have been the whole reason why you seen those results and like you said adding them can be a pita.

If u were to test your results with lets say 3 or 4 cats manufactures that would even be better but thats a lot of work

EG….

Green Cats
GT500 CATS
There’s another Highly Efficient Cat?

Would love to see those results

But still in the end, if your tuning isnt spot on and any cat melts and bricks up on boost, BAD THINGS CAN HAPPEN real fast

 

[Prodigal]

This thread is exactly why I can’t convince myself to pull the trigger on FI and probably never will. It was all Greek to me. Yeah, I would love the power but I have no idea what lambda and half of the other stuff mentioned even means and I don’t have the time or resources to learn it all. I would need to have a tuner on retainer to keep my car running optimally. Truth is a strong and reliably running NA car is probably best for me. For you guys with the skills and know how, I am slightly jealous.

 

[andrewtac]

This thread is exactly why I can’t convince myself to pull the trigger on FI and probably never will. It was all Greek to me. Yeah, I would love the power but I have no idea what lambda and half of the other stuff mentioned even means and I don’t have the time or resources to learn it all. I would need to have a tuner on retainer to keep my car running optimally. Truth is a strong and reliably running NA car is probably best for me. For you guys with the skills and know how, I am slightly jealous.

it applies to NA as well

 

[Stymee]

This thread is exactly why I can’t convince myself to pull the trigger on FI and probably never will. It was all Greek to me. Yeah, I would love the power but I have no idea what lambda and half of the other stuff mentioned even means and I don’t have the time or resources to learn it all. I would need to have a tuner on retainer to keep my car running optimally. Truth is a strong and reliably running NA car is probably best for me. For you guys with the skills and know how, I am slightly jealous.

U don't need to read or tune using lambda for a reference, a wide band reads the same A/F…I never ever used Lambda reference on my LS cars when Don Kinder helped me tune the car.

(Slowhawk Performance)

I used a LM2 with a w/b sensor in the header, LS1 Edit and a data logger and he sent me updates

It worked flawlessly

That said, imho a Boosted Mustang needs a good fuel system. No cats, great tuning, free flowing exhaust and very good fuel (93 Pump gas imho is playing with fire unless u mix some 260 in the car or octanium because it’s unpredictable) Keep boost under 12lbs and you’ll have a killer setup on a tire and supporting susp. mods

Sponsored