What's new
Van's Air Force

Don't miss anything! Register now for full access to the definitive RV support community.

How to measure power gain?

blaplante

blaplante

Well Known Member
I'm thinking of some small changes that should increase the io-320's power output, potentially 1 or 2%. I'd like to be able to measure the performance, or relative power, to confirm the change actually did help.

Does anyone have a good idea on how to do that?

Current thoughts: measure WOT top speed, 3 way course. It needs to be done at the same density altitude for both tests, and the change isn't quick to do. Plus, any power improvement only increases top speed by the cube root (approx), so a 1% power increase will show as much less than a 1% speed change, which gets it down in the noise.

Rate of climb is much more sensitive, as it is proportional to excess power (above minimum flying power req'd). So a 1% power change gives at least a 1% increase in Rate of Climb. But it is totally sensitive to weight (I expect a 1% change in weight would also make a 1% ROC change). And, the rate of climb needs to be done across the same density (or pressure?) altitudes. Again repeatability within 1% seems difficult and so the change could get buried in the noise and measurement error. 1% of 1500 fpm is just 15 fpm.

Ideas? Strain gauges?

Thanks, Bryan

PS - fixed pitch prop, but I do have MP read out to 1/10 "
 
Last edited:
Fuel Burn?

How to you get the mixture re-set to within 1%? My change may affect the air/fuel ratio, so simply running full rich may not get the same AF ratio before and after the change. I guess I could install an oxygen sensor, which would work (for a while) before the lead kills it.
 
You won't be able to measure valid HP gains of 1% via flight testing, looking at speed gains, especially if you change more than one variable- AFR, humidity etc.
 
Be organized.Very!

You will have to know your planes performance vs density, gross weight, temperature, and humidity, and use the engine correction factors. Not an easy task, but doable.

BTW is is very difficult to accurately measure and correct engine power to 1% and be repeatable year round. Not to mention the other factors for flight testing.

Despite how this sounds, I do testing the same way, but try to keep the configuration changes quick and test on the same day or the same time the next day. Then attempt to hold as many variables as possible the same. Just keep mind to create a test program to repeat the configuration and see if results follow. I did engine development in a lab for a couple of years after doing performance analysis for 4 yrs. Test repeatability for a string of changes is extremely helpful in progressing in a known manner. Otherwise one just gets confused.
 
Last edited:
DanH said:
Works very well, but not cheap, as you need telemetry to get the signal off the rotating shaft. Good news? Excellent equipment is available.
Click to expand...

Where would you put the strain gages, Dan, inside the crank nose, it would need to be hollow in the front bearing area? That might be the cleanest location, for length, to get torsional data. Then the data processing would get interesting but certainly doable. We used strain gage resets on a machines shaft section, and picked up the signals with a ring antenna around the section. The strain gages were self powered (induced I suppose).
 
1% power changes are within the scatter of most decent engine dynos and even each individual power pulse varies by more than this typically if you plan to use strain gauges.

1% power gain would make no perceptible or useful gain in speed in any case.

In testing, apply the scientific method for the most valid results- change only one variable at a time. This means same CHT, same oil temp, same OAT, same baro, same AFR, same ignition timing, same altitude, same humidity, same speed, same rpm, same MAP etc.

Now you see why it would be so hard to draw valid conclusions about changes in hp of 1%.
 
Last edited:
BillL said:
Where would you put the strain gages, Dan, inside the crank nose, it would need to be hollow in the front bearing area? That might be the cleanest location, for length, to get torsional data. Then the data processing would get interesting but certainly doable. We used strain gage resets on a machines shaft section, and picked up the signals with a ring antenna around the section. The strain gages were self powered (induced I suppose).
Click to expand...

All he wants is torque, not frequency, so the easy approach is a short shaft extension (a spool) with the Wheatstone bridge and transmitter pre-installed.

Ross, did you notice much live torsional telemetry at Reno this year?
 
DanH said:
All he wants is torque, not frequency, so the easy approach is a short shaft extension (a spool) with the Wheatstone bridge and transmitter pre-installed.

Ross, did you notice much live torsional telemetry at Reno this year?
Click to expand...

I had talked to at least one team who said the system was having issues returning accurate data. Still working on it.

Certainly very valuable for these folks if they get it to work reliably and with good repeatability.

I believe there are at least 2 different systems being used at Reno.
 
Last edited:
I've tried various methods over the years - the quickest, most repeatable and simple method i've used is to simply line up the aircraft on the runway numbers on a flat calm day, firewall it, and read the gps speed over the numbers at the far end from 20' off the ground.

My home runway was almost 1km. The runway altitude, 20' climb and runway length never changed, the gps speed was always accurate, the "flat calm" could be judged very accurately, even the tower became an inadvertant helper (by giving timely & accurate QFE and ambient temp).

My thinking was that by reducing the myriad of variables involved in "at altitude" testing and making the actual test consistent (just firewall it and read the speed) the result would be accurate and very consistent from test to test.

As I made changes I certainly found this method very reliable (with no change I could get EXACTLY the same result 10 times in a row) And yet also see a measurable difference from even a small change such as a 10kg weight change or air filter mod
 
strain gauge and other feedback

I agree a 1% change not going to be 'significant'. But a few 1% changes can be valuable. If I put 5 changes on, and measure a 5% improvement - what do I know? Heck, one of the changes could have caused a 1% regression and we'll never know!

Exercise for the reader - how much % would be a 1" prop pitch change?

Regarding pancakes - haven't had one in years. My BMI is 21.5 ;)

We're just looking a *relative* power here (hence relative torque, relative force) vs. an absolute measurement ... (precision vs accuracy).

How about a strain gauge on the engine mount diagonal brace? Force is turning the prop, but also being resisted by the engine, via the mount, via the airframe. I have no idea if there's a measurable force - but there should be A force.

Glad to see I've got some wheels turning in the group :)
 
Pittsartist said:
I've tried various methods over the years - the quickest, most repeatable and simple method i've used is to simply line up the aircraft on the runway numbers on a flat calm day, firewall it, and read the gps speed over the numbers at the far end from 20' off the ground.

My home runway was almost 1km. The runway altitude, 20' climb and runway length never changed, the gps speed was always accurate, the "flat calm" could be judged very accurately, even the tower became an inadvertant helper (by giving timely & accurate QFE and ambient temp).

My thinking was that by reducing the myriad of variables involved in "at altitude" testing and making the actual test consistent (just firewall it and read the speed) the result would be accurate and very consistent from test to test.

As I made changes I certainly found this method very reliable (with no change I could get EXACTLY the same result 10 times in a row) And yet also see a measurable difference from even a small change such as a 10kg weight change or air filter mod
Click to expand...

Ground speed affected by wind which might not be a constant during your run. You still have to control all the variables I previously mentioned to be sure they are not skewing the results.

Measuring torque directly via strain gauge with good averaging software is the best. Speed measurement without long a stabilization period and reciprocal or triangular course using GS, the worst for accuracy.
 
Static full-power RPM?

OK, this assumes the changes you plan to make affect engine power at all power settings and that you have a digital tach (EMS) with reproducible results.

You should be able to figure out if using RPM as a surrogate for power will work by making a graph of full-power static RPM vs DA then using the graph to predict full power static RPM for a run. If predicted RPM from the graph equals measured RPM plus/minus 5-10 RPM I think you'd be able to detect changes in power output down to roughly 1%.

Assuming you do the tests at the same spot on the same field and have an accurate DA any increase in power should result in an increase in RPM on a fixed pitch prop.
 
A simple point, but one often forgotten here on VAF: when you have measurement error, as we do when evaluating aircraft performance, just because you can't measure any increase in speed due to a mod, doesn't mean that mod had no effect. In statistics we call the failure to detect a real effect a type II error, and we always keep in mind that lack of evidence for an effect is not evidence for lack of an effect. We should also remember that the cumulative effects of attention to many individual details, each with undetectable small effects, can indeed add up to a significant speed increase.
 
I just went through this with measuring the effect of ignition advance on cruise speed.

Two data points: ROP max power. The only change was the ignition timing: no effect on TAS.

The second one was lean of peak, economy cruise. Ignition timing optimization gave me a 2-3 Knot cruise speed increase, representing a 5.2% increase in horsepower or a 1.7% increase in efficiency (mpg).

No other variables were involve... rpm, mixture, altitude, mixture were left alone.


V
 
I've been playing with my a/c system to see what kind of decrease in airspeed there is while compressor is engaged. Was looking at IAS on my G3X because its resolution is 10ths of a knot. The compressor draws about 3HP, which at 23/2300 is a ~3% power drag, and IAS varied by about 0.2 kts. Had autopilot engaged and let speed normalize for several minutes with compressor engaged/disengaged to try to get reasonably valid data.
So a 1% power difference would be exceedingly hard to measure.
 
Years ago when I was modifying my plane for racing I used the three course gps program to record a LOT of speed mods. Results were inconclusive due to variances in the weather. Temp, rising or falling pressure systems, humidity etc etc.
I took all of these variable out of my testing by flying beside a buddy in his airplane.
Fly beside your friend so that you are exactly the same speed. Record exact power settings, fuel flow, manifold and rpm. Get your friend to do the same in his airplane.
Go and make your changes; to the engine, airframe, what ever. Now fly beside your friend again using the exact power settings for each airplane as you used in the previous flight. If you are pulling away from him your mods worked.

The beauty of this system is that it takes all the weather variables out of the picture and clearly shows a difference in performance.
 
Because you can't control all the variables from one flight to the next you don't know which ones hurt speed and which ones help. Stack ones that help and you think you made a gain but you really didn't and vice versa.

Yes, several 1% gains can lead to measurable results but individually, you'll be hard pressed to see each 1% change so that you can prove that one as a keeper.

You're splitting hairs with measurement errors and trying to control a dozen variables to get valid data.
 
aturner said:
A simple point, but one often forgotten here on VAF: when you have measurement error, as we do when evaluating aircraft performance, just because you can't measure any increase in speed due to a mod, doesn't mean that mod had no effect. In statistics we call the failure to detect a real effect a type II error, and we always keep in mind that lack of evidence for an effect is not evidence for lack of an effect. We should also remember that the cumulative effects of attention to many individual details, each with undetectable small effects, can indeed add up to a significant speed increase.
Click to expand...

In other words. "You don't know what you don't know";)
 
hgerhardt said:
I've been playing with my a/c system to see what kind of decrease in airspeed there is while compressor is engaged. Was looking at IAS on my G3X because its resolution is 10ths of a knot. The compressor draws about 3HP, which at 23/2300 is a ~3% power drag, and IAS varied by about 0.2 kts. Had autopilot engaged and let speed normalize for several minutes with compressor engaged/disengaged to try to get reasonably valid data.
So a 1% power difference would be exceedingly hard to measure.
Click to expand...

Yep. And the 1/10's that you G3x display mean squat. The data's accuracy is limited by the most inaccurate input. I am not sure which that is. I doubt Garmin's speed is that accurate off of pitot data. Maybe filtered WAAS GPS is that accurate, but I doubt it.
 
The challenge remains :)

Comments on some ideas - static rpm. I thought about that, but you'd need to lean to peak power (to be at a consistent mixture), and running full power for a while as I fiddle with the mixture isn't something I'm that comfortable with. [Hint my change has shown in automotive testing to have a bit of effect on the mixture.] Plus you need the density altitude (and oil temp) to be the same as the prior test. Better have calm air too.

Speed over the numbers... boy at that point these planes are accelerating quickly - trying to read the airspeed at the same exact point sounds challenging. And you need to be at the same mixture setting every time, which has the problem noted above.

I'm also starting to see that even if you have an engine dyno, measuring a 1% difference would be a big challenge. Density altitude, oil temp, mixture settings are all variables trying to mess you up.

Clearly the best (like the air conditioner test) is where you can do a/b testing *on the same flight*. If anyone has an idea on how to swap an oil sump in flight, let me know! :D

No reactions to a strain gauge on the engine mount?
 
Last edited:
You must log in or register to reply here.
Back
Top