Thrust and Impulse variance in commercial APCP motors

[Steve Shannon]

I may not be following this, but that is just NAR certified motors. The full list with TRA/CAR is
https://www.nar.org/SandT/pdf/CombinedMotorsByImpulse.pdf
Cheers, Alex

Thanks, Alex, you're right. That's the link I wanted to post.

 

[boatgeek]

The

I have heard of top TARC teams buying large numbers of motors, and then weighing them to try to weed out any that might have a tad bit more or less propellant.

The TARC team I mentor opens several motors (CTI, 24mm 2G), weighs grains, and then mixes and matches to get consistent propellant mass. That seems to work well. The max-to-min variation is on the order of 1-2% of total grain mass, including ignition pellet and casting tubes.

 

[jqavins]

In a funny sequence of events, I did some googling to find just this sort of information (though I was looking for thrust variation, not impulse) and stumbled onto a NASA page about something else, the reason for the typical shape of BP engines' thrust curves. I found the answer I was looking for at ThrustCurve.org, then I came here to the propulsion forum to post a link to the NASA site, and here's this thread about motor performance variation. Look for X, stumble on Y, find X, share Y, stumble on X again!

Peak thrust and burn time will usually vary much more than total impulse though.

According to John, the limit for thrust variation is 5%. Actual typical variation I don't know.

 

[sriegel]

This is one of the reasons full K and larger motors must be used with electronic ejection systems. We learned in the early days that pyro delays in large motors burned much longer in high altitude flights than when tested closer to sea level.

I had a J motor delay run a full 7 seconds longer than drilled in Alamosa, CO, last year. They're up around 7K MSL.

 

[Grog6]

I just reread this thread, and I noticed something that caught my attention: apcp motors burn longer at elevated altitudes.
I have to wonder WTF? as the burning environment for the propellant is pretty isolated from atmosphere; it's at hundreds of pounds of pressure, separated from the rest of the world by a nozzle.
Any Ideas?

 

[AlexBruccoleri]

I just reread this thread, and I noticed something that caught my attention: apcp motors burn longer at elevated altitudes.
I have to wonder WTF? as the burning environment for the propellant is pretty isolated from atmosphere; it's at hundreds of pounds of pressure, separated from the rest of the world by a nozzle.
Any Ideas?

After burnout the chamber pressure is quite low. Not sure how the delay grain/residual propellant pressurize the chamber though.

 

[Johnly]

If the gas velocity through the nozzle throat is supersonic, the outside atmospheric pressure has no effect of the propellant combustion process.
The term often used for this condition is "a choked nozzle". Once the velocity though the nozzle throat goes subsonic it becomes "unchoked" and the atmospheric effects on the burn rate become apparent. as evidenced by the change in the burn rate of the tracking/delay grain at high altitudes.

 

[Alan Whitmore]

I tell beginning rocketeers that the commercial APCP motors we fly from Cesaroni and Aerotech can vary in Thrust and Impulse for any two of the same motors, maybe by about 10%. I had heard that from someone else, or maybe read it somewhere, though I also seem to remember that the certifying authorities require the values measured to be within 20% of the claimed/published values. The message that I try to bring to beginners is why we work with safety margins, like a minimum 5:1 thrust to weight ratio.
Rocketry isn't an exact science, and there is probably more variance between flights because of wind or other factors than motor variance.

I'm curious as to what contributes to any variance, and whether it does vary much between motors. If you take five identical motors from the same manufacturer, with different dates or batches, some brand new, some 3-4 years old, even motors from the same batch, and bench test them, will there be significant variance? 10% seems like a lot, 20% even more unbelievable, though I don't have the luxury or equipment to make the tests myself.

I'm also curious as to what may cause any variance, and whether some propellant formulations are more prone to variance than others. Do slow burners vary more than fast burners, for example.

Just something to talk about after the launch....

I just spotted this thread. If you check in on www.thrustcurve.org you will find links to lots of the official certification letters. The data in those letters includes standard deviations from the mean for all of the measured parameters, if enough motors were tested to make such a calculation meaningful.
I'm sure the statisticians on this forum can give a better explanation for why we use standard deviation rather than variance for our qualifications. In my professional career, I tended to use variance and standard error when dealing with proportions (x/n) and standard deviation when dealing with numerical quantities. Might be wrong, that's just the way I did it.
There was one experiment that we (the TMT committee) did back in May of 2017 that is relevant, and which has never been reported. I had had some discussion with manufacturers about whether the orientation of a motor during the test burn affected the actual delay element timing. So, Tripoli bought me 12 AT I161W reload kits, and I fired 4 of them pointing up, 4 pointing sideways, and 4 pointing down, the orientation under which most rocket flights take place. Setting up the load cell so that the motor hung from the cell rather than pressed on it was a challenge, but we worked it out. Bottom line, orientation has no effect on delay element timing.
However, 12 motors gives a number that can be profitably analyzed statistically.
parameter average standard deviation
burn time 2.379 sec .086 sec
Total impulse 455.69 N.s 11.0 N.s
Average Thrust 191.65 N 7.18 N
Max Thrust 271.14 N 12.56 N

Alan

 

[Spacedog49Krell]

Attached is variance data I've recorded from Aerotech G40-10 motors. Some of the variances are extremely short lived and can only be detected with high speed flight computers.

 

[OverTheTop]

Variation

 

[kjhambrick]

Attached is variance data I've recorded from Aerotech G40-10 motors. Some of the variances are extremely short lived and can only be detected with high speed flight computers. View attachment 597916

@Spacedog49Krell --

My eyeball planimeter says the total Impulse MIGHT be a little higher for the 'Anomoly Motor' ( the red line ) ...

I used to see similar differences in the olden days when I extracted Thrust Curves from 8-bit AltAcc Data.

Have you tried integrating the area under the two curves ?

After adjusting for the 1G offset and Drag-vs-Velocity I'll bet the variation is negligible or at least within spec.

-- kjh

 

[rharshberger]

@Spacedog49Krell --

My eyeball planimeter says the total Impulse MIGHT be a little higher for the 'Anomoly Motor' ( the red line ) ...

I used to see similar differences in the olden days when I extracted Thrust Curves from 8-bit AltAcc Data.

Have you tried integrating the area under the two curves ?

After adjusting for the 1G offset and Drag-vs-Velocity I'll bet the variation is negligible or at least within spec.

-- kjh

a sample size of two motors is too small to call a single motor an anomaly.

 

[kjhambrick]

a sample size of two motors is too small to call a single motor an anomaly.

I agree @rharshberger

And the total impulse for @Spacedog49Krell's 'Normal Motor' and 'Anomoly Motor' look pretty close by the eyeball method of integration.

-- kjh

 

[JohnCoker]

What's most important is total impulse, so you really need to integrate. Here are (my excerpts) from the NFPA regs:
https://www.thrustcurve.org/info/certification.html
That said, I would expect commercial motors from the same batch to vary much less than that, but none of the manufacturers publishes quality metrics that I am aware of.

 

[OverTheTop]

A quick way of integrating and comparing the outcomes is to print the graph twice and cut out each graph and the zero line. If you have accurate scales you just weigh the cutouts. Integration is just the area under the graph, which is proportional to the density of the paper. That's how chemical analysis using spectrometers used to be done before the advent of digital computer interfaces.

 

[Spacedog49Krell]

@Spacedog49Krell --

My eyeball planimeter says the total Impulse MIGHT be a little higher for the 'Anomoly Motor' ( the red line ) ...

I used to see similar differences in the olden days when I extracted Thrust Curves from 8-bit AltAcc Data.

Have you tried integrating the area under the two curves ?

After adjusting for the 1G offset and Drag-vs-Velocity I'll bet the variation is negligible or at least within spec.

-- kjh

The total-impulse between the two motors is within specifications. This is 16-bit data recorded every 2.5 ms of either slag or a small chunk of unburned propellant passing through the nozzle based on the time into the burn. My question is, "How many of these millisecond events have occurred? Are these events common or rare? Are there other millisecond events that we should be concerned about? Could CATO's be related to a nozzle blockage issue?"

 

[kjhambrick]

I like your data @Spacedog49Krell

Thanks for sharing it !

I don't know the As to your Qs but there are all kinds of unimaginable chemical and physical events that will combine in unknown ways to make each motor as different as are snowflakes -- they're all different but they're much the same in the important ways ( total impulse and general shape of the thrust -vs- time curve ).

One thing I found useful when flying an accelerometer is to be sure to record the Motor Manufacturer's BATCH ID from the paperwork ... you might see something useful in that info.

One thing I see is that the deceleration after burnout is significant in your data plots. I am not sure of the units ( ??? Gs ??? )

I've got an awk script that I wrote specifically for 8-bit, 16 Hz AltAcc Acceleration Data that more-or-less automates the process of extracting thrust curves from raw accelerometer data and a launch parameter file.

In keeping with the Black Sky naming scheme, the script is called PROPULSE.

I plan to port so it can process Blue Raven Data when I can get around to it.

In a nutshell, the script takes the rocket characteristics and launch site conditions and then back-calculates Force -vs- Time from the raw acceleration data by factoring in the force due to drag during the motor burn phase.

There are some invalid assumptions in PROPULSE.

The most glaring issue is that PROPULSE assumes drag -vs- velocity during the burn phase is the same as it is during the coast phase. It is not.

Another less important thing I never bothered with was the rate of consumption of the delay grain and the effect of the hot smoke from the delay on drag during the coast phase. But I figured it was insignificant with large enough rockets and motors.

The script is very specific to AltAcc Data Sets but porting it for other Accelerometers is absolutely doable,

Be happy to share it if you want it.

-- kjh

p.s. @OverTheTop --

Please stop tickling my lab tech nostalgia bone. I sold my micro beam balance when I moved from the SF Bay Area to San DIego ( but I did keep my K&E Polar Planimeter )

EDIT: these are PROPULSE plots from AltAcc Data for three G40's I flew at Ocotillo, CA in 1997 ...

 

[FredA]

The most glaring issue is that PROPULSE assumes drag -vs- velocity during the burn phase is the same as it is during the coast phase. It is not.

What's different? Base Fill?

 

[kjhambrick]

What's different? Base Fill?

I am not sure @FredA

I am not sure what Base Fill is ...

But there will be a pressure differential behind the rocket when the motor is burning vs smoke only after burnout during the coast phase ...

Maybe it does not matter ?

It's a Q someone asked me back in the rec.models.rockets days that I couldn't answer

'How do you account for different base drag during burn and coast phase ?'

As for 'it is not' there was a long discussion and I had no arguments that drag was was the same ...

-- kjh

EDIT: pressure -> a pressure differential

 

[waltr]

The total-impulse between the two motors is within specifications. This is 16-bit data recorded every 2.5 ms of either slag or a small chunk of unburned propellant passing through the nozzle based on the time into the burn. My question is, "How many of these millisecond events have occurred? Are these events common or rare? Are there other millisecond events that we should be concerned about? Could CATO's be related to a nozzle blockage issue?"

First, I agree with everyone else that the Total Impulse of those two motors are close enough to be within specs..

If the 'spike' was caused by "a small chunk of unburned propellant passing through the nozzle", my concern is that it may cause a "Vectored" thrust. I and others have seen enough flights where the rocket left the rail straight then at a distance above the rail the rocket made a fairly sharp turn (maybe only a few degrees but sometimes much more).

Also have been plenty of posts asking why the rocket made a sharp turn. This may be the reason.

 

[kjhambrick]

Good conjecture @waltr

I am with you.

One reason I am so exicted by the Blue Raven Data is that it records both gyro and acceleration with 3-DoF each.

At the very least, the sharp turn should be documented in the tilt data.

And if the rocket makes a sharp turn due to vectored thrust then there should be evidence in the Accelerometer and Gyro Data -- maybe a spike like @Spacedog49Krell recorded accompanied by a sharp tilt ?

I dunno yet but I sure do want to look for such events

-- kjh

 

[Spacedog49Krell]

I like your data @Spacedog49Krell

Thanks for sharing it !

I don't know the As to your Qs but there are all kinds of unimaginable chemical and physical events that will combine in unknown ways to make each motor as different as are snowflakes -- they're all different but they're much the same in the important ways ( total impulse and general shape of the thrust -vs- time curve ).

One thing I found useful when flying an accelerometer is to be sure to record the Motor Manufacturer's BATCH ID from the paperwork ... you might see something useful in that info.

One thing I see is that the deceleration after burnout is significant in your data plots. I am not sure of the units ( ??? Gs ??? )

I've got an awk script that I wrote specifically for 8-bit, 16 Hz AltAcc Acceleration Data that more-or-less automates the process of extracting thrust curves from raw accelerometer data and a launch parameter file.

In keeping with the Black Sky naming scheme, the script is called PROPULSE.

I plan to port so it can process Blue Raven Data when I can get around to it.

In a nutshell, the script takes the rocket characteristics and launch site conditions and then back-calculates Force -vs- Time from the raw acceleration data by factoring in the force due to drag during the motor burn phase.

There are some invalid assumptions in PROPULSE.

The most glaring issue is that PROPULSE assumes drag -vs- velocity during the burn phase is the same as it is during the coast phase. It is not.

Another less important thing I never bothered with was the rate of consumption of the delay grain and the effect of the hot smoke from the delay on drag during the coast phase. But I figured it was insignificant with large enough rockets and motors.

The script is very specific to AltAcc Data Sets but porting it for other Accelerometers is absolutely doable,

Be happy to share it if you want it.

-- kjh

p.s. @OverTheTop --

Please stop tickling my lab tech nostalgia bone. I sold my micro beam balance when I moved from the SF Bay Area to San DIego ( but I did keep my K&E Polar Planimeter )

EDIT: these are PROPULSE plots from AltAcc Data for three G40's I flew at Ocotillo, CA in 1997 ...

Wow, old Blacksky AltAcc 1 data graphs, back when the G40 was 112 N*sec. Around 2000, Scott heard that I was flying 12-bit data flight computers that recorded acceleration every 10ms and altitude every 40ms. A coding structure I still use today, just faster. I had a few conversations with Scott on writing new code, I believe, for the AltAcc 2(?). My day job travel schedule didn't work out for the project.

The graph units are in G's. I also found it useful to record the motor batch numbers. Aerotech requested them for the two anomaly data files I have sent them. From my experience, motor anomalies are not common. They are interesting to study when they occur. One event that is common to all motors has been the ignition sequence. Recording data in the millisecond realm, I see ignition differences between BP and APCP motors that are interesting to study. Another interesting area that requires further study has been ground impacts when parachutes fail to deploy. Those few milliseconds of data reveal interesting events.

Thanks, but I have an Excel file that does the same as your PROPULSE. I'm working on a sensor cluster that will measure the actual base drag conditions during the flight. No more assumptions on base drag. It will be in the recorded data set for post flight analysis.

 

[kjhambrick]

Very nice @Spacedog49Krell !

That would have been the AltAcc 2C -- 2000 was about the time I had to give up rocketry for a new job in L.A.

I would love to see your base drag data !

Please post it if you can share it ?

Thanks !

-- kjh

 

[Spacedog49Krell]

First, I agree with everyone else that the Total Impulse of those two motors are close enough to be within specs..

If the 'spike' was caused by "a small chunk of unburned propellant passing through the nozzle", my concern is that it may cause a "Vectored" thrust. I and others have seen enough flights where the rocket left the rail straight then at a distance above the rail the rocket made a fairly sharp turn (maybe only a few degrees but sometimes much more).

Also have been plenty of posts asking why the rocket made a sharp turn. This may be the reason.

I agree with you Walt.

I too have received several requests for analysis of rocket flights that suddenly veered off course after launch. One was definitely the fly-away-rail-guide striking a fin, clear video evidence. A second rocket, using the same model fly-away-rail-guide, was a probable fin strike. The others were unknown causes. One sent on-board flight computer data, but it did not record any unusual accelerometer events.

A minimum sampling rate to capture, in detail, a thrust vector event would be 200Hz. My graph data was at 400Hz and there were no unusual lateral G loads recorded or flight changes in the video. Until more of us are flying high speed flight computers, >250Hz, with gyro data we may never know or resolve some of these mystery issues.

 

[Spacedog49Krell]

Very nice @Spacedog49Krell !

That would have been the AltAcc 2C -- 2000 was about the time I had to give up rocketry for a new job in L.A.

I would love to see your base drag data !

Please post it if you can share it ?

Thanks !

-- kjh

I only have a rough layout of the sensor cluster. There is a limited gap between the 29mm motor mount and the BT-60 body tube for sensors and electronics. This project started three months ago, but was interrupted with grandkids for the summer.

 

[kjhambrick]

I only have a rough layout of the sensor cluster. There is a limited gap between the 29mm motor mount and the BT-60 body tube for sensors and electronics. This project started three months ago, but was interrupted with grandkids for the summer.

No worries @Spacedog49Krell

I certainly understand having the grandkids and being able to spend time with them !

I would love to see what you come up with when it's ready to share.

-- kjh

 

[Buckeye]

A quick way of integrating and comparing the outcomes is to print the graph twice and cut out each graph and the zero line. If you have accurate scales you just weigh the cutouts. Integration is just the area under the graph, which is proportional to the density of the paper. That's how chemical analysis using spectrometers used to be done before the advent of digital computer interfaces.

Very clever and an interesting bit of history!

Though, a trapezoid rule or Simpsons rule written on the data would be the easiest way to integrate.

 

[OverTheTop]

Very clever and an interesting bit of history!

Though, a trapezoid rule or Simpsons rule written on the data would be the easiest way to integrate.

Assuming the person has the relevant programming skills, yes. Otherwise they just need a scale and to be able to operate scissors .

 

[SolarYellow]

The most glaring issue is that PROPULSE assumes drag -vs- velocity during the burn phase is the same as it is during the coast phase. It is not.

Another less important thing I never bothered with was the rate of consumption of the delay grain and the effect of the hot smoke from the delay on drag during the coast phase. But I figured it was insignificant with large enough rockets and motors.

Another interesting thing to think about is that that difference will likely be variable in context of anything that affects base drag. MD or rockets with an efficient boat tail might have proportionally less base drag under thrust, in addition to less base drag when coasting.

 

[MJ D]

Air in propellant should be treated as an additional particle type in the mix (air). But how do you adequately control for the percentage of that particle and its morphology? So of course if there are bubbles present the consistency is going to be lower than if they are not present. More bubbles and/or larger bubbles -> faster burn rate at a given pressure if the propellant has a classic burn rate curve. Bubbles -> lower, and probably inconsistent, density. IMHO any such propellant belongs on a burn pile.

Any decent propellant is vacuum processed, in commercial circles ambient mixing is unheard of and air bubbles are a great way to sensitize propellant.

I have a couple of CTI F120 29-1G reloads that state "use within one year of date". Only motors I have seen this on.

Vmax propellant is very storage stable and retains it's mechanical and ballistic properties for years. It uses no migratory burn rate catalysts such as ferrocene derivatives. It handily passed a wide range of tests to qualify it for various commercial applications. The only reason to put a "use before" date on such a product is to satisfy a rule or regulation that requires such. I would not hesitate to light one that is 20 years old.