A few days ago, AIU and I decided to compare the NECO load modeling program, with measured results from PressureTrace and a chronograph. I sent him my measured results from a set of 8x57 loads, and he ran them in NECO. These are the results.

gage PSI.. model PSI.. chrono fps.. model fps
>53790...........56421...........2755............2816
>50845...........53371...........2682............2768
>47840...........50487...........2609............2721
>44866...........47756...........2536............2673
>41891...........45171...........2464............2625

A friend with some experience in the field also offered this:

quote:

The Pentagon shut down the Frankfort Arsenal's efforts to use algorithms for forecasting pressures because as one contract researcher said, "We had at best, reasonable results for Milspec .223, 30-06 and 308 using Milspec IMR Powders."

With mid-sized cases, normal chambers and standard powders QL is often quite close, as were the old Frankfort Arsenal and Homer Powley
methods. Unfortunately there is more variation in civilian components and barrels. Not to mention differences in hand loading techniques. The further you deviate from 308/30-06 case proportions, military loads and production barrels; the more the data will deviate from
actual measurement.

Denton, The results, both PSI and velocity, appear to be with in 1% of the QL projection which is quite close, surely enought to keep one out of trouble.
It`s interesting the case size used will cause problems with the data. I would have thought the type of case, bottle neck vs straight wall, might cause problems but never the capasity.

Denton,
this confirms (externally) what I've said for awhile... QL (neco) over estimates the pressure, and thereby the vel, of a load. BUT the "RSI" matches the pressure of a FPS.

What I am saying is, if you look at 2755FPS from the chrono... which reported a 53790 psi... which, if you look at the QL PSI for about 2755 (2768) you find the pressure to be 53371 psi.

So, here's my point.
while quickload DOES tend to project <5% higher pressure and velocity for X load.. it reports HIGHER, which means, if you go by that reported pressure, your actuall loads have lowe prerssure.

Beats the heck out of reporting pressure too low.

jeffe

Now, wait just a cotton-picking minute! Quickload is not a static model. Until you take into account the variations in powder lots, bullet hardness, primer brisance and flame duration, and the individual firearm, it is nothing better than a "average of averages".

You are comparing a direct measurement method (unbenchmarked!) against an mathematical model. Interesting, but truly comparing apples to oranges. How is a comparison of a prediction tool to a measurement tool feasible?

If you want to do a proper evaluation of Quickload, calibrate the model first (i.e. adjust the settings to where the observed speed and pressure match the predicted pressure), and THEN change the components (increase or decrease powder, different powder, different bullet weight, etc).

All of a sudden Quickload will predict rather closely to the actual observations. As a matter of fact, it will PREDICT a lot closer than PressureTrace will PREDICT Dutch.

Dutch,
one question...

do you calibrate, from reference blocks, EVERY TIME you use your calipers?

Or, one statement... *I* think, for *MY* money, that Quickload and RSI are a FAR better pressure reading tool than reading a primer and good luck

jeffe

Jeff,

Huh?

Dutch.

i think everyone is in agreement. absolute accuracy vs relative accuracy. 5% absolute is quite good and my feeling for sometime is Mr. Broemel put in a little safety margin for all the obvious reasons.

i use chrony and adjust burning parameters in QL to fit the velocity and use that pressure. Only 2 powders so far, XMP5744 fit well and XMR4350 has not fit as well (could not adjust parameters to fit the velocity) same rifle, different bullets.

rgds,

steve

PS link to another datapoint
pressure trace Lapua

Man, my comp blows... anyhow, in the whole schmeel that I had typed a minute eariler to only disapper into cyberspace, NOT to get posted...

There was a discussion a few weeks earlier(more like trolling for trolls) about how to or how not to calibrate the pressuretrace to a "known".

Ultimately, they agreed to disagree on how a system that cannot be calibrated to a "known" can come up with consistent numbers as long as you do not fudge the numbers too bad. Sone folks cannot accept the fact that the thing works without a calibration from a factory round, or whatever. If my synopsis was wrong, please correct me, no need to mislead folks...

Denton,
Quickload provides a good starting point for a load, much like a load manual, and it can go one step better by taking into account the exact bore and groove dimensions, case capacity, and seating depth. Most people just use the default Quickload settings for each cartridge, which are conservative, which leads to higher-than-actual pressure and velocity predictions. Your 8 x 57 predicitons would have been more accurate if Quickload had been set up for your particular gun. If it is a milsurp rifle, chances are it has generous dimensions.

I have been pretty happy with Quickload but do notice some quirks. For example, it is hyper sensitive to seating depth, which makes sense when loading density is near 100%, but doesn't make sense and doesn't agree with actual results at lesser loading densities.

Quickload is also overly sensitive to what it calls the "start pressure", which depends on how far the bullet is seated off the rifling and the type of bullet (and the type of rifling and the type of throat and so on). As demonstrated in the May issue of Rifle magazine, the effects of the throat length and seating depth may actually be quite small.

Quickload does not take into account the bullet friction and how that affects the powder burn, which can actually be quite important, especially with cast rifle bullets.

If you are using a store-bought bullet in a popular cartridge in a rifle that has normal dimensions, then you don't need Quickload. On the other hand, if you like to shoot custom bullets, wildcats, or cast bullets, or if you are one of those geniuses who thinks loadbook data is too wimpy, then you can get your money's worth from Quickload. Just remember it is not the final answer.

If you don't calibrate your calipers to a KNOWN size, and then measure over the range of the calipers, then how can you KNOW it's accurate?

the answer? you DON't.. you know it's consistent in WHAT it measures, but do you know that it's accurate to .00001? how about .0001? or even .001? ... sure you can read what the dial says to .001 and estimate to the half of 1/10000.

Apply the same concept to the RSI... does it ACTUALLY measure 55000? is that actually 56250 or 53750 psi? in my case, as I paid *MY* money for it, is who cares...

the "resolution" of anything else.. primer reading, casehead expansion, etc, isn't even within 10% consistanly

Sure, you could spend several tens of thousands for a better machine... but, for a home user, that's called PII.... pursuit of irrelevant increments..

like wanting 11FPS more from a stout load, or wanting a hunting rifle to have .05 smaller groups, just because you want a "true" 1/2" rifle

jeffe

Jeff,

Why do you keep talking about calipers?

Dutch.

quote:

Originally posted by Dutch:
Jeff,

Why do you keep talking about calipers?

Dutch.

ah...

guess i didn't make that logical bridge....

a caliper is a precesion measuring instrument.
a GAUGE for all practical purposes
an RSI is a precesion measuring instrument
a GAUGE for all practical purposes

You are talking about the "requirement" for calibration of a gauge.... selectively.

I am asking if you calibrate your caliper (gauge) every time before you use it, as you are asking if the RSI is calibrated every time before it can be used.

Just trying to level set

jeffe

Based on the lot-to-lot variation I have seen in powders, I suspect that no QL type program is ever going to be very accurate, unless powder manufacturers start supplying some kind of "activity index" with their powders, and I think that is slightly less likely than a Hindu being elected Pope.

Dear Posters:

You can look at this data and say “the gas tank†is either half empty or half full. I choose to look at the data and say the “gas tank†is half full. I think the agreement is remarkable.

Denton, showed the data from QL when the program was used to predict velocity and pressure solely from total powder charge. He didn't show you the ability of QL to predict pressure from the velocity – an actual measurement. The latter is how I use the program, since I chronograph all my loads.

Shown below are the data Denton showed. The (#%) is percent QL difference from Denton’s measurements (by the way how do we know which measurements are correct – we don’t know).

gage PSI……. model PSI..………chrono fps… model fps
>53790...........56421 (4.9%)...........2755............2816 (2.2%)
>50845...........53371 (5.0%)...........2682............2768 (3.2%)
>47840...........50487 (5.5%)...........2609............2721 (4.3%)
>44866...........47756 (6.4%)...........2536............2673 (5.4%)
>41891...........45171 (7.8%)...........2464............2625 (6.5%)

Correlations of measured MV with NECO predicted PSI & NECO predicted Vgt grs for that velocity.

grs...gage PSI...….MV.....…….NECO PSI……..NECO grs
53....53790.4.……2755...……52489 (2.4%)...……51.7
52....50815.6.……2682...……48290 (5.0%)...……50.2
51....47840.8.……2609...……44423 (7.1%)...……48.7
50....44866...…….2536...……40858 (8.9%)...……47.2
49....41891.2.……2464...……37570 (10.3%)...… 45.7


I use the program to predict PSI from measured velocity given caliber, bullet, barrel length, and powder type.

Also, I use it to find the "best" performing powders for a given bullet, caliber, barrel length, up to 110% max fill, and max average PSI. I did this for the 8x57mm Mauser using the 175 gr. Sierra FB spitzer in a 24 inch barrel with max average set at 55,000 and up to 110% max. case fill. The following top 6 powders with expected performance came up out of the 100s available. Some will be compressed charges.

Somchem 5365 ~61 grs. with 2919 fps.
Ramshot hunter ~62 grs. with 2873 fps.
Acc. XMR 4350 ~59 grs. with 2873 fps.
V550 ~59 grs. with 2868 fps.
W760 ~60 grs. with 2866 fps.
IMR4831 ~61 grs. with 2860 fps.

Also, below is the analysis that Denton did with the data and Denton's comments:

“First, thank you very kindly for the analysis of more optimum powders. That alone probably justifies purchasing the software. Right now, finding the best powder is kind of a SWAG.

Here are the results of the statistical analysis. You can post them if you like, or we can just enjoy the outcome...

Five loads for the 8x57 Mauser were evaluating using the NECO software, and were compared with a prediction model based on strain gage measurements. To generate the strain gage model, 10 cartridges were fired, and the data were analyzed to produce a straight line models of charge vs. peak PSI, and charge vs. muzzle speed. This will produce less statistical noise than comparing with the raw data.

For peak PSI:

Gage PSI = -5730.1 + 1.058 x NECO predicted PSI.

R^2 = 99.89%.

The standard deviation of the residuals (error) is about 150 PSI, which is remarkably good.

In order to obtain these results, both systems must be highly repeatable. If you have repeatability, all other measurement problems are solvable.

The systematic difference between the two systems might come from any of several sources, including a batch of powder with characteristics different from average, a military chamber (Yugo 48) looser than a test barrel, or systematic errors in the hoop strain equation or properties of steel used in the PressureTrace calculations.

For feet/sec muzzle speed:

Chronographed fps = -15429 + 1.526 x NECO fps.

R^2 is so close to 100%, that the software package reports it as 100%.

The standard deviation of the residuals ia about .7 fps.

Again, the correlation indicates remarkable repeatability from both systems. The major errors are all systematic.

There is some reason to believe that the chronograph may be more accurate than the NECO model. If so, you'd still have to say that the NECO model produces very useful estimates of muzzle speed, particularly at higher speeds, where most rifles operate.


Prove all things; hold fast to that which is good.â€


Warm regards and happy shooting, AIU

Well, AIU, I guess I'm more of a pessimist than you are! (And that's fine.)

The good news out of the calculations is that practically all of the errors are "systematic" (can be corrected) rather than "random" (cannot be corrected). So, in theory, it is possible to eventually make a good mathematical model.

The bad news is that lot to lot variation in powder is significant, and manufacturers don't and won't supply an "activity index". Until you can get something like that, the best model in the world will be very limited. Add in chamber differences, bore differences, seating differences, etc., and it gets nothing but worse. You can build a good model for any one lot of powder, for any one firearm, but you can't generalize.

Is NECO a help? Sure. Especially since AIU pointed out that it can dramatically cut down on the Easter-egging involved in finding an optimum powder for a given cartridge.

I think the other problem is that case geometry has to be a factor, and that is not accounted for in a simple water capacity model. If you have a short, fat cartridge, the primer may be able to ignite a line of powder down the center of the cartridge, reaching almost to the bullet. In that case, combustion spreads from the centerline out. If you have a long skinny cartridge, the combustion more nearly burns from the back, forward. Same capacity, two different results.

Deton, Quickload is not quite that simple. There are things it does not allow for, that it should, such as primer vigor, etc.

I just spent a little time with three cartridges in Quickload, and my loading records. The 223WSSM, the 7mag and the 358 Win.

By manipulating the start pressure values (to account for the HV bullets moly coating and their low engraving force), and by manipulating the case shape factor (which is used to determine the amount of unburnt powder and gas that follows the bullet into the barrel), I was able to bootstrap Quickload to come within 50 fps of the observed velocities.

Next, I used those software settings, and weight charges of different powders I have in my records to predict the velocities of those charge weights. Presto, all predictions were within 75 fps of observed velocities.

Jeffe, you either have me confused with someone else, or you are not reading, or both. Not anywhere in the last 6 or so years have I ever suggested to use a caliper for anything other than measuring case length or cartridge length.

That said, any complex system should be benchmarked to make sure it works as designed. In other words, the Pressure trace DESIGN doesn't need calibration. Put it on a rifle, and you introduce a human factor that simply must be controlled for. Sloppy glue jobs, connections, measurements, etc, etc. No one in his right mind would hang their well-being on a theoretical calibration.

Firing a couple of factory rounds (or handloads) with an approximate known pressure value to benchmark the system "as used" is nothing to get one's panties in a wad over, and may just keep some of Mr. Murphy's ghouls appeased. JMO, Dutch.

quote:

Firing a couple of factory rounds (or handloads) with an approximate known pressure value to benchmark the system "as used" is nothing to get one's panties in a wad over, and may just keep some of Mr. Murphy's ghouls appeased.

Dang good idea.

Denton/AIU, any chance you could get exact chamber/bore data into that simulation and see how much of the difference is 'corrected'?

It was interesting that the variance grew as the velocity/charge/pressure dropped...dunno what it means for sure (could take a guess, but why), but if it does that for multiple cartridges it could be interesting.

It is also worth noting, along the lines of 'which is right', that the more data we have (manuals, simulations, measurements) the better our confidence in our results and the safer our hobby becomes. These tools are cheaper than body parts for sure!

Direct quotes from Denton below.

quote:

The systematic difference between the two systems might come from any of several sources, including a batch of powder with characteristics different from average, a military chamber (Yugo 48) looser than a test barrel, or systematic errors in the hoop strain equation or properties of steel used in the PressureTrace calculations...Again, the correlation indicates remarkable repeatability from both systems. The major errors are all systematic...There is some reason to believe that the chronograph may be more accurate than the NECO model. If so, you'd still have to say that the NECO model produces very useful estimates of muzzle speed, particularly at higher speeds, where most rifles operate.

Denton, from you comments, I thought you were impressed with the NECO program...but, from this thread you seem to have become more negative. Hmmmmm.

Oh, I think the program is useful alright.

Before you use a tool like that, it's good to think about the places that it might be useful, and the places it might less useful.

For me, I think the main use could be doing a quick screening of possible powders, as you have suggested.

For shooting IMR powders in 30-06's, it sounds like you get a good model. And Dutch says he has tweaked his model to account for some of the finer issues, and gotten good results. That's good.

In theory, the errors are almost all correctable. That's also good.

We just have to recognize the limits of where we can apply what we have today, and use it right.

After a day at the range, I have to take back some of the nice things I said about Quickload.

I was shooting 50.3 gr. H4895 behind a 180 gr. cast in an '06, and also 49.0 gr. 3031, based on a Quickload prediction of 58 ksi.

Turned out, both loads were 73+ ksi according to Pressure Trace. Velocity was high, too, and one case came unglued.

I double checked my Quickload calculations and couldn't find an error. Quickload made very accurate predictions for these same cans of powder in another cartridge. I hate to admit it, but Quickload was very wrong in this instance, and wrong in an unsafe way.

I'll continue to use Quickload, but with more caution.

I don't think QL is any good for mini-bores
and cast bullets, unless bullets are treated
to be as hard as bullet jackets.And I have
bullets here that hard in 458 and they acted identical to the jacketed bullets of same weight.Are these the bullets you are getting
extra spike readings from?Ed.

I doubt I can add much here, but...

I recall reading the numbers used to describe the powder burn characteristics are derived from pressure barrel firings. That these barrels and chambers are cut to minimum dimensions may explain why QL tends to overpredict P and fps for a given charge. Perhaps the shot start pressure needs to be backed off even more for typical sporter barrels?

It was disappointing to see QL's prediction of P for a given fps fall off so much at the lower pressures. While QL's author warns that the modeling of the burn at lower P is known to be off, the results seem to match fairly well with the reduced loads available in some load books. Indeed, QL seems to give fair predictions for fps and P for cartridges like the SAAMI 8x57 and the 35 Rem. (Do keep in mind, QL's author recommends a 10-15% fudge factor on the P, probably to accomodate the random shot to shot variations in P.)

Anyway, thanks to denton and AIU for this information. I hope you fellows have time to try it with some other cartridges -- both high and low pressure -- as well.

quote:

Originally posted by hubel458:
I don't think QL is any good for mini-bores
and cast bullets, unless bullets are treated
to be as hard as bullet jackets.

There is no such thing as a cast lead bullet that is as hard as a gilding metal jacket, but the bullets in question were heat treated WW, 28 - 32 BHN.

QL has been amazingly accurate for cast bullets in my revolvers.

i was interested in the shape curve of Quickload and made this comparison (PT data supplied by a member in alaska). since 300/338, first pass thru i just used 308 case with increased capacity and the fit was not as good. Then model the exact cartridge dim. and by only adjusting the start pressure to match the data, this was the result.



The data set integrated out a velocity of 3431fps and Quickload 3238fps. And you can see why the velocity was low due to the shorter tail in quickload.

Anyhow, i am a fan of QL and given all the programs out there, quickload at least has complex burning model and most of the time, good match. And therein lies the problem for some people. If not 100%, then how does one know that a load is not poorly modelled. So why bother.

Well, for most of us, we gain confidence with a particlar load. And once achieved, then QL becomes very useful. So for the guy that was using the 300/338 Lapua improved, now that a match has been made, can use quickload to extrapolate that load to other bullets (yes, the numbers might be a bit off but when no reload data is availabe for a wildcat, not bad starting point).

I have no experience in internal ballastics and am learning a lot from this forum, but i wonder why it is that we can model complex structures such as a stealth jet, atomic fission, sun dynamics, even chaotic weather systems but for some reason, not accurate internal ballistics?
Is the answer that just not worth the effort and easier to just build and measure?

rgds

steve

Steve, I think the answer is that the important variables aren't well controlled by most users. Bore size, groove depth, land width, throat length, primer vigor, bore resistance, bullet engraving force, throat erosion, etc.

However, I agree with your premise. Once the parameters in the model are set correctly for a particular firearm, future predictions are rather robust. Certainly better than the PREDICTIONS made by a chronograph or a strain gauge!

Incidentally, most people here take the PSI readings from the strain gauge and the velocity readings from a Chrony as hard truths. There is at least consistent anecdotal evidence that the Chrony readings tend to be high. Usually between 75 and 100 fps faster than Oehler. Which is right? Chrony, Oehler (or Quickload...). Only a ballistic pendulum could tell us, for sure. JMO, Dutch.

quote:

just not worth the effort and easier to just build and measure?

Yes, to a point. The other issue is the existance of an accurate model that takes into account EVERY significant factor, and the computing power required. Take your weather system example, the computer power to make that model run is not what the normal person has on their desk at home (or at work, for that metter). Besides, how accurate is the weather prediction you see every evening on the news...out say 3-5 days?

The final question is one we are well familiar with...limited returns. How much better is the model if we double the amount of work (and $$$)going into it? 5%? 1%? Is it worth it? Remember, thiese are not governments that can just work to a predetermined accuracy point, this is a company that has to sell a product and make money to stay in business. Not easy given the relatively limited range of customers!

I think you've nailed the central issue: information per unit cost. The idea is to know what grade of information you're buying, and to use it appropriately.

I want to get a Pressure Trace,BUT if it is too
accurate making those funny traces being
shown and talked about by some, do I want it and MORE important,do we and
do I want to know that much.Also will it sell,
giving that much and kind of info???If that
info is true, of course we need them so that we can correct loads that show up like that,
or make changes to guns and components to
make corrections....Ed.

steve505, I'm not sure that a P-t trace can reliably give one fps. To do so, you must make an assumption about drag between bullet and bore, just as QL does. So which assumption, if either, is correct? Does the PressureTrace ask you to enter a chrono reading to try to adjust its calculations?

As for ever perfecting modeling, a P-t trace in the old VihtaVouri 2nd (or was it 3rd?) reloading manual convinced me it was rather hopeless. I'm sure the guys with the PressureTrace have seen this as well. At ignition, the pressure can vary widely and rapidly. This should vary with the primer's "spark" characteristics. I have no clue how one could even begin to model this behavior.

Then there is the issue of modeling the powder burn. My understanding is that the tests which determine this are fairly insensitive at low pressure. The problem is that the burn itself is very sensitive to pressure. The burn rate is so relatively slow at low P that it is difficult to discern changes in it. Take this inability to accurately sense the initial burn rate, combine it with the primer effects, and then throw in the vagaries of bullet engraving, and it seems hopeless to me.

Dutch: are those cheap chronos really that far off. I'll never look at my Competition Electronics job the same again. (sniffle)

ASDF, we don't know WHICH ONE of them is off. All we can do is line them up and say "geez, this one reads higher than that one".

Which one is right? No one without a calibrated unit knows. Even though we know all the cycle rates, etc, etc. in the electronics and we know the readings are according to the theory, we just don't KNOW. All we know is that it is "pretty close". FWIW, Dutch.

quote:

I think you've nailed the central issue: information per unit cost. The idea is to know what grade of information you're buying, and to use it appropriately.

Denton, this nicely sums up the problem. determining the grade of information is the tricky part.

steve

quote:

are those cheap chronos really that far off

I sure hear about guys lining up 2-3 different chrono's and getting results within 5fps on them all a lot. I have yet to hear about someone getting 100+ fps difference in such an experiment. It's probably happened, but I haven't seen or heard about it.

There is a confidence factor (denton could explain the statistics better I think) involved with measurements. One cheap way to GREATLY increase the confidence in your readings without a true calibration (*) is to place 2 otherwise identical measuring units in series to measure the same value. The closeness of the measurements shows accuracy...when the measurements diverge, one is off, but when they are both nuts on, you can have a high confidence that you have a valid, accurate reading. This in the one feature of the Oehler (with the third screen) that is really worth anything at all, IMHO.

(*) Note that even a professionally calibrated, NIST tracable instrument is not 100% reliable, as measuements can drift, and we all know that 'stuff' happens. Always use a bit of common sense with any measurement system.

CDH, you might find this discussion interesting:

http://www.securedsend.net/forums/showthread.php?t=21080&highlight=oehler+test+chrony

FWIW, Dutch.

Wow, a 24 ft screen spacing! However, another shooter there noted his Chrony gave the same fps as an Oehler tested against it. I suspect the big error was not so much due to the accuracy of the 24 ft screen spacing, as it was some mistake in setting up the Chrony. With only a foot or two between screens, errors due to illumination etc. become larger as a percentage. One of the reasons I chose the model I did was that it didn't fold. Still, the report should make one more cautious in handling the basic chrono's.

This is off topic some but:

Denton,

Not sure how a mathematical program can have "random" errors. Any errors are all "systematic."

Dutch,

I do not see how there can be a "known pressure" factory load for any given rifle by which a strain gage can be calibrated.

Strain gages can be used to give predictions of approximate pressure, but they still are only approximate.

asdf, I agree. One should be a little cautious with numbers. Just because the read-out displays something with 4 digit precision, doesn't mean it is necesarily so. I have a lot of confidence in technology, but not near as much confidence in the two legged component that puts it to use.... And it doesn't really matter which two legged component!

Will, I said benchmark, not calibrate. But, like WOW, a strain gauge that can PREDICT. I got to get me one of those. Dutch.

Dutch, interesting link. 24' screen spacing would definately reduce the error due to pulse counting...but the large number of shots should average out the inherent velocity random error. Still, it is a test on one unit, but not favorible for sure. The Oehler vs. Chrony vs. XYZ has been well hashed out many times, and when working (big qualifier there, I know) they are all good enough for 99% of us out there. Those benchrest guys sure can take things to extremes, can't they!?!?