Experimental Design

[memilanuk]

Anyone here familiar with experimental design aka design of experiments (DOE)? Have you ever used factorial or fractional factorial experiments to optimize a load? What about response surface modeling?

Been reading and thinking about this stuff for a while; just never seem to get around to really putting it into practice. I know of one or two people that have in the past. The one example made for an interesting read; talked to the other person for a good while on the phone but never got to the point of implementation.

Monte

 

[25AI260]

factorial experiment may be done if you had polynomial model to start which is still only an approximation attain as a specific target can be used to optimize the combinations in a full factorial design would not be logistically feasible

just my two cents

Sheldon Cooper

 

[Warren Dean]

Anyone here familiar with experimental design aka design of experiments (DOE)? Have you ever used factorial or fractional factorial experiments to optimize a load? What about response surface modeling?

Been reading and thinking about this stuff for a while; just never seem to get around to really putting it into practice. I know of one or two people that have in the past. The one example made for an interesting read; talked to the other person for a good while on the phone but never got to the point of implementation.

Monte

You must be really bored, big guy. ;D

 

[R.Morehouse]

Cabin Fever..........I Got It To....

 

[memilanuk]

Sheldon, thanks for the reply.

At this point I'm mainly curious... there are a few variables that seem like there might be interactions present. I agree that a full-blown factorial design of every variable out there could get out of hand very quickly, with limited value.

I was thinking more in terms of a simple screening design to weed out some things and see if there are indeed interactions between others, and then a possible follow-up RSM test to explore those that do matter.


The logistics of randomization, replication, blocking, etc. is where I start getting vapor-locked...

Monte

 

[acloco]

6 sigma

 

[jpretle]

An orange is what color???, again.?

 

[Alexander-M]

Monte,

Many moons ago (1975), while as QC Engineer at General Electric Wire Mill, I conducted an experiment using Scheffe's Test to evaluate how several parameters affected enameled (magnet) wire. I can't remember now how many parameters there were, but at least annealing temperature of the bare wire, two of the annealing atmosphere characteristics, speed of the wire, enamel temperature, and the type of enameling dies used, were closely monitored and measured or classified. I wrote two programs back then, one in Basic and the other in Fortran for the different computers we had available. The end result was that we were able to determine how significant each parameter was to the quality of the finished magnet wire.

I can see doing a similar experiment including slight variations in powder weights, jump distances, base to ogive dimensions, bullet weight, etc., and shooting in a calm day to determine which of those parameters or factors has more significance in the group size obtained. Maybe include shooting at different distances as well?

Only problem is that I would spend much more time conducting the experiment and doing the analysis than I think I would care to do.

Just my 2¢

Nando

 

[GerryM]

some people are hard to understand.......

Its like trying to understand the color 9.

 

[memilanuk]

Nando,

The parameters you mentions are perhaps a bit more subtle than I'd originally envisioned looking at; still, they sound like good material for a later experiment.

No doubt about it, this will take a bit more time/effort than my usual testing method. I'm okay with that, partly because like some mentioned, it is a slow time of year around here, and partly because it just seems to make sense that there might be a little room for improvement that isn't necessarily being captured by standard OFAT testing. Maybe it is and it will all be just a good excuse to go to the range. Oh, darn

Any thoughts on how to tackle things like replication and blocking? Not sure if there will be enough 'power' due to the factorial nature of the designed experiment, or if additional replicates would be needed. Also, I'm going back and forth on how to handle blocking - some things, like wind, ambient light, temperature, humidity, etc. I can't control. Also there may be some constraints such as barrel temperature, fouling, etc. that may require a break in the testing process for cool-down, cleaning. Plowing onward through changing ambient conditions and with an ever increasingly hot barrel may not be the most objective plan. At the end of the day, I do want a 'robust' load that will not be unduly affected by most of the above - I just don't want it to end up confounding things during the actual testing.


Monte

 

[CharlieNC]

DOE are efficient to cover the space for combinations of variables especially when using other than full factorial designs; many good options available. Analysis for responses such as velocity, group size, etc is straight forward. However I do not know of a good option to mathmatically determine nodes, which would be best done graphically. PM me if you want details.

 

[25AI260]

i find in just one frame [ X's ? ] the parameters will change greatly in each attempt to clairify a constance
using just three of the elements thats mentioned by Nando cannot be controled in a short period .. my thoughts were not logistically feasible originally mentioned the variables in itself would take a life time an still not be conclusive !!!!!

i was hoping one with experience in Scientific Engineering [ maybe lurking] lenerd hoffstedda comes to mind hoping he'd chime in ?

Sheldon C

 

[clams243]

For optimalization performance do your loading in a bariatric chamber during electrical storms

 

[expiper]

OP,,,when I want to optimise a load ,,I start touching the lands and work my way out after finding the max powder charge,,,,guns are like women,,,they work way different in the field than they do in the lab,,,Roger

 

[OIF/OEF]

Don't forget the next MENSA meeting gents...

 

[25AI260]

haven't been to England for the IQ approved intelligence tests, but maybe one day.. is Chicago, closer an around Halloween they feature a costume party for which many members create pun-based costumes an the annual Mensa Mind Games competition.. Oh Yea !

whaddayathink
Sheldon C

 

[Joe Wade]

Monte

John Stranahan has a good book "Precision Rifle Rebarreling and Case Preperation on a Home Shop Lathe"
He has a chapter dedicated to "load development, factorial design, accuracy loads. He charts bullet jump versus powder charge to plot best accuracy. Once you understand the principals, you can use it for other reloading components and variables. I use it to develop my loads. I think it saves time and the number of rounds fired to find the best load.

Here is a link to an article on his book.
http://forum.accurateshooter.com/index.php?topic=1748330.0

Hi is a forum member. Just search his name and send a PM
or e-mail to: JohnStranahan@aol.com

Joe

 

[OIF/OEF]

No kidding you guys are extremely intelligent I love to learn but I know when to tap out also lol. Me head hurts must quit trying to figure out topic discussion

 

[clams243]

The winter has been toooo long

 

[memilanuk]

Yes, apparently it has. :-X