Hello,
I am interested in learning more about the Art vs Science of reloading. It seems that most people who reload "tinker" a little bit looking for the perfect combination that matches their rifle.
I'm wondering if anyone has used response surface full factorial modeling to identify the "perfect" bullet combinations for the most desirable results?
For those of you who don't know what RSM modeling is here is a breif description. Basically, you define the control input factor ranges like bullet grains, powder grains, seating depth, shooting distance to target, head space...basically anything you want that you can "control"
Then you identify the "names" of measurable responses that are make up the desirable results like precision (ie averaged shot group size), accuracy ( distance from the bulls eye at the defined shooting distance), muzzel or chronograph velocity, etc.
From there you generate a series of test of your input factors and fire your rifle, measure and record your results. From there you put the data into a software program that generates a 3D model and prediction profiler.
Think of it like a cube with a topographical image of a mountain that you can look at from all sides. You can then have the software tell you what your optimum load combinations are and how big of range you have before performance deminishes at various combinations. It will also predict how your rifle will perform at virtually any combination within the cube size, and how All of the reponses combinations will perform. So If you have a 175 grain bullet with x seating depth, Y grains of powder, Z distance to target you will know what your accuracy, group size and muzzel velocity will be for that load combination befor you even shoot the round or even build it at your reloading bench.
It is extreamly important to know that garbage in equals garbage out. So you have to be an obsessive compulsive nut like myself to generate a good model. Everything must be measured and matched accuratly while you go through the process of building the loads and recording the results.
However, the beauty of it is this. You generate a model that maps out all of your parameters with minimum shots fired. So a basic model can be as low as lets say 20 shots but the model will be equillivant to 3,340 shots or something like that. The more levels you have...obviously the more shots you have to make. So a huge model in triplicate shots would be something like 300.
This is something I'm going to do with my FNH SPR 300 WSM and my .308 once I can afford a progressive reloader.
If one or two of you would like to try the modeling method let me know. I would be willing work with you to generate the model if you are willing to load and measure the results with extream precision. (garbage in will be garbage out I promise you that) I cannot do any more than one or two because I don't want this to be "work". I am just "curious" how well it will match up and thought it might be interesting and useful for some of you as well.
Once the model is built I can send you the results and model in pdf format with the "ideal" input factor settings. Then you will need to go out and shoot at the recommended setting to prove out the model and report back how well your system performed.
I am interested in learning more about the Art vs Science of reloading. It seems that most people who reload "tinker" a little bit looking for the perfect combination that matches their rifle.
I'm wondering if anyone has used response surface full factorial modeling to identify the "perfect" bullet combinations for the most desirable results?
For those of you who don't know what RSM modeling is here is a breif description. Basically, you define the control input factor ranges like bullet grains, powder grains, seating depth, shooting distance to target, head space...basically anything you want that you can "control"
Then you identify the "names" of measurable responses that are make up the desirable results like precision (ie averaged shot group size), accuracy ( distance from the bulls eye at the defined shooting distance), muzzel or chronograph velocity, etc.
From there you generate a series of test of your input factors and fire your rifle, measure and record your results. From there you put the data into a software program that generates a 3D model and prediction profiler.
Think of it like a cube with a topographical image of a mountain that you can look at from all sides. You can then have the software tell you what your optimum load combinations are and how big of range you have before performance deminishes at various combinations. It will also predict how your rifle will perform at virtually any combination within the cube size, and how All of the reponses combinations will perform. So If you have a 175 grain bullet with x seating depth, Y grains of powder, Z distance to target you will know what your accuracy, group size and muzzel velocity will be for that load combination befor you even shoot the round or even build it at your reloading bench.
It is extreamly important to know that garbage in equals garbage out. So you have to be an obsessive compulsive nut like myself to generate a good model. Everything must be measured and matched accuratly while you go through the process of building the loads and recording the results.
However, the beauty of it is this. You generate a model that maps out all of your parameters with minimum shots fired. So a basic model can be as low as lets say 20 shots but the model will be equillivant to 3,340 shots or something like that. The more levels you have...obviously the more shots you have to make. So a huge model in triplicate shots would be something like 300.
This is something I'm going to do with my FNH SPR 300 WSM and my .308 once I can afford a progressive reloader.
If one or two of you would like to try the modeling method let me know. I would be willing work with you to generate the model if you are willing to load and measure the results with extream precision. (garbage in will be garbage out I promise you that) I cannot do any more than one or two because I don't want this to be "work". I am just "curious" how well it will match up and thought it might be interesting and useful for some of you as well.
Once the model is built I can send you the results and model in pdf format with the "ideal" input factor settings. Then you will need to go out and shoot at the recommended setting to prove out the model and report back how well your system performed.
