Notesoft

How it works

______This can be divided into two groups: the stuff you see, and the stuff you don't see. Except for a small comment on the math, we'll focus on the stuff you can see.
______One of the first things you'll notice is that entering the data is very similar to entering the data on a paper QA form. But it's easier. Your tech doesn't have to do the calculations, eliminating one major source of mistakes. Also, the electronic form is more secure because you can backup thousands on a single CD, and store it off site. Fire at the office? You still have your data; ten thousand test results on a ten cent piece of plastic. If you take advantage of the backup features, your data will survive things that could destroy your building.
______We are not yet decided on this, but are considering whether to offer on-line backup options, so you don't even have to burn the occasional CD. If we do this, it will be added as a one-click feature in the program.
______Your technician starts using AgMon the moment the first test result is available. When he finishes with one aggregate type, the statistics can be run, emailed to the production personnel who control your production process so they can see, very quickly, what the current trends are and make tweaks to keep the production as near to the center of the specified gradation as practical. Of course, nearer to the center means that you stand a lower risk of product hitting the stockpile out of spec. If your QA technician notes an alarming abrupt change, the email can and should be backed up by a telephone call.
______In addition to the usual mean-based analysis of sample results, AgMon-8 allows you to click once to see trends, visually, on a sieve by sieve basis; and to click again to see an estimate of the composition of your "well-rolled" pile. (The first was also done by AgMon versions 1.0 through 1.10 and is well established.)

A Little Math Stuff

______The trend analysis performed by AgMon is adapted from standard techniques. (If you want to read more about it, go to the National Institute of Standards and Technology web site, math section, for a thorough discussion.) The mean line is determined by a conventional "best fit" approach. The Standard Deviation about the mean line is estimated to equal the SD about the ordinary mean. When the slope is small - as in a well controlled process - the error is trivial and conservative. When the slope is large, the error increases somewhat, but it doesn't matter because your process is clearly out of control.
______This analysis estimates the probability that your next sample, for each screen, will be "in spec." Then, making the conservative assumption of statistical independence, the program computes the probability that the next sample will be in spec for the entire gradation.
______Estimates made by statistical methods are just that, estimates. Estimates made by AgMon are slightly conservative within the range where calculations are useful. If your goal is 96% across the board gradation spec compliance, and AgMon reports that you are achieving this level, you are actually doing slightly better than indicated. If your process is poorly controlled, AgMon will return estimates that are progressively more conservative. That is, if you are actually running at 80% compliance, AgMon may report 70%. If you are running at 50% compliance, AgMon may report 30%. Considering how AgMon is used illustrates the insignificance of this sort of error. When your process is in good control, the more accurate estimates allow you to fine tune, and to exercise control proactively. When your process is badly out of control, your interest is in the direction you must shift the results. Fine tuning is not very important when the house is on fire.