Jump to content
ADI Forums

# Show me yours I'll show you mine... (27 CFR 19.356/600)

## Recommended Posts

I've attached our newly created spreadsheet for our 19.600 compliance (yes I know we're missing a few columns)... it's a work in progress. We have both a Mori 4-spout filler and the expressfill time-based filler. This spreadsheet is for the latter.  All in all, we're doing OK. Not sure what to think about the intra-spout differences. Spouts 2 and 3 are less underfilled than 1 and 4.   I'm going to add some calculated columns for the corresponding volume deltas at 30F and 90F (thanks @meerkat for making this so simple).

I have no damn clue how to address this:

Quote

There must be approximately the same number of overfills and underfills for each lot bottled

since this filler consistently underfills but well within limits.   I have a theory based on how these are plumbed.  The last stats class I had was in the fall of 1985 so forgive my errors. I look forward to your thoughts.

Expressfill Filler Audit.xlsx

#### Share this post

##### Share on other sites

It is fully adjustable, so it shouldn't consistently under-fill...

#### Share this post

##### Share on other sites

I’ve heard it said underfilling is better than overfilling.

#### Share this post

##### Share on other sites
Posted (edited)

I have added Column L to your spreadsheet and attached it here.  A value that has an equal number of occurrences above and below it is called the median (the midpoint of the frequency distribution in stats-speak).  Cell L2 initially contained my guess for the median.  I have set the values in Column L to be +1 if the actual net fill value in Column H is more than my guessed value for the median, or to -1 if the actual value is less than the guessed median.  The total of all the + and - ones in Column L should be zero if there are an equal number of aboves and belows.  Excel's Goal Seek function makes it easy to find the actual median using its internal trial and error functions.

The median turns out to be 701.6.  It is useful to compare this with the mean (the usual understanding of "average") which I have shown in Cell H27.  The fact that the mean (701.7) is so close to the median (701.6) is an indication that your data points are normally distributed (i.e. follow the typical bell curve) and this is a "good thing" (tm).

Because the median is 4.6 grams lower than your target fill of 706.2 and the median and mean are so similar,  I suspect that if you simply increase the target to 710.8 (=706.2+4.6) then you will get a mean value very close to 706.2 and your overs and unders will be roughly equal.

Edited by meerkat
typo

#### Share this post

##### Share on other sites

I notice that you have a target mass set to one value. Do you adjust that during the bottling run? We take the density before each test of the fill and the spreadsheet calculates the target mass for that read.

I've never figured out how much a change in proof during the bottling run will impact the target weight. Not even sure it's worthwhile but we get the density when we take the proof anyway. So I just added it into the spreadsheet as well.

## Create an account or sign in to comment

You need to be a member in order to leave a comment

## Create an account

Sign up for a new account in our community. It's easy!

Register a new account

## Sign in

Already have an account? Sign in here.

Sign In Now

×

• Calendar