Principles of the average net weight system

Hi everyone,

we were trying to sale our products to Canada. I need to ensure if our products net weight is compliance with SFCR and have a procedure in place for our process. I have been trying to get my head around the weights and measures.

Principles:

Tolerances for net quantities declared in metric units of mass for catch weight products

more than 60 gr to not more than 600 gr - Tolerance is 6 grams

1. The declared quantity on a package should accurately reflect the quantity being supplied, so the average net contents of the packages in a lot may not be less than the declared quantity

2. The control over production should be such that the individual packages are within allowable tolerances. No more than 2.5% of the lot may have a negative error larger than the tolerance.

3. The number of packages which may have excessive negative errors is limited. Not more than one package may contain less than twice the permitted tolerance.

As an example I am thinking of a 500 g pack (brownie- bakery products) in a lot of 20000 packs. It is required the tolerance should be 6 gr. The tolerable negative error would be 2.5% of nominal weight. How can I calculate the average net quantity in a lot?

Do you not have a checkweigher on the line?  How are you verifying weights right now regardless of Canadian import requirements?

In order to satisfy CFIA and Weights and Measures Canada do the following:  this is what CFIA inspectors use for verification

The "lot" wouldn't be the whole production run, it would be by shipment. As CFIA can't/won't verify your entire production run, they will consider the contents of the shipment a "lot".

I hope this helps

And here is a link to the actual regulations where you may find more information

https://laws-lois.ju...1605/index.html

For the purposes of this inspection procedure, a "retail packaged product" is any product that has been packaged, weighed and labelled in a retail establishment prior to purchase by a consumer, and for sale exclusively in that establishment (such as meat, fish, poultry, fresh fruit and vegetables).

The number of catch-weight retail packaged products in a lot may be very few in some retail establishments. In such cases, different products with same tare value may be combined to represent a lot.

For catch-weight food packaged at retail:

• if there are less than 10 packages in the lot, select the total lot.
• if there are more than 10 but less than 300 packages in the lot, select a sample of 10 packages from 3 randomly selected commodities.
• if there are 300 or more packages in the lot, select a sample of 30 packages from 6 randomly selected commodities.

Note: a random sample could be chosen from a lot consisting of only 1 commodity when there has been a history of net quantity problems with 1 commodity.

• For each sample selected, separate the packages into groups of the same tare and the same price per unit of measure to enable packages to be weighed with limited adjustments for the tare.
• Measure each sample, ensuring the tare has been set on the scale.
• When the tare weight value used represents the value of only 1 tare determination or the value from a predetermined tare weight table, the actual tare should be confirmed for each package which is found to be defective in quantity.

 

Do you not have a checkweigher on the line?  How are you verifying weights right now regardless of Canadian import requirements?

 

In order to satisfy CFIA and Weights and Measures Canada do the following:  this is what CFIA inspectors use for verification

 

 

The "lot" wouldn't be the whole production run, it would be by shipment. As CFIA can't/won't verify your entire production run, they will consider the contents of the shipment a "lot".

 

I hope this helps

 

And here is a link to the actual regulations where you may find more information

https://laws-lois.ju...1605/index.html

Appendix 2: retail prepackaged products sold by weight (catch-weight food)

For the purposes of this inspection procedure, a "retail packaged product" is any product that has been packaged, weighed and labelled in a retail establishment prior to purchase by a consumer, and for sale exclusively in that establishment (such as meat, fish, poultry, fresh fruit and vegetables).

The number of catch-weight retail packaged products in a lot may be very few in some retail establishments. In such cases, different products with same tare value may be combined to represent a lot.

For catch-weight food packaged at retail:

• if there are less than 10 packages in the lot, select the total lot.
• if there are more than 10 but less than 300 packages in the lot, select a sample of 10 packages from 3 randomly selected commodities.
• if there are 300 or more packages in the lot, select a sample of 30 packages from 6 randomly selected commodities.

Note: a random sample could be chosen from a lot consisting of only 1 commodity when there has been a history of net quantity problems with 1 commodity.

Net content determination

• For each sample selected, separate the packages into groups of the same tare and the same price per unit of measure to enable packages to be weighed with limited adjustments for the tare.
• Measure each sample, ensuring the tare has been set on the scale.
• When the tare weight value used represents the value of only 1 tare determination or the value from a predetermined tare weight table, the actual tare should be confirmed for each package which is found to be defective in quantity.

According to the NIST Handbook 133 specifies that the average net quantity of contents in a lot must at least equal the net quantity declared on the label. we consider each day production run as a lot, and every hour take 10 samples. get the average weight at the end of the shift which should not be less than the net weight declared on the label.

I am really confused. can you give me an example of tolerance according to the net weight?

Thanks,

Tresa

I think you're over thinking this!  What your doing is more than sufficient

The 500 gram package CANNOT weigh less than 494g in order to comply with weights and measures.  So your average at each weight check that you are doing cannot be less than 494g, because by default, you'd have at least one package that weighed less than the allowable variance of 6 g.

501 + 499 + 499 + 502 + 494 + 495 + 503 + 497 + 499 + 500 = 4989 g/10 pks = 498.9 g

I think you're over thinking this!  What your doing is more than sufficient

 

The 500 gram package CANNOT weigh less than 494g in order to comply with weights and measures.  So your average at each weight check that you are doing cannot be less than 494g, because by default, you'd have at least one package that weighed less than the allowable variance of 6 g.

 

501 + 499 + 499 + 502 + 494 + 495 + 503 + 497 + 499 + 500 = 4989 g/10 pks = 498.9 g

The 500 gr package CANNOT weigh less than 494g. Do you mean each box or average in a lot? what I understand is the average lot shod be within the tolerance of 6 gr not each box. right?

right now for the 500gr box we have the 6gr tolerance in a lot. it means by calculating the average net weight of the samples, it should be within our 6 gr tolerance but for individual box net weight we get 490-488-510-492-491-

No, the average per PACKAGE

Your example above would NOT meet requirements as per below

Your average is skued by the one package that is WAY over weight 

A lot will not meet the requirements of the Consumer Packaging and Labelling Act and Regulations if in a sample the number of units containing less than the declared net quantity, by more than the prescribed tolerance, exceeds the number permitted by the Regulations.

Hi Tresa,

Weight control is mathematically, a quite complex statistical exercise.

The Canadian requirements look similar, albeit IMO less clearly spelled out, to the methodology used by EU to support their "standard" weight logo. the statistical consequences of the latter's setup have been highly analysed and detailed in published documents. Together with explicit lot sampling routines.

Regardless one, IMEX, customary operational solution (unless perhaps having a nice checkweigher option) is to simply overpack by an amount such that you can forget the mathematical intricacies and simply utilise the average package net weight/defects (should be very few) as determined/confirmed by sampling at appropriate intervals.