30 replies to this topic

[Charles.C]

Hi All,

 

Based on previous posts and various IT documents, eg USP Guidance on Food Fraud, I have put together an abbreviated attempt at a Guideline Vulnerability Assessment. One more for the growing stack.

 

 Vulnerability Asessment,3.0a.xls   19.5KB   1198 downloads

 

Comments welcome.

 

PS - for a few comments on the action interpretation of the final risk matrix, see last para this post and following -

 

 http://www.ifsqn.com...-v7/#entry95466

 

PPS - a 5x5 matrix version and  a Guidelines Table regarding  Likelihood Levels for the Excel O/D Factors are in posts 11, 12 respectively.

[teodymayojr]

this is excellent! 

[sue.c]

Hi. Looks pretty good but by applying an average you could be diluting the risks. A good theory though as otherwise can get so complicated and over the top. 

[Charles.C]

Hi. Looks pretty good but by applying an average you could be diluting the risks. A good theory though as otherwise can get so complicated and over the top. 

 

Hi Sue,

 

Thks for yr comment and Welcome to the Forum !

 

It’s a valid observation in certain situations. Alternative methods of summation are available which can reduce such an effect but involve some computational inconvenience. General practice seems to try and keep it simple as long as possible unless there is an obvious disconnect.

 

Some procedures include criteria for the individual components to avoid ignoring  significant “outliers”. I daresay most users would anyway be alert for such occurrences and act accordingly.

 

A few other textbook caveats also exist but these seem to be “smoothed over” on the grounds of pragmatism if useful results are obtained. The inherent subjective factors in these methodologies are always in the background too. Hopefully BRC will agree.

[Charles.C]

Hi All,

 

 

Just noticed that I forgot to “activate” the “averaging” cells in column L of Excel sheet in VERSION 2.0a in post 1. The activation is simple but seemed easier for new readers to update the original file. I also corrected a minor typo. in the text.

 

I have activated the averaging cells and and re-uploaded as ver.3.0a

 

Sorry for omission and please update accordingly (note - the results from ver.3.0a will be identical to version 2.0a + the activation step previously given in this post).

Edited by Charles.C, 12 November 2015 - 11:16 AM.
edited/updated

[boediprasetyo]

Hi Charles,

 

Just comment for your vulnerability assessment : really very good

 

I have tried to make another approach, but finally yours is more simple.

 

So, I send the attachement of yours with little beat additional and vulnerability assessment of mine (need advise too).

 

Once again thank you for be inspired me

 

Warmly regards

 

Boedi

Attached Files

•  Mini Vulnerability Asessment 2.0_10.xls   57KB   728 downloads
•  VARM 151112_1.xls   946KB   708 downloads

[syju28380]

Hi Charles,

 

Can you please upload " Vulnerability Asessment,3.0a.xls " again; the attachment is not working (the other attachments in the thread are fine) when clicked. 

 

Thanks,

Syju

[Charles.C]

Hi Charles,

 

Can you please upload " Vulnerability Asessment,3.0a.xls " again; the attachment is not working (the other attachments in the thread are fine) when clicked. 

 

Thanks,

Syju

 

Hi Syju,

 

Seems to work OK for me ?

 

Try again ?

[Charles.C]

Hi Charles,

 

Just comment for your vulnerability assessment : really very good

 

I have tried to make another approach, but finally yours is more simple.

 

So, I send the attachement of yours with little beat additional and vulnerability assessment of mine (need advise too).

 

Once again thank you for be inspired me

 

Warmly regards

 

Boedi

 

Hi Boedi,

 

Thks for all the variations and yr kind words. Scaling is indeed popular but also adds another layer of effort. Swings and Roundabouts ?.

 

Should note that sheet 1 in yr 1st Excel is same, I think, as my original upload 2a so still contains unactivated cells. Activated in sheet 2.

 

It might also be worth noting that using a 3x3 matrix may produce more results in the "Medium" area compared to a larger matrix.

 

As Ulrich previously emphasised, the slog factor remains in the data details. Not to forget validating the various "opinions".

 

IMO your original approach  also has benefit if one can allocate values for the “impact”. An example of that method is in the 2013 link below. Also illustrates the potential complexity of evaluating supply chain risk  if one chooses to “dig (too?) deep”.

 

http://www.supplycha...nd-reduce-risk/

[boediprasetyo]

Hi Charles,

 

Thank you for your advise and another reference.

 

Warmly regards,

 

Boedi

[Charles.C]

Hi All,

 

I have added a 5x5 version using same procedure as for 3x3 plus included a small modification  so that procedure should be fairly readily adaptable to any other matrix sizes  / zonal layouts if so required.

 

One (obvious) deduction  is that if you (100%?) sample every lot and have near-perfect detection you can be  invulnerable !  Also, but not quite, true for the 3x3 scenario. Alas, not a very realistic option in practice.

 

The basic mathematical logic employed in this method is fairly well documented for other applications, eg auditing, but the use of detection in a 2D risk matrix is afaik new territory.

 

Please inform if any computational error(s) encountered, I’m not a frequent Excel user.

 

 Vulnerability Assessment, 5x5,1.0a.xls   19KB   1069 downloads

[Charles.C]

Hi All,

 

I have constructed  an Excel  “low / medium / high” Table of  (Likelihood vs VA Factor Characteristics) to assist  the 3x3, 5x5, VA excel sheets  in posts 1,11 from various data sources, particularly the well-known USP document below.

 

 vulnerability,vul1 - USP guidance on food fraud mitigation.pdf   1.08MB   870 downloads

 

 Likelihood Levels vs Factor Characteristics for VA use, 2.0c.xls   18KB   934 downloads

 

The cell comments are inevitably subjective and occasionally speculative. May also be some typos in which case pls inform. I hope may  give an  idea of possible “ways to go”.

 

As far as databases for fraud history are concerned, a few have been given in various other threads but the links IMEX often do not work or are not free/publically accessible, eg

 

http://www.ifsqn.com...3815#entry93815

 

I also noticed there are some individual tables in the literature which offer rankings for countries’  “business reputabilities”. I will add later if I can find any free ones.

 

I also noticed this quite neat tabular overview from a database –

 

 food fraud incidents by method - food category.pdf   517.73KB   935 downloads

[Charles.C]

addendum to previous post -

 

PS - Please note that I have updated the original Excel version 2.0b slightly to 2.0c

[prembibo]

this is excellent! 

Good

[Kehlan]

Reading your "likelihood levels V factor characteristics for VA use" document... can I ask how you would judge point 8 in my case?

 

Our ingredients are fresh, cut herbs with no processing whatsoever other than harvesting and packing into crates. Each herb is very distinctive in its leaf shape, taste and smell and hence instantly recognisable. I am struggling with the point about sophistication of detection techniques... on one hand, the human eye is the most sophisticated detector out there, there's no machinery that can duplicate what it can do, but looking at something is not of course, an accredited laboratory technique.  so how do I score it for this question?

[Charles.C]

Reading your "likelihood levels V factor characteristics for VA use" document... can I ask how you would judge point 8 in my case?

 

Our ingredients are fresh, cut herbs with no processing whatsoever other than harvesting and packing into crates. Each herb is very distinctive in its leaf shape, taste and smell and hence instantly recognisable. I am struggling with the point about sophistication of detection techniques... on one hand, the human eye is the most sophisticated detector out there, there's no machinery that can duplicate what it can do, but looking at something is not of course, an accredited laboratory technique.  so how do I score it for this question?

 

Hi Kehlan,

 

I interpreted the BRC text via USP, eg  -

 

A simple approach for making this estimation [USP's terminology of "susceptibility"] is to determine how well and specifically a method and specification or suite of methods and specifications characterizes an ingredient, and how well the tests and specifications exclude fraudulent ingredients or components from passing the QA test.

 

i summarized the above as  "efficiency"  as "quantified" in the excel Table with respect to the features mentioned.

 

Yr query may also overlap factor 4.

 

I'm not familiar with yr product/its intended usage (ie its specification).

 

How well does your visual method score in respect to efficiency + specification ? Can you validate yr opinion ?

 

PS - also note that afaik "detection" does not solely refer to the final user, the contrary applies. If you are purchasing from a broker,etc  there is a shared responsibility. This is the point of "Supply Chain".

[Kehlan]

the product is fresh cut herbs... that's it.  The herbs arrive in crates, it is packed into plastic sleeves to the required weight, the sleeve is sealed, labelled, packed in a box and goes out again to the customer.  All very low tech.

 

From a point of view of fraud and substitution, there are no substitutes that would not be instantly recognisable.  coriander is coriander, there's nothing else that looks, smells or tastes like it. 

 

Its the bit about "sophistication of test methods" that worries me although maybe I am worrying too much... overall its a low risk product for a lot of reasons.  In terms of "is it coriander or is it something else?" then yes, I'd say a visual inspection is pretty infallible, to a trained QC at least.  But if by "sophistication" (which is the word used in the BRC standard) an auditor means verifiable lab results, then no, it isn't sophisticated at all.

[Charles.C]

Hi Kehlan,

 

I hypothesize that what BRC actually meant was nearer -

 

Is the "detection methodology" sophisticated enough to identify adulterants ?

 

Just for example, if the coriander product was considered adulterated if mixed with, say 5%, of some other (presumably cheaper)  herb X, and the sampling / visual examination routinely applied could only detect a difference when, say >10%, was present, the current test procedure would presumably be considered  inadequately sophisticated.

 

Just speculating, I have zero knowledge of the herb trade or whether it has any History of Fraud. Does it ?

[Kehlan]

No, there are zero incidences of known fraud in fresh herbs.  Processed is another matter of course... once a herb has been dried and reduced to powder it could be anything.  while its hard to tell what is really in your Earl Grey teabag, a harvested plant that is still in one piece, is much more obvious.

 

You do get some weeds mixed in of course, its noted on the QC intake sheet and the packers remove the weeds while packing.  so on packing every single stem is handled and is examined visually by the people packing.  So I think we have a case for being able to say that our methods are sufficiently sophisticated.

[Charles.C]

No, there are zero incidences of known fraud in fresh herbs.  Processed is another matter of course... once a herb has been dried and reduced to powder it could be anything.  while its hard to tell what is really in your Earl Grey teabag, a harvested plant that is still in one piece, is much more obvious.

 

You do get some weeds mixed in of course, its noted on the QC intake sheet and the packers remove the weeds while packing.  so on packing every single stem is handled and is examined visually by the people packing.  So I think we have a case for being able to say that our methods are sufficiently sophisticated.

 

Hi Kehlan,

 

Yes, sounds quite impressive. I didn't know that herbs are that finely distinguishable.

 

PS - dried herbs seems to be a different kettle of fish -

 

http://www.foodmanuf...aud-among-herbs

[BrummyJim]

As long as you don't make specific claims for your herbs (thyme and not lemon thyme etc) then you seem to be sorted. We make specific claims for variety (fruit juice) and so we submit products to the lab for authenticity testing to be sure it's alphonso mango and not tommy as an example.Similarly we take care with Sicilian lemon.

 

If there are low cost methods for authentication, you might want to consider them to demonstrate that your methods are adequate.

[Kehlan]

Brummy Jim, how are you testing for variety authenticity and how often? do you test every batch or just randomly based on risk assessment?

[BrummyJim]

We require our suppliers to be able to provide authenticity results for each batch they deliver, and we also have a risk based back up where we will send samples for authenticity analysis. Not cheap, but better to be sure. The requirement on suppliers is standard in the industry as there has been quite a lot of adulterated juice in the past.

[Charles.C]

As long as you don't make specific claims for your herbs (thyme and not lemon thyme etc) then you seem to be sorted. We make specific claims for variety (fruit juice) and so we submit products to the lab for authenticity testing to be sure it's alphonso mango and not tommy as an example.Similarly we take care with Sicilian lemon.

 

If there are low cost methods for authentication, you might want to consider them to demonstrate that your methods are adequate.

 

Hi Brummy Jim,

 

Just curious. Is the motivation for substitution / adulteration  purely cost or are other factors like availability additionally/solely relevant ?

[BrummyJim]

Economic. Sugar is much cheaper than juice, and since some juices are used solely for their sweetening capabilities, it's not easy to taste the difference. It also makes expensive juices like pomegranate go further.