12 replies to this topic

[Bakerworth]

I work in a bakery manufacturer, and we have now had 2 instances where the crumb cake was rejected from the metal detector but no metal was found in the product via visual inspection, and magnet. Has anyone experienced this? MD is a CCP and is causing us to discard entire lots, but we cannot find root cause. 

[Scampi]

What are your thresholds?

 

And does it contain cocoa powder?

Edited by Scampi, 10 December 2024 - 03:50 PM.

[Planck]

This product is quite a difficult application in a metal detector and needs to be handled by particularly specialised technicians, making it almost impossible for the general user to adjust it successfully.

 

When you say it happened 2 times, over how long a period of time? Or in total how many products happened 2 times?

 

Generally perform the following steps:

1, Confirm that the metal detector itself is stable. This check confirms that the device itself as well as the absence of interference from the outside world.

2, Still on empty, confirm that your test piece can achieve reliable or acceptable detection sensitivity.

3, Get your product through the metal detector and do everything you can to get the product's signal low enough.

Also verify your test piece, the signal obtained by the test piece must be high enough or at least acceptable.

How to get your product's signal low enough is at the heart of parameter tuning.

And there are different methods for different brands of machines.

 

 

Acceptable here is generally not defined.

However, we consider that it is desirable for the false rejection rate and the leakage rate to all be less than one in 10,000

Of course, it also depends on the size of the test piece you use.

 

A case can be given here as a reference:

A factory supplying bread to McDonald's and KFC uses LOMA's metal detectors and passes through approximately 24 slices of bread in film bags at a time.

Detection sensitivity is Fe 1.5, NFe 2.0, SS316 2.5.

 

Lastly, may I know how long it was or how many products happened 2 times?

 

After the MD is adjusted to an acceptable state, you can calculate the false reject rate from the signal value of the product and the threshold you set.

 

If the calculated false reject rate is close to the actual false reject rate, then this is normal.

If the calculated theoretical false reject rate is much lower than the actual false reject rate, then this could be metal, or some sort of interference.

If the theoretical false reject rate is high... Don't worry, it's not going to happen.

Because there is no way you can accept that in this case, there is no need to calculate it at all.

 

The fact that you asked this question shows that your factory is using MD very seriously, good luck!

Edited by Planck, 10 December 2024 - 04:00 PM.

[Bakerworth]

What are your thresholds?

 

And does it contain cocoa powder?

1mm Ferrous, 1.2mm Non-Ferrous, and 2.5mm Stainless Steel.

More than 4 metal pieces or packages within 1 hour

 

There is no cocoa powder in the product. 

[Bakerworth]

This product is quite a difficult application in a metal detector and needs to be handled by particularly specialised technicians, making it almost impossible for the general user to adjust it successfully.

 

When you say it happened 2 times, over how long a period of time? Or in total how many products happened 2 times?

 

Generally perform the following steps:

1, Confirm that the metal detector itself is stable. This check confirms that the device itself as well as the absence of interference from the outside world.

2, Still on empty, confirm that your test piece can achieve reliable or acceptable detection sensitivity.

3, Get your product through the metal detector and do everything you can to get the product's signal low enough.

Also verify your test piece, the signal obtained by the test piece must be high enough or at least acceptable.

How to get your product's signal low enough is at the heart of parameter tuning.

And there are different methods for different brands of machines.

 

 

Acceptable here is generally not defined.

However, we consider that it is desirable for the false rejection rate and the leakage rate to all be less than one in 10,000

Of course, it also depends on the size of the test piece you use.

 

A case can be given here as a reference:

A factory supplying bread to McDonald's and KFC uses LOMA's metal detectors and passes through approximately 24 slices of bread in film bags at a time.

Detection sensitivity is Fe 1.5, NFe 2.0, SS316 2.5.

 

Lastly, may I know how long it was or how many products happened 2 times?

 

By 2 times, I mean 2 production runs, approximately 2 weeks apart. Each time we have had 6-7 pieces rejected in the middle of the run within one hour but not all at once.

 

Maintenance reviewed the machine and confirmed it was in working order. Verification was performed using the test wands hourly to ensure the machine is working properly as well both with and without the product. 

 

We are still in the process of Validating our MD as this is a new process and was just installed about 3 weeks ago. We have not yet had a chance to perform validations on the crumb cake specifically, right now we are running based on a brownie/blondie as it is considered a "like item" at our plant. 

 

It is looking like without this validation, we cannot run this product. I was hoping to find some alternative reason that the MD would be set off to justify releasing the product due to only 7 pieces of the lot being rejected. 

[Planck]

 

1mm Ferrous, 1.2mm Non-Ferrous, and 2.5mm Stainless Steel.

More than 4 metal pieces or packages within 1 hour

 

There is no cocoa powder in the product. 

 

Can you pleae go ahead and tell us the size of the aperture in this MD? The aperture is measured this way:

 Width and height of apertures.jpg   517KB   0 downloads

thanks.

[Planck]

By 2 times, I mean 2 production runs, approximately 2 weeks apart. Each time we have had 6-7 pieces rejected in the middle of the run within one hour but not all at once.

 

Maintenance reviewed the machine and confirmed it was in working order. Verification was performed using the test wands hourly to ensure the machine is working properly as well both with and without the product. 

 

We are still in the process of Validating our MD as this is a new process and was just installed about 3 weeks ago. We have not yet had a chance to perform validations on the crumb cake specifically, right now we are running based on a brownie/blondie as it is considered a "like item" at our plant. 

 

It is looking like without this validation, we cannot run this product. I was hoping to find some alternative reason that the MD would be set off to justify releasing the product due to only 7 pieces of the lot being rejected. 

"Each time we have had 6-7 pieces rejected in the middle of the run within one hour but not all at once."

This looks very bad indeed.

 

“We have not yet had a chance to perform validations on the crumb cake specifically, right now we are running based on a brownie/blondie as it is considered a "like item" at our plant.”

For your ‘like item’, do you use the same aperture size MD? The same test piece?

What's the false reject rate there?

 

Frankly, that's a pretty high sensitivity requirement, and I haven't seen such a high sensitivity requirement.

Of course, I don't know your aperture size yet.

[Scampi]

 

"We are still in the process of Validating our MD as this is a new process and was just installed about 3 weeks ago. We have not yet had a chance to perform validations on the crumb cake specifically, right now we are running based on a brownie/blondie as it is considered a "like item" at our plant. 

 

It is looking like without this validation, we cannot run this product. I was hoping to find some alternative reason that the MD would be set off to justify releasing the product due to only 7 pieces of the lot being rejected. "

 

 

You're assumption is correct, you should NOT be running actual product UNTIL the MD is validated.  I sure hope you've got the manufacturer on site to assist

 

Blondies/Brownie's should NOT be considered like products as they are incredibly dense, and a crumb cake is quite airy.  

 

Your thresholds are incredibly small, and will likely create many headaches for you

Edited by Scampi, 10 December 2024 - 05:06 PM.

[kingstudruler1]

Metal detectors will reject "falsely" some times.    Vibration, power flucuations, product flucuations, many things can cause false rejects.  

 

I dont think I would create a procedure that product was automatically discarded due to a an unconfirmed reject.  Just like I would not release product during a known metal incident just because a metal detector was present.   I would determine disposition of product based on the entire investigation.   

 

Metal detectors are a great piece of technology.   However, as technology they are not perfect.    They can/will miss metal depending on composition, size, orientation, etc.  They can/will also dectect when no metal is present.    Metal detectors are sometimes treated a magic box.   However, food safety personnel need to understand the technology and its limitation(s) in order to make informed decisions.  

[G M]

Something as simple as a variation in moisture or salt content could throw false rejects with tolerances that tight.

 

I wouldn't use the machines results to make any decisions until you had a validated procedure.

[Dorothy87]

Hi  

 

do you have aluminum baking trays? how did you inspect the cake? I had a few situations which the aluminum baking trays were damaged. A tiny piece of aluminum was found... magnets won`t detect aluminum. 

[jfrey123]

Lots of good advice coming here.  Also want to double check with you OP:  you said nothing was found in the rejected product, but did you re-run the rejected trays, plastic, and label to see if it trips a rejection?  More than once I've had metal found embedded in a plastic tray, cardboard boxes, and once I even had an ink change by my labeling manufacturer start to trigger false rejections.

 

But otherwise, looks like from other replies the crumb cake is a new product for you guys.  I think you should consider completing a full validation on it, preferably with a rep from the manufacturer to supervise, as the different density from your routine products is most likely triggering the false rejections.  

[Bakerworth]

Thank you all for your replies! I appreciate the helpful thoughts and tips. I read through all the comments and checked everything from packaging to labels and determined that the machine is just not set up properly for this item. We are putting everything on hold and pausing crumb cake production until we can complete our validation. Better safe than sorry.

 

Thanks again! It is great to be part of such a helpful community!