Mettler-Toledo International Inc. (MTD) Earnings Call Transcript & Summary

It's very important the commissioning of a new metal detector is carried out by the equipment manufacturer or their appointed or authorized agents, i.e., a competent or qualified engineer. Mettler-Toledo do this by following a documented process detailed in our IPac. So here's an example of the documentation that our engineer will follow. This will all be filled out during the commissioning process. So what happens is the engineer arrives on site. The first thing he does is he will check the equipment condition and the packaging just to make sure that there's been no damage to the equipment or the components during transit. He will then check the mechanical installation. So is the equipment level, has the equipment been positioned properly, and is the system height correct, is there anything in close proximity to the metal detector e.g. metal structures or trunking that may potentially affect the performance of the machine. Then he will look at the electrical installation. Is there a clean, uninterrupted power supply to the equipment to ensure good operation, does it need a mains filter, that's something that our engineer can actually advise when he gets on site. Then we look at the equipment location. Are there any cables or devices positioned close to the detector that could cause interference, is there anything that could cause interference due to heavy vibration, for example. Then he'll move on to operational checks. So he'll power up the machine. Does everything power up as expected, is there power to both the metal detector and the conveyor, is there pneumatic pressure to any reject devices if they're fitted. And then he'll move on to a functional check, a systems function check. Is everything working, is the stop-start system working on the conveyor, are the emergency stops working, are the sensors working, is the reject device working, any due diligence features that might be on there. He checks all the sensors and the switches. And he'll also check all the safety switches and guards associated with the machine as well. Then he'll go on to the performance checks. So he'll set the optimum detection sensitivity and reject timers for all of the products provided. Products will be set up with a maximum product signal of 50% detection trigger level. He'll actually determine the test piece sizes required to ensure a maximum -- sorry, a minimum of 150% of trigger level or confirm detection of the customers' sensitivity specification meets 150% of trigger level. So he'll actually take the customers' test samples and make sure that we're as a minimum at 150% of trigger level. Then he'll do some operational training to a basic level. So he'll do training for operators, training for QA operators or -- and training for engineers, if required. Training attendees are logged and documented in the IPac folder. Any additional training can be delivered at a more detailed level and can be erased at a later date if required. Okay. Then he goes on to the performance verification. So he'll confirm detection sensitivity with the calibrated test samples, ferrous, nonferrous, stainless steel 316, and aluminum if required. Test samples are tested at the front, middle and rear of the pack and positioned in the pack to travel through the center of the aperture, although be put on the top of the pack if the top of the pack is below the center line of the aperture of the metal detector. For reactive or wet type products with product signal, it's absolutely essential that freshly prepared products should be used for the test packs. He will then confirm correct operation of the reject device in all positions, confirmation of correct functionality of any due diligence failsafe features, for example, reject confirmation systems, et cetera. And then once he's happy, everything is tested and everything is functioning as it should be, a performance verification certificate would be issued, which is normally valid for at least 12 months. And that -- all of that documentation is all kept in -- kept together with this -- with the IPac.

Thanks, Ian. So Ian has taken us through the importance of commissioning your metal detector and some of the actions required. Now we'll look at the validation. Your qualified metal detector installation team have set up your products, either to your site specification or to the metal detector's maximum capability. So now it's time for you, as site personnel, to validate your metal detector. That means checking the settings for each product in order to demonstrate due diligence and ensure that the metal detector will operate in accordance with the site specification. You're ensuring that the metal detector will reject contaminated products on detection. In addition, you're checking that any warning or signaling devices are effective and working. You are also confirming that the installed fail safes are functioning correctly. And lastly, you're about to confirm the acceptable false reject rate for your plant. So false reject rates, 1 in 100,000, 1 in 1 million. They're perfectly acceptable false reject rates and you're working towards those. The verification procedure should ensure that the -- whether it's your production line or your particular product sensitivity standard is being complied with, and basically answers that question, will this piece of equipment do the requirement that we bought it for. The results should be documented, of course, and communicated to all and available for anybody to audit, et cetera. So to carry out the validation, you will need to test products. You need the actual products that you're going to put through and use in normal operation. And you need to do this validation with every single product that's going to go through and be inspected with the metal detection system. And obviously, the products that you're going to do the test with need to be contaminant-free. They need to be the same weight, size, et cetera, exactly as a normal production product would be. Then in order to validate your metal detector, we need to run through the process. So here is the process. So firstly, you're going to take clean non-contaminated products and put them through 30 times. So why do we do it 30 times? Different companies have different standards and use various different numbers, and you're free to use your own. From our standpoint, when we assess this in conjunction with the probability of detection, we believe that 30 passes gives you a high degree of confidence that the test is accurate and a high enough confidence, at the same time, it's not too onerous on the actual test itself. So, I mean, you could pass the products through 5 or 10 times. You'd have a lower level of confidence in the equipment. Whereas if you passed it through 100 or 150 times, you'd have a very high level of confidence. But to do that for every product and every contaminant test piece is an awful lot of work. So on balance, we think that the 30 test passes is a good trade-off between a high confidence level and not a ridiculous amount of work, so to speak. So we think that 30 passes is sufficient. So now to move on to what the test looks like. Firstly, we take the 30 clear products and run them through. And while we're doing this, we're actually monitoring the display on the metal detector. And what you want to see is a nice consistency of the product. The product should be run below about 50%. So on our metal detectors, that means about less than 5 green bars. So a long way from any potential trigger. And if you have the ability to record this, some companies like to record each of these test pieces, the levels of each of them. So when you put the 30 test pieces through and it passes without any reject, then it's passed the clean product test. So next then, you want to be able to do the ferrous test pieces. So to do that, you need to get your product and then put it through 30 times with the ferrous piece at the front. Then 30 times again with the ferrous test piece in the middle and then 30 times with the ferrous test piece at the rear. And you do this for each of the metal types. So also the same for nonferrous and stainless steel. So that means there's 90 tests for each metal type. So you pass the ferrous through 90 times, ensuring all of them are rejected. And also, again, like you did with the clean products, you want to monitor the display and ensure that there's a consistency with the display so that the rejects, once again, we recommend about 5 red bars on our display, so that's a trigger plus about 50%. So the metal detector has really seen the product and is giving you a very strong detection and hence, rejection. So that means then that you've got 90 ferrous test pieces rejected, then you do the same with the nonferrous and then you do the same with the stainless steel. Recording the results, but also, you need to record that the product that has been tested, the time and date of the test, the test sample type and sizes used, maybe the names of the people carrying out the test and the confirmation that the metal detector is fully functioning. So this is validating each product that this metal detector system is fit for purpose, and will do the function of inspecting the products that's required. Now we're going to do the demonstration of the validation procedure. And for that, I've got my assistant, Ian, to help me out catch the packs. So as I said previously, the process is to put 30 clean packs through and these are real products, so you have to do this for every single product. And while you're putting them through, you're monitoring the display and ensuring that the packs are going through cleanly, and they're nowhere near triggering or rejecting. So we would recommend that the clean packs are going through less than 50% towards a trigger. When you're putting through the test packs with the test samples, we would recommend that the trigger plus about 50%. So with our metal detectors, that's about 5 red bars. So non-test piece packs, less than 5 green bars; test piece packs, 5 red bars or more. You can look at that using our display, or in this metal detector, it's got a great facility where the -- it indicates the actual level, the peak level of every single pack that goes through. So this is a real great feature that you can use during this validation process. So as I said, the first objective is to test the clean uncontaminated packs. So this is what I will use to represent 30, but I'm actually only going to put 3 through. So the first process would be to monitor the screen. Take the 30 good test piece -- test packs, uncontaminated and put them through. And all 30 should go through without any reject or even being close to reject. As I say, we recommend well below the 50%, so 4, 5 green bars maximum. Then you follow the same procedure with the ferrous products, with the ferrous test piece on. So we do 30 with the test piece at the front. And ensure that all 30 are rejected, and again, ensuring that we get a good strong signal, at least 5 red bars. Then we do 30 with the ferrous with the test piece in the middle. Again, 30 rejections ensuring that we get a good strong signal in all 30. And then finally, we do 30 with the test -- ferrous test piece on the rear. And again, we want 30 good, strong rejects. So that's the validation process. We'd follow the same through that with the nonferrous, 30 front, 30 middle, 30 back; and same with the stainless steel, 30 at the front, 30 at the middle, 30 at the back and ensuring at all times that the test piece display is 5 red bars or more. So that is the demonstration of validation process to ensure that the metal detector per product is working per specification and all the associated fail safes and reject mechanisms are all functioning correctly as well.

Thanks, Paul. Now the metal detection system has been installed, commissioned and validated, it's now ready for use in production. However, consideration now has to be given to regular monitoring of the detector's performance whilst it's in production. First point to consider is how often should the metal detector be performance tested. This frequency of testing will normally be determined by: one, if the metal detector is a CCP, critical control point, and test frequency will be determined by a risk assessment in line with the HACCP-based standard that you actually follow, or it could be in line with your customers' code of practice which may already specify frequency of testing as a minimum. Secondly, what tools are required to carry out the performance testing? So first of all, calibrated test samples for each metal type of the correct size and type of carrier appropriate for the product, e.g., for example, we have test cards, such as this. We have test sticks. So you might use a test card if it was a nice dry product, a nice packaged product. If it was a wet product, then we would recommend you use something like a test stick where the test sample is actually encapsulated inside this plastic carrier here. Okay. Test samples need to be positioned in the product to pass through the weakest point of the metal detector, which is, of course, the center of the aperture. And they need to be positioned front, middle and rear of the pack to fully challenge the reject time as set up. If the top of the pack is below the center line of the detector aperture, then the test sample should be positioned on the top of the pack. The third thing to think about is how will the results of the performance testing be documented. So there are a number of ways. The old way used to be via old paper record. So the results were recorded on paper sheets. This is becoming less and less common now as most organizations, most food manufacturers want to try and reduce the amount of paper that's actually being taken out onto the factory floor. One of the opportunities that we have is a printer, so you can connect it to -- directly to a printer and print out the results of testing and performance monitoring, or we can use a USB stick such as this. This can be connected to the metal detector and again, all of the results can be actually downloaded onto the USB stick and then they can be uploaded onto your systems -- your computer systems directly from the stick. And then the third option is to actually have all of the results automatically downloaded onto -- via a network connection or by dedicated software, which can be provided by the manufacturer. So for example, Mettler-Toledo can provide a data acquisition and monitoring system that we call ProdX, and that's a very good system for actually monitoring, recording and documenting all of the results associated with your metal detection systems, and. We'll talk about that later on in the presentation.

Thanks, Ian. That was really interesting. Okay. So, so far, we've discussed the importance of correctly commissioning your metal detector system, then we followed that by the importance of validating the metal detector system with the products that you're going to inspect during its normal operating life. And then Ian just did a great job of taking us through the issues around performance testing and monitoring of your metal detection system. So now it's time to move on and look at the test themselves. So we're going to demonstrate the main tests that we recommend. That being the consecutive test and the large metal test. We'll also demonstrate the memory test. And to begin with, I'll demonstrate the standard test. Okay. So we've discussed the need to install and commission your metal detection inspection device correctly, very important, and we've demonstrated how to validate your metal detection system. And Ian has just advised us the considerations to be taken into account when we start to operate the equipment and what is required at verifying sensitivity, the test frequency, where the test samples are placed, et cetera. So now it's time to start performing a test on the test equipment as in normal operation. So first, we will look at the standard test. Before we start, I just have to say that our metal detectors already have the test routines programmed into them. Therefore, there's very little training required and the operator only has to follow the instructions on the screen. So this is great, the metal detector will guide the person through the test process. If your metal detector does not guide you through the process, then you'll need to codify the procedure that the quality operator has to follow. So the standard test comprised of putting the 3 test pieces through the metal detector, randomly during normal production, insert it into product flow. So we have the good packs. We'll then pass through the metal detector, and the test phase packs will obviously be rejected. So we can see, at our metal detector, that we already have an overdue indicator from our test procedure. So this machine now requires a test, as indicated by the overdue indicator. So as the quality operator, I would come along and insert my passcode and then select the performance test icon. And now I'm presented with the options of which tests to perform. So in this case, we're going to do the standard test, so I select that. And up pops the 3 metal type tests, so the ferrous, nonferrous and stainless steel. I can do these in any order, but I guess we should use the ferrous first, so I'm going to select ferrous. And the screen is telling me to pass a 1.5 ferrous test sample through the product flow. So obviously, normally, you can't walk behind the metal detector, but thankfully, on this time, I can. So I take the first test piece that's located at the front of the test pack and put it into the production flow and ensure that it's rejected as it was there. And as you can see, the screen indicates that the test was passed, so I can acknowledge that and move on to the next test. So in this instance, I'll use the nonferrous. So the display is saying that I need to get a 1.8 nonferrous test piece, which is located in the middle of the test pack, as we have it here. So I put that into the production flow. Again, I can see and confirm that the test pack has been rejected. The display also confirms that, so I can confirm it as well. And I'll move on to the final test sample, which is the stainless steel. So that's a 2-millimeter stainless steel test piece that's located at the rear of the pack. Again, I put it into the production flow. Once again, it's rejected, which is indicated here on the screen, so I can acknowledge that and acknowledge that I've carried out the whole test. And the displays tell me that the test is complete and everything's good. So we've passed the standard test. Over to Ian.

Okay. So Paul has just demonstrated the standard test. Many retailers have a requirement to test the rejection of packs consecutively to challenge the performance of the detector head and the reject mechanism in the case of multiple contaminations in the product. As you can see now, the detector is ready for a test. So what I'm going to do now is just run it through a consecutive test. Before I do that, I'll just show you the test packs. So you can see the ferrous test pack. We'll have the test sample positioned at the front. The nonferrous test pack, we'll have the test sample positioned in the middle of the pack. And the stainless steel test sample, we'll have the sample position at the rear of the pack. Okay. So I'll just start the machine. So I'm going to log in now as a QA operator, select "performance test." And then as you can see now, I can actually select a consecutive test on the screen. And now I can carry out the test. So I need to make sure that I put the test packs onto the conveyor system with the spacing identical as if it is in production and also at the same speed. So as you can see, all 3 packs were rejected. And on the screen here, we have a green tick, which indicates that the test was successful. You can also see, just at the top of the screen here, that there's a count. So we've actually counted the 3 packs through the system. So that was a successful consecutive test. We can select the test passed, green tick on the screen there. And log out again, and that's the end of the test.

Okay. Now for the memory test. Some retailers have a variation of the consecutive test, and that is the memory test. Again, this is a further requirement to challenge the performance of the metal detector and the reject mechanism combined. However, in this case, the test comprises of 3 packs with test samples separated by 2 packs without test samples. The test runs all 5 packs, so the theory is that the 3 packs that have the test samples are rejected and the 2 clear packs are not rejected. In certain cases, the pack speed, pack spacing or even metal detector sensitivity may mean that this test isn't really appropriate. And we all here think that the consecutive test is the best way to test this rejects and metal detectors together. Now if we look at the test pieces as they go through and the test packs, I'm going to enlist Ian to help me demonstrate these. The first ones through will be the ferrous test pack, and then there will be a clear one. And then it will be the nonferrous test pack and that will be completed by a clear pack, and then it will be the stainless steel with the test piece at the rear. So thank you for that, Ian. so once again, I need to log on as a quality operative. I'll use my own passcode this time. And you can see, I've got the performance test icon available and the tests that are available. And this time, we know we're doing the memory test and the display is telling me to actually put 5 packs though, which we know, which is good. So I'll start the metal detector and I'll start the test. So first will be the ferrous, then the clear, then the nonferrous, then the clear and then the stainless steel. Okay. So we can see clearly that the 3 test -- rejected test piece packs are in the bin. Ian, hopefully, has the 2 clear packs? Yes.

Yes.

The metal detector display's indicating that they passed the test, so I can accept that. The test is complete, and that's the conclusion of the memory test.

Okay. So Paul has just demonstrated the memory test. Many retailers have a requirement in their codes of practice to check reject timers are correctly set up where a side reject device is being used, for example, a pusher like on this machine here or an air blast. And in these -- when we use these types of reject systems, it is absolutely essential that we use a photogated timer setup. Incorrect setup or incorrect use of the correct type of timer can allow large contaminants to be missed by the reject system and allowed to continue down the line for final packing. The large metal test will confirm that photogating is properly enabled and prevent this situation happening. Okay. So before I do the test, let's just talk about the test pack. So we'd use a normal test pack such as this, as we've used before. And inside this test pack, we have a large metal sample. And when I say a large metal sample, we're actually using a 20-millimeter diameter ferrous ball. Just to give you an indication of the size of the test sample, I'll just show it to you. There, okay? So that's inside the pack. So now I'll just run the test. I'll just log in as a QA operator, as usual. I'll select performance test. And as you can see, we've got the option to select a large metal test here. Okay. So now I'll just pass the test pack. And as you can see, that was correctly rejected into the reject bin, and that's exactly what we want to see. If the photogated timer system wasn't actually switched on, then what would probably have happened is that the reject system would have operated well before the packet actually reached the reject system. And in certain situations, that can actually allow the pack to get past the reject system and go on down the line for final packing. So essential that the photogated timer system is used with the side reject, and this is a good way of testing to make sure that it is actually operating correctly. So as you can see on here now, we have confirmation that we passed the test. So I'm just going to log out of the system, and that's a successful large metal test completed.

Now we will take a look at reporting. Obviously, if you followed the best practice in testing your metal detector, you want to be able to match that by having the best practice in recording results. Clearly, this is one area where common practice probably varies most from best practice. Probably the widest practice today is still that of recording the results on pieces of paper. Clearly, this is not best practice. Yes, it can seem the most convenient, but it has a lot of limitations. Surely, in the not-too-distant future, we will move beyond this method. Obviously, it's easy to do on paper, and I have an example here. But what happens if the paper gets lost or it's damaged, which could happen easily. Inherently, there's difficulty to prove the accuracy of the results. And there's always seems to be a question of the authenticity of the results. You just ask any auditor, and they will tell you stories of instances where the paper-recorded details have authenticity issues. Also in my experience, there always seems to be very little information in recorded results as is the example here, it just states the machine, the test sample size, the day and time and whoever carried out the test. So clearly, this is not best practice. We want to look at being able to use a system where no human intervention is required in recording the results. That would be best practice. A system whereby the results cannot be lost or damaged, where no human intervention in recording results then means the accuracy of the results is not in question. If there's no human intervention recording the results themselves, then the authenticity of the results is never in question, plus lots of data was recorded. And I have an example here. So machines whereby all of this information is stored automatically and is either available by print out to a printer, or as in this instance, it's available on a USB or has an electronic connection. The data covers all of the areas of the test. So in this example, it has the time and date, as you can see in your right panel below. It tells you the serial number of the machine, who did the test, which line. It has the details of the product that's in production and then goes into great detail about the ferrous, nonferrous and stainless steel tests that were undertaken. So clearly, this is moving towards the best practice. However, there is an even better method than this. And again, if there is a best practice, we want to look at being able to adopt an online system that's continually managing and monitoring the system, a system that records all of the test data, a system that can present this data quickly and easily to any personnel that's authorized to see it, a system that holds the entire life cycle information of the metal detector will be great, and a system that can, at the press of a few key strokes, present the last hours' worth of test data, the last week, the last month, the last year, et cetera, very easily and quickly. Furthermore, a system that can then take this data and send it to other systems via electronic systems, e-mail or text, et cetera. And this machine -- this metal detector is connected to one sort of system. So we will take a look at that now. Now we're going to take a look at the ProdX software. So this is the application that's taken all the data from the metal detector or the test -- performance test information and recording it and storing it for users to interrogate. So we're going to look at the system. As you can see at the moment, we're looking at the metal detector, the live information from the head, but we want to look at the events data. So I'll navigate to this events list. So the information now is presented to me in a format, whereby I can look at live data or I can look at historical data, and I can focus on in a particular period of time and date. Obviously, this is our test system, so we don't have test running all the time, so I ran a series of tests on the 22nd in order to store the data for this demonstration. So now if we look, anybody who's looking at the system could see that we can go through a historical view of the tests and confirm that they've been carried out with the test information is presented in the terms of it's got the time and date stamp. It's got what the test was being performed and that it was passed or failed, who actually did the test, who carried the test out and the actual results from test and the size of the test pieces that we used and whether the reject was confirmed. All this data is then stored. Automatically, it's taken from the system without any human intervention. So if we go through it and just take an example, we can look through and we can see, historically, we started by doing a consecutive test. The QA operator logged on 12:58, on the 22nd. They did the test by inserting a ferrous of 1.5 millimeter, nonferrous 1.8 and stainless steel of 2 millimeters. We've got the maximum signal strengths for each of the test packs and they passed the performance test, and so on and so forth. We can then go into and interrogate all of the system. And you may notice here, there's a few entries that are not black. They are amber or red. So in this instance, we can see that a test was actually due or is being requested. And actually, no action was taken and therefore, it went into a test overdue. So you can see when tests are overdue. Likewise, you may have noticed this other red indicator here, and this is telling the users that the test has failed. So what happened? Well, at 13:40, a memory test was performed by operator Paul. That was me. I put a 1.5 ferrous test piece through and then a clean pack, a nonferrous test piece of 1.8 and a clean pack. And they all did as was supposed, either rejected or not rejected as required. But when the stainless steel test pack went through, it didn't trigger, which is 100%, and therefore, the test failed. And so I did this in order to indicate what would happen when a test has failed. Straight away after that, I did the test again ensuring that it would pass. I used the correct-size test piece in order to do that this time. And you can see that the test has failed -- sorry, has passed. So the system is there. It's been able to store the data. It's non-human intervention in order to store the right information. And more importantly as well, users cannot go in and manipulate or alter the data. For example, I cannot go in and change this, "test failed," to a, "test passed." It doesn't allow me to do that. So the system is storing the data. The user can select the period of time that they want to look at, whether tests were performed and then they can ascertain whether the tests were performed, whether they're passed and all the associated test information. So that's what a product -- this product system does on our electronic online system, taking the data and storing it. There is one other feature that it has in that it can provide reports out straight from the data. So if I go into the system and go into these reports, I actually already have a report stored. So I'll go into my metal detector line 1 PV test report that I have stored, and this is asking me to look at the -- or to report on all the performance tests that were carried out in a certain period of time. That's why I've configured it for my line. So if I view that data now, the report will be created by going into the database and extracting all the data for the reports that correlate to the time that I asked for, which was on the 22nd or the fourth, between midday and midnight. Obviously, I didn't do tests on all of those. But all of the tests that were carried out along with date and time stamp, the user who did it, whether the test was passed or failed, et cetera, whether the reject was confirmed or not is all available then in a report format that can be electronically sent to people via e-mail. And in fact, this report can automatically be sent as well, so you could have this report, automatically send out the information every day, months, year, et cetera. So the system is able to not only store the data and present it back in a very clear and quick and concise manner, it can also send out reports to third parties. So that is what we would consider to be a best practice of the storing of the test result data. Thank you.