Mikkel Svold (00:12): Michael Sievers, hello and welcome to the show.

Michael Sievers (00:16): Thank you, Mikkel.

Mikkel Svold (00:18): Just for those people who don’t know you, you are the global key account manager for dairy at FOSS Analytical. I think many dairies around the world, if not all, know what FOSS is and the company FOSS, but can you just shortly describe what is it actually that you do what your daily work?

Michael Sievers (00:38): My daily work is to coordinate all our inline, online activities, meaning real-time measurements in process, supporting our daughter companies and our strategic partners like Au2mate, for example.

Mikkel Svold (00:54): Okay. We spoke, I think it was yesterday, just briefly and you said that about 80% of All milk worldwide is analyzed by FOSS sensors, which is quite astounding.

Michael Sievers (01:08): Exactly.

Mikkel Svold (01:09): I think that’s also why I think it’s very relevant to have you talk here because today’s topic is optimization of dairy production down to what we would say the second decimal. To me, optimizing a production down to the second decimal when we are talking a live raw product seems so crazy that you’re actually able to do that. But can you maybe just start by explaining me why is it interesting? Why is it smart to be so optimization focused?

Michael Sievers (01:53): Let me start a little bit with the history. I graduated around 35 years ago as a dairy engineer. I’m working now since 30 years in FOSS and I can remember when I started at FOSS, we placed a revolution. That’s the MilkoScan. That’s the instrument you mentioned that 80% of the milk is analyzed for payment and process optimization. When we launched this instrument 30 years ago, it was the first time the dairies could measure fat, protein, total solids in real time. They had to take a sample, bring it to the lab. They got the result in real time and then they adjusted the process. The reason why this is so important is 80% of the costs in a dairy are related to the raw material milk. Then on the other hand, you have to remember, at least in Germany, 80% of the retailer business is in the hand of four companies, so you cannot just increase the price. Milk butter are commodities, so the only way to stay efficient is by the perfect utilization of the ingredients in raw milk.

Mikkel Svold (03:17): Also, at least in Denmark, you can’t control the influx of milk as well. You have to take it from the farmers. So you have no real control over your or no day-to-day control, so to say, over your supply and you’re telling me you actually also can’t have control of your price. You can’t have control of your demand as well. So you have to find your business within that.

Michael Sievers (03:43): Exactly. Now you could say when we launched the MilkoScan, the lab sensor 30 years ago, why now online? When I started my career 1985 in dairy, we had in Germany 617 sites producing dairy products in 2023 and the milk volume has not really changed. You only had 210 sites left. This is now quite stable, so I don’t expect a huge reduction, but now you can imagine for the same type of milk, we have significant less sides and that’s even more important in Denmark. So that’s why we say every drop counts.

Mikkel Svold (04:33): Every drop counts. I think just to get it clear, because obviously there’s going to be some listeners out there sitting at some of the really huge dairy productions and they maybe know already because they may have already implemented this type of monitorization and optimization of the raw milk and how to control its way through the production. However, there’s going to be still quite a few of the medium-sized small dairies out there trying to look into what does it actually mean that you can optimize down to the decimal? I don’t know if you have any examples that you can give. What would that be?

Michael Sievers (05:18): It’s not easy to make a definition what is a small or medium-sized dairy. When I started decades ago, 500,000 kilos milk intake a day was a huge dairy. Today you would call this a small, medium-sized dairy, and for such a medium-sized dairy… Let’s give a practical example, for butter. Typically, if you control your butter process with lab instruments, meaning you take a sample, run to the lab, have the quick result, go back to the churn and adjust. Important to say is dairy is one of the most controlled industry in the world. Everything is defined-

Mikkel Svold (06:08): By regulation.

Michael Sievers (06:08): … by the Codex Alimentarius. Butter is not allowed to have more than 16% moisture. So if you now control such a process with a lab instrument, that gives you real-time lab results. That’s still much better than the old way where you have to wait 15, 20 minutes, but still you have some variations. Now you could say you have the opportunity to get the same quality of result every three seconds with the same performance like a lab instrument, meaning with the lab setup, you can in average reach 15.7% moisture because you will have some variation and not a single butter should be above 16.

(06:53): If we now say with this online system and a great control loop, for example, Au2mate where we work very closely together, you can reach 15.85. So now you could say this dairy is now able to add 0.15% more moisture. It sounds nothing, but if you produce five million kilos of butter a year… And that’s a small side, I have to say. The large sides have 20 million, 30 million kilos of butter a year. So only five million kilos of butter a year with the actual butter price of over seven euros, this small dairy could save around 70,000 euros a year by increasing the yield. Now, you-

Mikkel Svold (07:47): Just by a judging by zero-point.

Michael Sievers (07:48): 0.15. Exactly.

Mikkel Svold (07:49): One-five. Wow.

Michael Sievers (07:52): Now, you can imagine if a huge dairy has five times more, it’s a no-brainer, but we can now see this trend and I’m always saying the threshold for such an online system is around 500,000 liters milk intake a day and then it pays back. Another example is consumer milk. Sorry, I’m German, and this varies from country to country, but you could say a semi-skimmed milk in Germany is allowed to have minimum 1.5% fat. So the same like butter, but not in this case that we want to put more moisture into it. We want to go to the limit as much as possible. With lab-

Mikkel Svold (08:45): Because we want to save the fat or-

Michael Sievers (08:47): Exactly, you want to save the fat. Every fat I’m saving in this milk production I can use for butter. So you could say it’s even a combination of several systems. Here, we talk about fat savings that are possible of 0.15, so even less than for the butter. Then people say, “But that’s rubbish. Why should I invest?” If you produce 250,000 liters of this type of milk a day and that’s small, I can really tell you that’s small, with the actual fat price we are having of around over six euros, you would save 90,000 euros a year. So you can see these tiny small adjustments very often if… Even my customer sometimes if I’m saying, “You can save 0.01 fat,” and then I get sometimes back, “So what?” But when I then discuss the mass balance, they are really impressed.

Mikkel Svold (09:50): Now, when we are talking small and medium-sized dairies, you say that the size of a medium-sized dairy has increased because the large dairies, they are super massive. So now the medium-sized are actually also pretty big.

Michael Sievers (10:06): Yeah, exactly. Yeah.

Mikkel Svold (10:09): Because one of my thoughts initially was well a small or medium-sized dairy, would they even be able to adjust their production like that? Would they be able to just take that fat and then send it in another direction and crank up the butter production overnight? How does that actually work in real life at a, let’s say, small or medium-sized company?

Michael Sievers (10:35): It’s not super precise, the numbers I’m telling you, but just to give you a feeling, this medium-sized dairy, they should invest around, let’s say, 200,000 euros in a sensor and an automation system, let’s say, from Au2mate, can be more, can be less depending on the complexity. Now, you can see 200,000 euros, you save around 90,000 euros a year. Already you could say after the second year, it’s a cash cow for you. That’s the reason why small dairies would all are now what we in former times called small dairies that are now really medium-sized, you could say, that’s the logic behind… If I have a return on investment below 24 months and in butter sometimes we talk about return on investment of less than one year, so that’s the logic why also medium-sized theories are now saying, “Oh, yeah, let’s go for it.” In this 30 years I’m doing this, it started with a handful of giants. Now really spread out to over the entire dairy industry. For sure, small farmhouse dairies are not able to use this, but 500,000 liters a day and we talk about business.

Mikkel Svold (12:09): Yeah. I’m just curious, where would you get most of your fat? I know you can do the same with protein, so those two being the main components that are really valuable because basically a lot of the rest is water, but-

Michael Sievers (12:29): You’re right.

Mikkel Svold (12:30): … where do you go? Where’s the better product? Is it butter? Is that the top shelf product price-wise where you get the most bang for your buck?

Michael Sievers (12:39): That’s also a huge benefit of a fully automated system because I cannot say in general put everything into butter. You always have to see where do I get more money for my fat and there can be business situations where you get the most money for fat, per kilo fat if I put it into cheese, then it would be stupid to produce more butter. So the dairies are also now able to react on market trends. Right now, we have the highest butter price since ever, so everybody is saying-

Mikkel Svold (13:16): Yeah, I can recognize that. I’ve got three children.

Michael Sievers (13:19): Right now, everybody is saying, “Let’s save fat in the cheese production and let’s put everything into the butter.” For a small dairy, let’s say, only producing consumer milk, they only have the chance to go as close as possible, sorry, I don’t know the Danish specs, but 1.5% fat, then they save fat. For them, it’s very easy. With the high butter prices, all butter producers in Denmark are keen to buy the cream. So for this small dairy, he can still produce you could say his consumer milk and sell all his cream to the big butter producers. For the small cheese makers, it’s the same.

Mikkel Svold (14:02): Yeah. So the point being here is that you don’t actually need to be able to adjust your own production. You can also just skim away the fat as precisely as you can and then sell it on to a larger dairy who can then use it for whatever product they want.

Michael Sievers (14:19): Exactly. That’s you could say that’s the network of the industry. Some need more fat, they buy it from other dairies. Some don’t need protein so they extract the protein, send it to other dairies and vice versa. That’s how the entire ecosystem in dairy is working.

Mikkel Svold (14:41): I remember Jesper Kjeldahl who was in the first episode here, he talked about… I can’t remember if it was actually on the interview or maybe it was before, but he said anyway that it makes sense for a Danish dairy to move in trucks, move milk or cream that grow also be hundreds of kilometers away. So, send it in a truck and then send it away to some other dairy that produces something else that maybe it may be the butter production facility or it may be a cheese facility or something like that just because of that, something that seems a little bit insignificant that 0.15% change, well, both in the fat that is in the milk, but also it could also be in price.

Michael Sievers (15:31): Exactly. That’s a very good example. I know a cheese plant in Denmark that is buying the cream in Germany because it’s cheaper than Danish cream, because the cream price is so high due to a very good brand of your biggest producer that he cannot afford to buy, you could say buy cream. From this company, he’s buying it from Northern Germany because it’s cheaper. Also, in Germany, they buy and receive milk, for example, from Poland. It’s a free market. A lot of people think everything is full of subvention. These times where we had milk lakes, butter mountains are completely over. This has European Union stopped completely. We are fighting on a world market against people from New Zealand that do this online, inline since decades. We Europeans are always conservative and jump on such a train later, but now on full speed.

Mikkel Svold (16:46): Just to be clear about that, so what you mean in inline is that in Europe and historically, these tests of the fat they’ve been done in a lab. So you would have to go take a sample, bring it back to the lab, physically bring the sample back to the lab, analyze it and then adjust the machine. What we can now do and what is being implemented across dairies are large and small is inline analysis of this type, right? So it’s basically instead of bringing the test to a lab, you would just do the actual test on like inline, in the production line. Can you explain if you’re a small dairy, I wonder what is the next step? How would you go about producing or implementing this type of automation?

Michael Sievers (17:45): Let’s use butter as an example because that’s a very easy nice situation. So you run your butter churn. You take a sample, go to the lab, you get your result. With this result, the operator goes back. I simplify now a lot, so for my dairy colleagues, I apologize, I do it very simple. They open a valve and put tiny little bit more water into the butter, then they take another sample, go back to the lab. So for a small dairy, in principle, everything is already existing but driven from people. So for the dairy, it’s just they can flange in this inline sensor that provides you every three seconds moisture fat, and so it’s non-fat. My problem so far was you could say I’m only providing the sensor the analytics and only with a sensor, only by measuring you don’t improve a process. That’s why this in partnership with Au2mate is so important because they use our results and then it’s very easy. There is a PLC, and instead of the operator adjusting it-

Mikkel Svold (19:02): Turning the knobs, yeah.

Michael Sievers (19:05): … the control loop, the PLC is doing this automatically and corrected every three seconds. That’s the huge benefit. You also have to be aware… Sorry, now I need to talk a little bit about statistics.

Mikkel Svold (19:23): Go ahead.

Michael Sievers (19:25): When you take such a butter sample and you get the result, then you have a bell curve distribution. So let’s say the result is 15.70, then you only can say within plus minus 0.1, you have one sigma. Plus minus 0.2, so there is uncertainty. But if you do it every three seconds, then it’s poor statistics. You can level out this random error and react in real times and this is how you could say this huge improvement possibilities, I call it huge, you call it decimals, are possible when you move from the lab to the online. Maybe another thing that is also for small dairy super important is you have heard now in the interview, I’m a really proud dairy engineer. So, what do you think is my son doing?

Mikkel Svold (20:27): He’s probably in dairy.

Michael Sievers (20:29): No, not at all.

Mikkel Svold (20:31): He’s a programmer.

Michael Sievers (20:32): Yeah. He’s a software developer and this is a huge, huge problem of the dairy industry worldwide, not only in Denmark, not only in Germany, it’s the same in United States and you will laugh even in India, highly qualified people want to do everything, but not working in a dairy, three shifts over the weekend, it’s wet and it smells. So the dairy industry today, the only chance to keep this high quality and improve. They need to automate because the young people do everything else influencing whatever, but for sure not working in a dairy.

Mikkel Svold (21:15): I think that’s true for all or for many production facilities that they’re actually… It’s really hard to get people and especially when you have a production like the dairy production where you don’t… Yeah, it is not something that you stop at five o’clock and then go home and then turn on the machines again next morning at 7:00. It’s through the night as well.

Michael Sievers (21:39): Yeah, especially when you have a spray dryer producing milk powder, you have to run it 24/7, because otherwise, starting up and shutting down a spray dryer is simply not possible to do this every day. So yeah, it’s a 24/7 job in dairy.

Mikkel Svold (21:59): I think the arguments for implementing automation into the dairy industry, they are pretty strong and they’re also fairly easy to understand. I think it’s just there’s something… You have to have the courage to take that next step. What are some of the uncertainties that you meet out there when you talk to dairies that are, I don’t know, on the brink of automating because there must also be some kind of fear of losing control or something like that?

Michael Sievers (22:30): I would say there are two. One thing is I don’t know how you phrase it in Denmark, but in Germany from experienced guys like me, you sometimes hear, “Why should I change? I have done this since 30 years. Everything is fine.” Dairy is extremely conservative. You could say they don’t want to change things so quick. That’s one thing. Then let’s be honest, the technology has improved. We launched in the year 2000 already our first inline sensors, but seen from today, very complicated, a lot of calibration work. PLC connection was not really set up. You had four to 20 milliamps. So a lot of these dairies have also, if I say, burned their fingers in the past with such technologies. This is still also in the mindset. It’s super complicated, but today we use artificial neural networks for calibration.

(23:47): The communication is not for the 20 milliamps, it’s OPC UA, all this cloud computing stuff. Things have changed. But I think you made the same experience, when you tried something sometime ago, even if things have developed, you stay critical because… So these are the two things, the old foxes that don’t want to change and you could say it’s bad experience from the past. I had one customer that was still saying, “Yeah. I don’t want to try it because I tried it 25 years ago.” I said, “Well, how was your computer looking 25 years ago?” So this is a little bit the mentality we have to overcome.

Mikkel Svold (24:36): Yeah. If you’re sitting out there and you want to start implementing, what do you do? You want to implement automation more in your facility. What happens then?

Michael Sievers (24:47): Let’s say in this case, my job is the easiest one because I provide the sensor. The partnership with Au2mate makes my life now also much more complicated because when… Sorry-

Mikkel Svold (25:03): I’ll let them know.

Michael Sievers (25:04): … off the opposite, makes it much more easy.

Mikkel Svold (25:08): Okay.

Michael Sievers (25:09): Much more easy, I’m really sorry, because when I, before this partnership, went to a dairy and explained all this and then they said, “Yeah. Michael, fine.” So you measure every three seconds, but only by measuring, nothing is changing. Who can do the automation? Because that’s the real game changer and this makes my life complicated. Now, the opposite with Au2mate, if I can explain in more details what we just discussed, then the next question is, “But who is doing the automation and can you guarantee the performance of the process and not how accurate you measure? I couldn’t care less how accurate you measure. My process needs to be spot on.”

(25:57): Now I have the opportunity to onboard Au2mate. When I have set you could say the basic seats, I invite Au2mate and we visit the customer together. They are the experts in the IT infrastructure in PLC setup, and now we can offer the customer a complete solution. Vice versa, it is very often that Au2mate is talking about automation and the customer wants inline sensors. Everybody wants inline sensors, password, Industry 4.0. Now, Au2mate has the chance to say, “Okay. We are the partner of the worldwide market leaders for sensors and together we can guarantee you a process performance.” This takes away a lot of the, you could say, fear of losing control because either they know us very good or they know Au2mate very good.

Mikkel Svold (26:58): Yeah. Yeah. Okay. I think now looking ahead, I’m looking at the time as well, our time is nearly up for this interview, but just quickly before we round off, what do you see becoming the next step in automation in the dairy industry? What’s going to happen? What’s in the future right now?

Michael Sievers (27:18): I use once more the butter setup because that’s very easy. The startup of a butter churn today is still very manual. You have a basic moisture content and then you start dosing automatically to do the fine adjustment, but you still need the operator. So the trend I’m seeing is not only to measure the butter at the outlet, but also the cream inlet, a lot of other sensors, and then as an add-on to the automation, using artificial intelligence by creating a digital twin, achieving the same result without any experienced operator, but that’s more a topic you have to ask then my colleagues from Au2mate. For me, once more, I provide the sensors and that’s the trend I’m seeing. Today, we measure the finished blend, you could say, and optimize. In the future, I see that you measure everywhere inline in a dairy, put this into artificial intelligence system and implement Industry 4.0. I think you and I can agree today we are at Industry 3.0, not 4.0, but that’s the trend I’m seeing at the moment.

Mikkel Svold (28:53): Yeah. Yeah. Michael Sievers, thank you so much for joining in this interview. It was really interesting. I think for anyone sitting out there in the dairies, small, medium, or even the large ones, the enterprises, I think it’s really about looking after this automation area right now, because otherwise, it seems like you will have to catch up later and that can sometimes be super hard.

Michael Sievers (29:25): Exactly. It’s a competitive world, and if you don’t do the step now, there is a huge risk that you are not profitable, and then in a world of sharks, there is a risk to get acquired from somebody else.

Mikkel Svold (29:45): Absolutely. Thank you so much. To the listeners out there, thank you so much for joining us as well. If you have any questions or comments or anything, please leave us a review where you post that or you can also send it to podcast@au2mate.com, and that’s Au2mate with a two, so it’s A-U two, with the number two, Au2mate. Yeah. So, podcast@au2mate.com. I think that’s it for now, and I just want to say thank you so much for listening.

Michael Sievers (30:13): Thanks, Mikkel. It was a pleasure talking with you.

Mikkel Svold (30:16): Thank you.

00:00 Introduction to Michael Sievers, Global Key Account Manager 00:38 Overview of FOSS and Inline Measurement Technology 01:53 Importance of Production Optimization in Dairy 03:43 Impact of Reduced Dairy Sites on Efficiency 04:33 Real-time Process Monitoring for Medium-sized Dairies 05:18 Practical Examples: Butter and Milk Fat Optimization 10:06 Financial Benefits of Automation for Smaller Dairies 12:30 Supply Chain Dynamics: Fat and Protein Utilization 14:19 Selling Excess Cream to Larger Dairies 17:45 Steps to Implement Inline Automation Systems 22:30 Overcoming Resistance to Technological Change in Dairies 27:18 Future Trends: AI and Inline Measurement Integration