Dairy data expert: How to use production data in a dairy facility – and actually trust it

Dairy data expert: How to use production data in a dairy facility – and actually trust it

In most production facilities, collecting data has never been easier. Sensors measure temperature, flow, pH, and pressure around the clock. And yet, for many dairy producers, the gap between collecting data and acting on it remains wide, because there is a step that often gets skipped: validating that the numbers being collected are actually right. In this episode of the Food Tech Podcast, we talk with Henrik Kæmpe, Operations Development Manager at Mammen Mejerierne, the second largest dairy in Denmark. Henrik has spent years building the data infrastructure that allows an organisation to make good decisions at every level, from the operator managing a batch on the floor to the site manager reviewing weekly performance. His perspective is grounded in practice rather than theory. The conversation works through the full stack of production data, from sensors and batch control through KPI dashboards and sales forecasting, before arriving at one of the more counterintuitive realities of dairy manufacturing. What makes Henrik's perspective valuable is that he approaches all of this as a people manager first. The goal, in his framing, is not a perfectly instrumented facility. It is an organisation where the right person has the right data, trusts it, and knows what to do when it changes. In this Episode * How decision loops in a dairy facility are structured, and why the data that matters at operator level differs from the data that matters at management level * Why traditional statistical process control fails in batch cheese production, and what that means for optimisation efforts * Why ownership of a KPI matters more than how it is designed * The first data points Henrik would introduce at any new cheese facility, and why yield sits at the centre of everything Chapters 01:43 How data use is structured across organisational levels 07:21 What production problems catch Henrik's attention and prompt new solutions 11:51 When sales forecasts go wrong and how production responds 15:57 How to ensure KPIs stay relevant to the people using them 24:06 Statistical process control and why batch cheese production makes it hard 29:30 Could dairy ever embrace natural variation and sell it as a feature? 36:00 How pH, salt, protein, and texture interact and complicate standardisation 36:32 The first data point Henrik would introduce to a new facility. About Henrik Kæmpe Henrik Kæmpe is Operations Development Manager at Mammen Mejerierne, the second largest dairy in Denmark with an annual turnover of approximately 270 million euros across three production sites. He has spent many years working with statistical process control and data infrastructure in cheese production, and his work sits at the intersection of operations, technology, and people management. Henrik's focus is on building the organisational and technical structures that allow teams at every level to make better decisions from the data they already collect. This podcast is brought to you by Au2mate [www.au2mate.com⁠]. This podcast is produced by Montanus [www.montanus.co⁠].

From Excel Hell to Automated Production | Ep 7

It starts the same way in almost every plant. A small team, a good product, growing demand. Someone builds an Excel sheet to track orders. Then another for recipes. Then one more for raw materials. Before long, the entire operation runs on spreadsheets, manual counts, and tribal knowledge. In this episode, Erik Søndergaard and Lars Linnet join the FoodTech Podcast to walk through what it actually takes to move a food or beverage production facility from low integration to full digitalization. From walking the factory floor and reading the posters on the walls, to implementing recipe systems, real-time dashboards, and automated job order management, this conversation covers the full journey step by step. What makes this episode especially practical is that Erik and Lars bring two different perspectives. Erik represents the management and financial side, focused on cost, compliance, and customer delivery. Lars brings the engineering view, grounded in what operators need and what the control systems must deliver. Together, they show how these two worlds must connect for digitalization to actually work. 00:59 Introduction to the episode and guests Erik Søndergaard and Lars Linnet 02:41 Establishing ground zero and why mid-range companies hit a glass ceiling 05:05 Walking a real Danish food production plant and identifying opportunities 07:30 What management needs from data versus what operations needs to monitor 10:43 Label validation, recipe systems, and where manual processes create real risk 15:53 What belongs in a MES system: batch IDs, traceability, OEE, and warehouse integration 19:32 The levels of integration from PLC and SCADA up to automated job orders 22:23 Dashboards, KPIs, and presenting data without creating overload 24:41 Change management, operator trust, and why involvement decides success 30:46 Quick win examples and why small steps build momentum for larger transformation Production This podcast is brought to you by Au2mate [https://au2mate.com/], This podcast is produced by Montanus [https://montanus.co⁠].

When empty belts burn money | Ep 6

We all know the image. Bottles racing through the line at full speed. Conveyors packed. Motors humming. It looks efficient. But when you step onto real production floors, the picture is very different. Belts run empty. Pumps throttle instead of slowing down. Motors consume energy even when nothing moves. In this episode, Gregors Geilager from Danfoss Drives joins the Food Tech Podcast to explain why energy efficiency is no longer a side project, but a core competitiveness issue for food and beverage producers. From frequency converters and pump laws to condition-based monitoring and empty conveyors, this conversation is packed with practical insights you can use immediately. If you are responsible for production, utilities, or technical decisions in a dairy or beverage plant, this episode will challenge how you look at motion, energy, and data on your lines. In this episode, you will learn: 1. Why empty conveyors quietly waste more energy than most people think 2. How small speed reductions can cut energy use dramatically 3. What frequency converters really do and why they matter everywhere 4. How pumps, belts, and motors reveal their condition through data 5. Where to start if you want fast payback on energy optimization Episode Content 00:10 The perfect production image versus reality on the factory floor 00:45 Why timing and balance matter more than raw speed 02:09 Energy prices, volatility, and why efficiency decides competitiveness 03:43 What frequency converters are and why modern plants need thousands 06:45 Why flexibility and frequent changeovers demand speed control 08:36 Why tiny inefficiencies matter at high production volumes 09:44 The affinity laws and why pumps are the biggest low-hanging fruit 10:34 How reducing speed by 20 percent cuts energy by half 13:08 Where frequency converters create value beyond simple speed control 17:28 Predictive maintenance using built-in machine learning 18:25 Cavitation explained and how drives detect it early 21:42 How drive data feeds SCADA and maintenance systems 25:19 Why most plants still miss easy energy savings 32:22 Where production managers should start their efficiency journey Production This podcast is brought to you by Au2mate [https://au2mate.com/da/automationsloesninger-mejeri-automatisering-proces-automation-plc/]. This podcast is produced by Montanus [https://montanus.co/].

How to not make a power glitch turn milk into a very expensive problem | Ep 5

What happens to your production when the power flickers, a server reboots at the wrong moment or a firewall rule opens a door you did not know existed? In this episode, Erik Søndergaard joins The Food Tech Podcast to unpack operational resilience in food and beverage production. From power drops and UPS age to backups, segmentation and NIS2, Erik explains how to think about uptime like an insurance policy: decide what an hour of lost production costs, then secure your systems to match that risk. If you run a dairy, brewery or any process plant, you will hear concrete steps to keep lines running and systems ready to restart safely when something goes wrong. In this episode, you will learn: 1. What operational resilience really means on the factory floor 2. Why power disturbances and aging UPS units are still the biggest real-world risks 3. How to use redundancy, backups and restore tests to protect critical servers 4. How network segmentation and OT/IT separation limit the blast radius of an attack 5. Why NIS2 is not just paperwork but a catalog of good uptime practices Episode Content 00:06 What operational resilience means in a digitized production 01:34 Real-world blockers of production and why power is enemy number one 03:38 IT vs OT - why five minutes offline is different in an office than in a cheese vat 05:44 Defining operational resilience as the ability to keep producing and restart safely 09:04 Calculating the cost of downtime and using risk analysis as an insurance model 11:20 Legacy equipment, isolation and why “air gaps” still matter for old systems 13:13 Why security is never “done” and the need for regular hygiene walk-throughs 16:05 The firewall rule that opened everything and what it teaches about everyday shortcuts 20:42 How segmentation limits the blast radius when something does go wrong 22:35 The basics to fix first - UPS age, server redundancy, backups and restore tests 26:23 Thinking in fire doors and zones for OT networks and systems 27:48 Securing vendor remote access without importing new risks 30:53 Clear roles when something breaks and anchoring responsibility at board level 33:31 Treating NIS2 as uptime engineering instead of box-ticking compliance This podcast is brought to you by Au2mate [https://au2mate.com/da/automationsloesninger-mejeri-automatisering-proces-automation-plc/]. This podcast is produced by Montanus [https://montanus.co/].

From gut feeling to guided zero-waste dairy production | Ep 4

AI is everywhere, but turning it into real outcomes in dairy and food processing is where the value is. In this episode, Anna Olsson, co-founder of Intelecy (a no-code platform for industrial AI), cuts through the hype to show what plants can do today: predict failures before they happen, optimize processes in real time, and capture expert know-how so it scales across sites. If you run operations, engineering, maintenance, or production IT, you will get practical steps to start fast, prove ROI, and avoid pilot purgatory. In this episode, you’ll discover: 1. The “learn, act, detect” framework for industrial AI 2. Why data quality and coverage beat big promises 3. How to move from pilots to scaled, maintained models 4. Where predictive maintenance ends and process optimization begins 5. How no-code tools bridge the IT–OT gap and protect operator trust Episode Content 01:57 After ChatGPT – expectations vs industrial reality 03:06 LLMs vs industrial AI and time-series sensor data 04:47 The “learn, act, detect” framework for process optimization 05:26 Predictive maintenance in practice and planning stops instead of reacting 06:40 Predicting future process states and adjusting before quality drifts 10:23 Tacit know-how and “knocking on pumps” vs data-driven models 12:21 Prerequisites for AI: stored sensor data and data quality 15:20 Case: how TINE detects bacterial contamination with AI 17:28 Energy optimization and small savings that add up 24/7 19:37 Why AI projects fail and end up in “pilot purgatory” 21:02 Build vs buy – scaling beyond the first AI model 31:25 Towards Industry 4.0 – closing the loop from prediction to automation This podcast is brought to you by Au2mate [https://au2mate.com/].This podcast is produced by Montanus [https://montanus.co⁠].