Eben van Tonder
2 August 2024
Helpful?
If this approach resonates with you and you want us to assist you in your meat plant, please visit our dedicated page: Consulting for Complex Meat Plant Operations.
Introduction
In 2018, I travelled to Nigeria to develop a central meat plant for the country’s largest retailer. What began as a mission to design a plant layout and train staff quickly became a journey of discovery into both ancient preservation methods and cutting-edge optimisation. From exploring African meat preservation systems that harness naturally occurring skin bacteria and groundnut fibre, to rethinking equipment, layout, and operations, the experience forced me to question everything I knew.
The deeper I went, the clearer it became: the true nature of a modern meat plant lies not only in its physical infrastructure, but in its data, its flow, interpretation, and real-time application. Through the lens of biological information flow and feedback systems, I began to see the carcass no longer as a set of predefined cuts, but as a dynamic mass of potential whose final form should be shaped by live market conditions and creative strategy. This philosophy, informed as much by my early understanding of bacterial ecology as by control theory and process engineering, came full circle when Kristi Berger and I began integrating Artificial Intelligence into the heart of the meat plant.
Together, we’re developing systems that treat carcass data, block tests, and market prices not as static records, but as fuel for an intelligent, evolving system that adapts in real time. This union of biology, chemistry, and AI is not only reshaping how meat is processed but it’s also redefining what a meat plant is.
The Data Centred Meat Plant
My biggest lesson has been in the sanctity of data in the meat plant. In Lagos, I had the opportunity to work with some of the best data scientists on the planet. This transformed my thinking! The leader of our group, a prolific data and systems man himself encouraged me to see the animal carcass as data which must be filtered to create usable information. How is this achieved?
Ensure Data are Captured
The recording and tracking of data are these days easily facilitated by comprehensive inventory systems. These systems provides comprehensive solutions for job costing, stock control, and process traceability.
The most powerful use of the concept of the meat plant as data emerges from a fresh approach to carcass evaluation.
Think Differently About the Carcass
I was blinded by my South African view of the block. The meat market in South Africa is a mature market with a rich history where butchers from the European nations and the UK laid its foundation from the start of the Dutch involvement in the Cape in the 1600s. West Africa challenged everything I thought I knew about the meat plant, including how I view meat.
When I thought about the carcass, I thought about it in terms of the South African way that it is managed and cut. Whether it was the South African way or the European, American, British, New Zealand or Russian way did not matter. The point is that my thinking was linked to a pre-defined set of parameters.
It took nearly two years to realize that no progress could be made without viewing the carcass as nothing more than an input mass that could be transformed in countless ways. With this, I go beyond the basic primals. My new approache led to astonishing results, increasing yield from 75% to 98%, with only 3% being bones. This nearly doubled profitability in terms of gross profit (GP).
The lesson I learned is that data is changed into far more meaningful information if I filter it through a lens that is not restricted to “forms” or outputs or products that are predefined. “Carcass = the cuts I know” is wrong! The thinking change is well summarised when I state: “Carcass = any output of any kind and any cut, limited only by our imagination and feedback mechanisms of NPD and marketing,” as I will discuss later. This change in thinking can not be overstated. It took me years to realise this!
Create Dashboards
The block test data now must be recorded and then represented in a dashboard that shows real-time changes in profitability based on current market prices for all cuts.
I am trying to convey the metaphysical concepts that will underpin the actual new set of dashboards. The traditional static block test must become dynamic, adapting as market conditions change.
What helped me to understand this was when I approached the matter from the perspective of a chemical equation. I viewed the carcass as a chemical reaction that required balancing, where the limiting factor dictated the outcome. This understanding allowed me to apply the same logic across other species, including lamb, goat, pork, and chicken.
Allow me to expand on my thoughts on the chemical reaction and how I apply it to the carcass.
Integrating Chemistry Principles: The Limiting Reagent Analogy
In meat processing, the concept of viewing a carcass as a chemical reaction that requires balancing is to me a powerful analogy for understanding and optimizing production. In chemistry, a reaction is said to be balanced when the number of atoms for each element is equal on both the reactant and product sides of the equation. This principle ensures that mass is conserved throughout the reaction.
A fundamental aspect of this concept is the idea of a limiting reagent. In a chemical reaction, the limiting reagent is the substance that is completely consumed first, stopping the reaction from proceeding further and thus limiting the amount of product that can be formed. For instance, in the reaction:
Here, if we have two moles of hydrogen (H₂) and one mole of oxygen (O₂), hydrogen is the limiting reagent. Once the hydrogen is used up, no more water (H₂O) can be produced, regardless of the amount of oxygen available.
Applying this analogy to meat processing, we can think of the carcass as a collection of potential “reagents” or primals. The process of deboning and the decision of whether to leave meat on the bone can be seen as manipulating these reagents. For instance, if the goal is to produce a certain type of cut (product), the availability of specific parts of the carcass (reagents) will determine how much of that product can be made. If a particular cut of meat is considered a limiting reagent, its availability dictates the maximum amount of a certain product that can be produced.
For example, if a specific tender cut from the loin is highly desired, the quantity of this cut will be limited by the number of loins available from the carcass. This is in the conventional view. But, if we can identify all other parts that can possibly give us the same look as the loin and we can manipulate their tenderness to simulate the loin, we have increased the number of limiting reagents and our products will be more. This simple example can be powerfully extended and applied across the entire carcass!
Designing the right dashboard is now key, where all these options are combined and data are translated into information that gives us all the options at a glance. I have wrestled with the “how” of what I just said for almost 8 months now and I finally have a very clear and simple model. I realise that doing this is as much an art as curing bacon or making great sausages. The art of designing an effective dashboard lies then in representing this complex data in a straightforward, easily interpretable manner. By arranging data such as the overall yield from different cuts, the current market demand, and profitability metrics, the dashboard can help identify which cuts are limiting the potential profit.
Let me say it again and state it a bit differently. The complexity is in the fact that even the reagents (the primals) change themselves depending on the final outcome required in terms of products (final meat cuts). The entire deboning process is altered in one extreme method that we developed and the results are dramatic. Then there is the mix of different cuts from the primals and then there are the final products. It is the clarity I got on how to integrate all this in a very simple way into one dashboard that is easy to maintain and does not require specialised and expensive programs or additional staff which makes it work. The clarity given by the dashboards now enables decision-makers to optimize deboning strategies and adjust product offerings dynamically, ensuring that the entire carcass is utilized efficiently and profitably.
Thus, just as a chemist balances a reaction to maximize product yield, a meat processor can “balance” the carcass to maximize profitability and efficiency, guided by a well-structured, data-driven dashboard. This approach not only enhances operational efficiency but also aligns production with market demands, ultimately driving business success.
Functional Feedback Loops
I had to find a way to integrate the various block tests (a set of completely different ways to debone) and the market requirements. Let’s say we have two blocks or completely unique ways to debone, A and B. If we dobone A, a set of final products will be produced. Same with B. The final products will differ markedly from black A and B. In any set, products will be produced in excess that does not line up with current sales. In plain language, the most profitable cutting method results in an excess of a certain kind that we don’t have a market for.
To manage this I turned to the system of feedback loops and mechanisms designed to accommodate this, which I became familiar with in my 20s. In particular, the one designed for the steam engine.
Here, the concept of functional feedback loops facilitated the implementation of a feedback control system known as the centrifugal governor. This device regulates the engine’s speed and pressure, ensuring stable and efficient operation.
The centrifugal governor was first applied to steam engines by James Watt, a Scottish inventor and mechanical engineer, in the late 18th century. Before Watt’s innovations, steam engines were inefficient and prone to speed fluctuations. His work between 1763 and 1775 significantly improved the efficiency and reliability of steam engines, making them more viable for widespread industrial use.
The centrifugal governor operates on a simple feedback loop principle. It consists of rotating balls connected to a spindle driven by the engine. As the engine’s speed increases, the balls are forced outward by centrifugal force, raising a lever connected to a steam valve. This movement adjusts the valve, reducing steam flow into the engine and consequently slowing it down. Conversely, if the engine’s speed decreases, the balls drop, opening the valve to allow more steam in, thereby increasing the speed.
This understanding of feedback loops is crucial in meat plant operations, where constant adjustments based on real-time data ensure optimal performance and efficiency. A transition from traditional practices to data-centric management will transform the way the plant is managed, resulting in driving innovation’s unlocking greater profitability.
This equation represents a proportional control system where the change in valve position (and thus steam flow) is proportional to the difference between the current speed and the desired reference speed.
James Watt’s introduction of the centrifugal governor around 1788 allowed for precise control over steam pressure, which not only improved engine efficiency but also enhanced safety by preventing dangerous over-speed conditions. The innovation of this feedback mechanism was a significant leap in mechanical engineering and control theory, laying the groundwork for future advancements in automation and industrial machinery.
I was contemplating these feedback mechanisms when I realised that two such essential mechanisms had to be put in place. The first was elucidated by the deboning manager, Mr. Jason (Chike) Nwaozuru. We were receiving pork one afternoon when he explained to me how in his experience customers’ preferences in a specific location could be met by ensuring that the cuts most favoured by them (identified through sales data) are produced in greater volume through creative cutting. What he meant is far from obvious. It is cutting traditional cuts in a way that we cut a portion from another adjacent section which is in less demand or commands a lower RSP (Retail Selling Price). One can find alternative cuts on the carcass which looks like the one in higher demand and with a higher RSP such as the mock loin from the blade. This logic can be extended to the most ultimate way where only 3% of bone is left after beef, sheep and goat deboning as opposed to the approximately 25% that can be expected with regular clocks.
The second dynamic feedback system involves viewing NPD (New Product Development) and sales as integral parts of a dynamic feedback mechanism with the deboning and processing departments. When the deboning manager reaches the limits of creative cutting, NPD and sales must develop products from the remaining cuts that can be sold at the required GP to maximise overall revenue.
Technological Integration and Automation
These revelations coincided with a paradigm shift in the meat processing industry, driven by the integration of advanced technologies. Automation, coupled with Artificial Intelligence (AI) and Machine Learning (ML), is revolutionising processes ranging from predictive maintenance of equipment to quality control and yield optimisation. These technologies enable more efficient processing, packaging, and distribution, while also improving supply chain management. The adoption of mobile auditing solutions further enhances real-time data collection and reporting, ensuring compliance with regulatory standards and improving overall efficiency.
These modern tools can accommodate the innovative thinking discussed, significantly enhancing operational capabilities.
Future Directions
The most powerful application of the concept of the meat plant as data emerges from a new approach to carcass evaluation—especially in the work Kristi Berger and I are developing. Together, we’ve built an AI-driven system that allows us to feed block test data and prevailing market prices into the platform in real time. Each morning, it generates a cutting configuration that reflects the most profitable position for that day. This enables the meat plant to move beyond fixed specs and toward a dynamic, data-responsive model, where AI doesn’t replace human decision-making but elevates it by aligning biological variation with financial optimisation.
Conclusion
What began with a handful of bacteria on a strip of beef skin and a notebook full of yield charts has come full circle into a powerful AI-driven system that reflects and responds to the complexity of the real world. Kristi Berger and I now work daily with systems that absorb live data such as block test results, market prices, retail sales and return, each morning, a dynamically optimised cutting configuration. We no longer speak of fixed specs; we speak of adaptive outputs, shaped by ever-shifting variables and designed to maximise value from every carcass.
Artificial Intelligence in our context is not a replacement for human ingenuity but a partner. It brings to the surface what is often hidden: the limiting reagents in a carcass, the invisible opportunities in underutilised cuts, the overlooked pathways to profit through smart product development and creative deboning. But to unlock this potential, the plant must be reimagined, not as a factory of forms, but as a thinking organism, responding, adapting, and learning.
The plant has become its own feedback system. From African skin cultures to real-time dashboards; from James Watt’s steam governor to recursive product development; from static butchery to dynamic yield balancing, this journey has revealed a simple truth: data is not the future of the meat plant. It is the meat plant.
And now, with AI, the meat plant finally knows how to listen to itself.
References
- StartUs Insights – “Top 10 Food Processing Industry Trends in 2025.” StartUs Insights
- FHA Food & Beverage – “Top Trends in the Meat Sector in 2024.” FHA Food & Beverage
- RizePoint – “2024: The Year of Tech-Driven Quality in Meat & Seafood Processing.” RizePoint
- Mordor Intelligence – “Processed Meat Market Size & Share Analysis.” Mordor Intelligence
- Earthworm Express – “Enhancing Operational Efficiency: Integrating Biological Information Flow Principles in Meat Plant Operations.” Earthworm Express
- Earthworm Express – “Beyond the Central Dogma: Evolving Genomic Insights and Their Relevance to Organizational Strategies.” Earthworm Express
- Earthworm Express – “Insights from Genomics: Relevance and Practical Applications in Meat Science.” Earthworm Express