Anyone hearing this for the first time rolls their eyes. Video games. University studies. Logistics. They don’t seem to fit together. Here is the proof of the opposite.
At its core, Factorio is a production and logistics simulation that never stops getting more complex. The player builds factories, optimizes material flows, fights bottlenecks, and scales capacity. Eventually, you find yourself sitting there at 2 a.m. realizing you are currently practicing operations management.
That is exactly what I used. Not as a gimmick, not to lighten things up. As a real learning lab. What came out of it surprised even me, because the students and I took completely different paths. With different results.
The factory was not optimized.
 |
Starting point: it runs, but yields very little. |
t the start, the students were given a running factory. That sounds fair. It was, but only at first glance. I describe the scenario in this post.
The factory produced. It ran. It wasn’t broken. But it was set up in a way that it just barely covered costs. Break-even. Nothing more. A factory that keeps itself alive but generates no profit.
This is no coincidence and no error in the scenario; that is exactly the task. In practice, as an employee or consultant, you are almost never confronted with a total ruin. Usually, things "sort of" work, otherwise the company would be bankrupt. And that is exactly the problem, because "just enough" is not a business model—at least not long-term.
The students get to know this situation. Not from a textbook. They experience it as a concrete, tangible system that they have to analyze, understand, and improve.
The mission was clear. The solution was not.
Optimize the factory. Maximize output. The target products are Utility Science Packs. Simply throwing more resources at it doesn't work. Factorio doesn't punish waste per se, but waste is against our rules. Conveyor belts have capacities and machines have cycle times. Anyone who ignores this builds expensive setups that still produce poorly.
So, the students analyzed and understood the factory before they started planning. Which machine becomes the bottleneck? Where is capacity lacking? What is the theoretical maximum output given the resources? Read the students' report here.
The students tore everything down.
 |
Tabula rasa—and then a fresh start. |
There was no step-by-step remodeling. No repairing. No optimization of what already existed. The old factory was completely gone within a very short time. Thanks to construction robots.
What followed wasn't a gut feeling; it was a plan. The analysis had shown what the optimal factory should look like. This became a blueprint—an exact digital layout that the construction robots implemented piece by piece. Automatically. Precisely. Fast.
The result is a factory built from the ground up for maximum efficiency. There are no legacy issues, no compromises, and no "grown" structures that you have to drag along. Clean. Maximum. Done.
In terms of pure production volume, this was the best factory in the room. Only: output isn't everything.
I didn't tear anything down.
That was a conscious decision. I left the existing factory standing and started improving it. These were small interventions. A belt here, an extra machine there, clearing a bottleneck, rerouting a material flow.
Kaizen: For those unfamiliar with the term, it means continuous improvement in small steps. No grand design, no reboot. It simply means consistently working on what is already there.
That sounds unspectacular. And it is. At least from the outside. From the inside, it looks different. Every small change has to fit into the running system. You can't just tear things down because everything is still operating. You have to understand what you are touching before you touch it. It’s slower. It’s more exhausting.
But step by step, the factory got better. It was never perfect, but it was measurably better after every iteration. One more step. Then another. Until the output almost reached the maximum. The last four save states are shown below. Every screenshot is a step forward—no revolution, just Kaizen.
 |  |  |  |
 |  |  |  |
Who wins? It depends.
By output: the students. Clearly. More output and built faster. If that is the key metric, there is no debate.
A factory is not an end in itself. It’s supposed to make money. And that’s where it starts to get interesting.
For the evaluation, the game data is extracted from Factorio via a mod and assessed economically in a real model (Excel sheet). We look at revenue, resource usage, absolute results, and net profit margin. These are the numbers used to evaluate a real company—not just unit counts.
And here the picture flips. The students operated profitably, no doubt about it. But their resource usage was higher and their margin smaller. If you build a factory from scratch, you have to pay for it—in the form of material costs, construction costs, and all other expenses required for the restart.
The iterative approach was more cost-effective. What is already standing doesn't need to be rebuilt. While you optimize, the parts that are running generate income. The net profit margin was higher in the end—despite the lower maximum output.
Produce more, earn less. Produce less, earn more. That’s not a paradox. That’s business administration.
Profitable. And still a problem.
My Kaizen factory earns more. But: production machines are placed wherever there was room. Belts take detours because the direct path was already taken. Machines stay where they are—not because it makes sense, but because they never had to move. This is organic growth. Every change on its own made sense. The overall picture is still hard to see through. The Kaizen factory needs someone to clean it up.
The student factory, on the other hand, is organized and the processes are more transparent. But it isn't fully optimized yet. Costs need to be pushed down to increase the net profit margin.
Both approaches need further development, whether through an external consultant or at least the honest will to change things that actually already work.
Both are real. I have experienced both. On one hand, there are companies that set everything up completely new—it was expensive and disruptive, but everything was clean afterward. And I have seen companies where you come in as a consultant and find structures that have grown over twenty years, where you first have to bring order before anything can improve.
Neither path is wrong. But neither is free.
Both—that is the answer.
That’s unsatisfying if you expect a clear winner. But it’s honest. And it matches the numbers. Here are the students' findings: “When Production Fails: A Logistics Problem in Disguise.”
Two cost factors are still missing: First, the actual remodeling costs. The students tore down and threw away a significant amount of material. I shifted things back and forth, remodeled, and touched things multiple times. This doesn't show up in the current calculation yet.
Second, the work-in-process (WIP). A running factory always has material in the pipeline. Supplies, semi-finished products, buffers. Depending on the approach, there is more or less of it. That too is capital. That too must flow into the return calculation.
The conclusion is: if you want to optimize a factory, you first have to know which strategy you want to follow. One answer is throughput. Profit margin is another. Scalability is yet another. The actual engineering task is to maximize all three simultaneously.
You don’t learn that from a textbook. You learn it when you’re right in the middle of it, making decisions and looking at the numbers afterward.
That’s what Factorio is for. Not as a game. As a learning lab.
Weiterlesen
Results of the students work
Jeffrey Olsen Neal, When Production Fails: A Logistics Problem in Disguise, März 26, 2026
Fabian Kleinke, Jan Grünfelder, Jeffrey Olsen Neal (2026) A Factorio Case Study in Industrial Efficiency, Tech-Report, Hochschule Kaiserslautern, Campus Pirmasens, GermanyDeveolopment
Mapping Real World to Factorio and vice versa (German), Februar 28, 2026
Factorio Logistics Controlling & Analysis (English), Januar 30, 2026
Factorio Logistik-Controlling & Analyse (Deutsch), Januar 28, 2026
Accounting Logic Must Not Follow Physics Too Closely, Januar 22, 2026
Using Factorio as a Logistics Simulation, Januar 05, 2026
Transparency Note
Blog: , Seite:
Version: 1.4 April 2025, Kontakt: E-Mail Martin Wölker
Pirmasens, Germany, 2018-,
ausgelesen am: , Licence
CC BY
Kommentare
Kommentar veröffentlichen