Milk Run in the factory: what a tugger train is and when it pays off
What is a milk run in the day-to-day practice of a production hall? It is a fixed delivery loop run by a tugger train along a set route, to the same points, at a fixed time interval. We do not drive "on call" — we drive to the rhythm of the process. A well-calculated system makes it possible to replace 3–4 forklifts with a single train. A poorly calculated one merely shifts the chaos from forklifts to a larger means of transport.
What is a milk run — in 60 seconds
The name comes from the milkman's round: the same route every morning, full bottles delivered to the same doors and empty ones collected on the same trip. A milk run in a factory works in exactly the same way. A tugger train — a tractor with trailers attached — travels around the hall on a planned loop, delivers material to the stations at the lines and collects empty packaging in the same cycle.
The key difference compared with reactive transport: the train departs on time, not when someone reports a shortage. If deliveries today work on call, there is not yet a process to automate — there is a series of reactions to shortages. A milk run turns that reaction into a planned rhythm.
Forklift vs tugger train — a comparison
The easiest way to understand a milk run is by contrast with a conventional forklift. These are two different philosophies of line supply: reacting to a shortage versus a planned cycle.
| Criterion | Forklift | Tugger train (milk run) |
|---|---|---|
| Route | Variable, "on call" | Fixed loop, fixed schedule |
| Load | Uniform, one type of load carrier | Mixed — many parts for many stations |
| Frequency | When someone calls | Every 15 / 20 / 30 / 60 minutes |
| Empty packaging | Separate transport | Collected in the same cycle |
| Safety | Highly variable traffic | Low speed, fixed route |
| Replacement ratio | 1 forklift = 1 trip | 1 train = 3–4 forklifts |
| Buffer at the line | Grows "just in case" | Shrinks to a calculated level |
| Working logic | Reacting to a shortage | Planned rhythm |
The "1 train = 3–4 forklifts" effect does not come from engine power. It comes from the fact that a single run serves many stations and many part numbers at once, instead of carrying one type of load carrier to one place. The train's trailers can be built as flow racks from the CS system, or you can choose ready-made CS system trolleys to suit a specific load carrier and pick-up height.
Logistics takt and loop cycle — where opinion ends
A milk run is not an "idea for an improvement". It is mathematics. Two numbers decide whether the loop has any right to work at all: the logistics takt and the cycle time.
Logistics takt — how often the line needs a delivery
Logistics takt [min] = container capacity [pcs] / consumption [pcs/min]. Example: a 600×400 KLT container holds 20 pieces, and the line consumes 1 piece per minute. The logistics takt is 20 minutes — every 20 minutes the line needs a new container. This is the clock to which the loop must adapt.
Loop cycle time — how long a round really takes
Cycle time = loading at the supermarket + travel + station servicing + safety buffer. Let us take a real loop: loading 5 min, travelling 800 m at 100 m/min takes 8 min, servicing 6 stations at 1.5 min each takes 9 min, buffer 3 min. A total of 25 minutes. And this is where a decision arises: the logistics takt is 20 minutes and the cycle is 25 minutes — the system is inadequate. A 25-minute loop will not meet a 20-minute requirement. There is no room here for "the operator will try harder" — either we shorten the route or we add a second train.
Hence the golden rule: the loop interval should be a divisor of an hour — 15, 20, 30 or 60 minutes. A loop every 23 minutes may be mathematically correct, but it is operationally difficult to maintain and scale.
Stop points, the "last metre" and the role of the Water Spider
A loop is not just a route — it is a set of stop points where material is exchanged. At each stop, one rule applies: "one in, one out" — the operator sets down a full load carrier and takes away an empty one. That is why every station must have space for a full and an empty container. Without this, empty packaging starts to pile up at the line and the loop falls apart.
The train delivers material to the zone — but not always directly to the workstation. This "last metre" — from the pick-up platform to a specific assembly cell — is handled by the Water Spider (Mizusumashi). This is not a helper for carrying things. It is a role for the most experienced operators, working to its own standard: a fixed sequence of workstations, a measured round time, clear logic for empties and a defined escalation point. Without this standard, the Water Spider turns into a firefighter putting out fires instead of systematically preventing them.
Nor can the train operator be a warehouse worker, picker and firefighter all at once. If they collect orders or search for material during the run, the cycle ceases to exist. Full load carriers must be waiting in the supermarket before departure, arranged in the order of the stops.
When a milk run does NOT make sense
A milk run is not a cure-all. There are situations in which forcing it through will only make the problem bigger.
- Too small a scale. If the line needs a delivery once every few hours and there are only a handful of part numbers, a fixed train will run empty. In this case, manual transport along a simple route is better.
- Process chaos. If deliveries work "on call", with no fixed set-down points, no markings and no 5S order — there is not yet a process to put into a loop. Order and a standard come first, and only then a milk run.
- Uncalculated geometry. Routes that are too narrow, loads not foreseen in the design, turning radii that a tractor with trailers cannot manage — these are risks that must be closed before the first train sets off. CAD almost always lies; the real width of routes is measured on the shop floor, with pallets and people.
The order of maturity applies: chaos → order → manual milk run → tugger train → AGV. Levels must not be skipped. Automating an unstable loop is the most expensive mistake — an AGV does not fix a milk run, it merely replicates its flaws faster and at greater cost.
Where to start — a PFEP audit
The starting point is not choosing a tractor or browsing a catalogue of trailers. It is data. Before you calculate any loop, you need a PFEP — Plan For Every Part — a table that gives, for each part number, the dimensions of the load carrier, the weight, the pieces per container, the consumption per hour and the delivery point. Without a PFEP, the takt cannot be calculated. Without walking the route physically (Gemba Walk), you do not know the real times. That is a topic for a separate article — the PFEP audit as the foundation of all internal logistics — but without it, talking about a train is premature.
A practical order for getting started:
- Take the layout and the PFEP for the most important part numbers.
- Walk the route physically: real widths, bends, stop points, traffic conflicts.
- Calculate the logistics takt for critical materials.
- Calculate the loop cycle time from real data, not from assumptions.
- Check the supermarket: are full load carriers waiting before departure?
- Choose one pilot loop, not the whole plant.
- Only then talk about the train, the trailers and possibly an AGV.
Frequently asked questions
What is a milk run, in the simplest terms?
It is a fixed delivery loop within a plant: a tugger train travels around the hall along a set route at a fixed interval, delivers material to the stations at the lines and collects empty packaging. Instead of reacting to shortages, it works to a planned rhythm.
How many forklifts does one tugger train replace?
A well-calculated system replaces 3–4 forklifts with a single train, because one run serves many stations and many part numbers at once. The condition is a calculated logistics takt and loop cycle — without these, the figure remains a promise, not a fact.
Does a milk run require an AGV straight away?
No. A milk run starts with a manual loop — an operator with a train on a fixed route. An AGV only makes sense once the manual loop has been running stably and predictably for several weeks. Automation is the last step, not the first.
