Material flow analysis

A material flow analysis records and evaluates the transport, storage and handling operations in a plant: which material moves between which places, in what quantity, along which routes and how often. It makes visible what nobody sees in one piece during daily operation: detours, crossings, overloaded aisles, areas that sit too far apart.

Two quantities carry it. Transport intensity is the frequency, or amount, of transports per unit of time; the distance says how far they travel. Intensity times distance gives the transport effort. Only that makes a layout judgeable: a small amount over a long route can cost more than a large one over a short route. Where mass is counted instead of amount, the term is transport performance, usually in tonne-kilometres.

It is therefore not a snapshot of operations but a statement about their structure: whether the arrangement of areas fits what actually flows between them.

Three things. A layout with the places material moves between: goods receipt, stores, production areas, assembly, dispatch. A route network, meaning the aisles and lanes that can actually be driven. And a transport matrix: one row per source, sink and amount for a period.

Optional but worth it: material, load carrier, mode of transport and the capacity per load carrier. Only these turn amounts into trips. And it is trips that occupy people, forklifts and aisle width, not unit counts.

The route network is where it usually goes wrong. Amounts exist somewhere in almost every plant; the routes rarely do. Leaving them out means calculating in straight lines. A straight line understates every real distance, because it runs through walls, racks and machines.

1. Set the boundaries. Which area, which period, which materials. What matters is not a minimum duration but a representative period: one that shows normal operation, without the shutdown weeks and without the run-up to year-end.

2. Collect the data. Movement data from ERP, production planning, shop floor data collection, warehouse management or tracking systems; in practice several sources are combined. Where none exists, recording by hand over a limited period helps. Its drawback is that it describes the period in which somebody was watching.

3. Capture layout and routes. The floor plan with its places, plus the route network. Both have to share the scale of the real building, or none of the metres below them hold.

4. Evaluate. Put the amounts onto the routes and look at where things pile up. This is the step this tool takes over.

5. Judge and decide. An evaluation is not yet a measure. The step from the picture to the decision, that is moving an area, opening a route or retiming a process, remains engineering work and cannot be calculated.

Methodical foundations for material flow investigations are described, among others, by VDI 2689 "Leitfaden für Materialflussuntersuchungen" (2019-02 edition); for material flow planning, VDI 2498 Part 1 applies as well. Both are guidelines, not standards. The procedure here does not follow them step by step.

Four views are common, and each answers a different question. They do not replace one another.

The heat map answers „where“. It colours places and routes by load and shows at a glance where handling and traffic concentrate.

The spaghetti diagram answers „how“. Every transport relation becomes a line of its own; tangles and crossings are not a metaphor there but the picture itself.

The Sankey diagram answers „how much“. The width of a band is the amount, which makes main and side flows comparable without lining up numbers.

The route evaluation answers „how far“. It gives the metres actually driven per relation, the basis for any statement about effort.

An unrepresentative period. An analysis over two quiet weeks in summer describes two quiet weeks in summer. The period decides everything that follows, and it is the choice least often justified.

Straight lines instead of a route network. The error is not spread evenly: it hits hardest exactly those relations that have to go around an obstacle, and those are the interesting ones.

Amounts without load carriers. Ten thousand screws and ten thousand engine blocks are the same number and not the same load. Without load carriers and capacity you measure unit counts and call it traffic.

Places that are too coarse. „Production“ as a single point hides every route inside production. The resolution of the places is the ceiling on what the analysis can show at all.

Forgotten empty runs. In a from-to matrix built from ERP or material posting data they are often missing: where no material moves, nothing is posted. In telematics, AGV or vehicle control data they frequently are present. So they have to be derived from those sources, observed or, where such data does not exist, estimated.

A material flow analysis shows structure, not behaviour over time. It answers whether the arrangement fits the flow. It does not answer whether a queue forms at a bottleneck at 9:40, how large a buffer has to be, or how a disruption propagates. That needs a discrete-event simulation.

It is also not a survey and not an approval. The metres it calculates are as accurate as the floor plan they come from, and a result replaces neither a technical review nor a decision.

That limit is not one to play down. An analysis that promises more than it can carry gets discarded at the first objection from the shop floor, together with everything about it that was right.