Value Stream Mapping for a Small Manufacturer: A Worked Example

What a value stream map shows

A value stream map follows one product family from raw material to dispatch. It puts two flows on a single sheet: the material moving through the shop, and the information that tells each step what to make. Underneath runs a timeline that separates processing time from waiting time. The method comes out of the Toyota Production System and reached most Western readers through "Learning to See" by Mike Rother and John Shook.

For a small manufacturer the useful part is that it needs no software. An A3 sheet, a pencil, a stopwatch and a day on the shop floor will get you a first current-state map.

This piece skips the icon glossary and works through one example from start to finish, so you can repeat the steps on a part of your own.

The example: a bearing flange

The company and its figures are invented for this example. They are chosen so that every calculation works out and you can check each one.

A machining shop with around 40 employees makes a steel bearing flange for a machine-building customer. The part passes four stations. It is sawn, turned and milled, and at the end one operator inspects and packs it at a shared bench.

The starting figures:

• The customer needs 2,400 flanges a month. With 20 working days, that is 120 a day or 600 a week.
• The shop runs one shift with 7.5 hours of available time, which is 27,000 seconds a day.
• The customer sends a monthly forecast and a firm weekly call-off.
• The shop ships once a week.

That gives you the takt time. 27,000 seconds divided by 120 parts is 225 seconds, so on average the customer needs one flange every 225 seconds. What counts as available time, and how takt relates to cycle time, is covered in how to calculate takt time.

Step 1: Walk the current state

Pick a product family, meaning parts that go through roughly the same stations. Here it is the flange. Then walk the route yourself, backwards from dispatch to raw material. Starting at the customer end makes you ask at every station where its material comes from.

At each station, write down what you see, not what the ERP says. Count the parts waiting in front of it. Don't ask anyone for the changeover time. Take it from the last few weeks of shop-floor bookings, or time one yourself. If the quoted figure and the measured one differ, record both.

Then map the information flow. How does the customer's call-off reach the shop floor? In the example, production planning builds a weekly plan out of the ERP every Monday and gives each station its own orders. Each station makes what is on its plan, whether or not the next station needs the parts yet. That is push production, and it explains much of the inventory you are about to count.

Step 2: Fill in the data boxes

Each process icon gets a data box underneath. For the flange it looks like this:

Cycle time is how often a finished part leaves the station. Changeover time runs from the last good part of the old job to the first good part of the new one. Uptime is the share of planned time the machine can actually run once breakdowns and maintenance are taken out. Batch size is the quantity made in one run after a changeover. In the example, the lathe and the milling centre change over to the flange once a week and then make the whole week's demand.

Now compare against takt. Every cycle time is below 225 seconds, with milling closest at 180. No station is too slow on its own, so the time must be going somewhere else.

Step 3: Count inventory and convert it to days

Between the stations you draw triangles for inventory. Divide each count by the daily demand of 120 parts, and a pile of parts turns into waiting time in days. Count the bar stock in front of the saw in parts too.

A part that joins the back of a queue waits until everything in front of it has been used. With 960 parts in front of the milling centre and 120 consumed a day, that is eight days. For a value stream map, this approximation is accurate enough.

Step 4: Calculate lead time and value-added time

The timeline under the map has waiting times on the top line and processing times on the bottom line.

• Lead time is the sum of the waiting times, 10 + 5 + 8 + 3 + 4 = 30 working days. Processing time barely registers at this scale.
• Value-added time is the sum of the cycle times, 40 + 150 + 180 + 60 = 430 seconds, a little over seven minutes.

The value-added ratio divides one by the other. Both need the same unit first. 30 working days at 27,000 seconds each is 810,000 seconds of working time. 430 divided by 810,000 comes to about 0.05 per cent.

So in this example a flange spends six weeks in the building and is worked on for a little over seven minutes. Count calendar days instead and the share gets smaller still, because nights and weekends come on top. No machine is too slow. The six weeks build up between the stations.

Why that waiting time never shows up in anyone's estimate is the subject of how to calculate lead time, which follows a single order through setup, processing, queue and move time.

Step 5: Design the future state

The future state decides where material may sit, how much of it, and what triggers the next job. For the flange there are five changes.

1. Halve changeover time and spend it on smaller batches. On the lathe, moving work to external setup brings the changeover down from 90 to 45 minutes. On the milling centre it drops from 60 to 30. The SMED guide covers how. The time saved goes into smaller batches, two runs of 300 a week instead of one run of 600. Weekly changeover time stays the same, 2 × 45 = 90 minutes on the lathe and 2 × 30 = 60 minutes on the milling centre.
2. A supermarket between saw and lathe. The saw stops working to a weekly plan. It refills a rack capped at 300 parts whenever the lathe takes a batch out. A kanban card on the container is the signal.
3. A FIFO lane between turning and milling. The lane holds at most 300 parts, exactly one batch. When it is full, the lathe does not start another flange batch. The milling centre takes parts in the order they arrive.
4. Inspection beside the milling centre. The inspection bench moves next to the machine. The operator inspects and packs each part while the next one is being milled. At most 60 parts sit between milling and inspection instead of 360.
5. Shorter rhythms with supplier and customer. The shop calls off bar stock weekly instead of monthly, which brings raw material down to 600 parts. Dispatch ships twice a week, leaving 240 parts in finished goods.

The information flow changes too. Production planning now schedules only the lathe. Everything after it flows through the FIFO lane, and the saw in front of it is pulled through the supermarket.

Inventory before and after:

Value-added time stays at 430 seconds, because no machine runs any faster. Lead time still falls from 30 to 12.5 working days. 12.5 days at 27,000 seconds is 337,500 seconds, which lifts the value-added ratio to about 0.13 per cent. There are 2,100 fewer flanges sitting in the building.

That is the plan on paper. Only the shop floor will show whether it holds. Decide who watches the caps on the supermarket and the FIFO lane, and put a date in the diary to map the stream again. For setting a future state without flattering the current one, see modelling the target process.

Mistakes on a first value stream map

Mapping the whole plant at once. Nobody reads a map with twenty product families on it. Start with one that runs often and causes delivery trouble.

Copying figures out of the ERP. The ERP holds planned changeover times and book stock. The map needs what is on the shop floor today.

Opening the future state with a purchase. None of the five changes in the example needs a new machine. Cap the inventory in front of and behind a machine before you buy another one.

Drawing the map and filing it. Without an owner and a date for each change, the future state ends up as a poster on the wall.

The next step

Pick a part that runs often and whose delivery dates give you trouble. On one day, count the inventory between its stations and divide each figure by daily demand. The sum is your lead time in working days, measured rather than estimated.

If you would rather have someone from outside map it with you, that is the scope of our process analysis. We map one of your core processes end to end, show where time and money are lost, and hand over a prioritised list of actions. Fixed scope, fixed duration, fixed price.