High-Mix Scheduling and Changeover Discipline
In a high-mix shop the machines rarely run long enough for their rated speed to matter. What matters is how often they stop and how long they take to start again. A line that changes over six times a shift spends a large part of the shift not producing, and the scheduling decision that governs the total output is the order in which the products are run, not the speed of the placer. Treating changeover as an unavoidable overhead instead of a design variable is the single largest loss in high-mix scheduling.
Why Setup Dominates in High Mix
Setup includes more than the physical change of feeders and stencil. It includes the time to retrieve and stage the materials, the time to load and verify them, the time to establish the first article, and the time to return the previous materials to storage. Where those activities are counted only as the minutes the machine is stopped, the real cost is understated by a factor that is often two or three.
The ratio of setup to run time is the figure that decides whether a product can be built economically. A product with a 30-minute setup and a 10-minute run is dominated by its setup, and every improvement in the placement rate is irrelevant to it. Where the mix is genuinely high, the productive strategy is to reduce the setup rather than to increase the speed, and to treat the setup as a process with its own standards and its own measurement.
Measuring Changeover Properly
A changeover measurement starts when the last good board of the previous product leaves the machine and ends when the first good board of the next product is confirmed. It is not the time the machine is idle, because the material staging that happens while the machine runs is not free either: it consumes operator time that could otherwise be used elsewhere. Recording both the machine-down time and the total operator time gives a picture that a single figure cannot.
The measurement should be broken into elements: stencil change, feeder change, program load, material verification, first article, and any waiting for a missing item. The waiting element is usually the largest and the most revealing, because it points at a staging or a kitting problem rather than at the changeover itself. Recording the elements per changeover and reviewing them weekly shows which part of the process to attack first, and the shift handover record is a natural place to capture the waiting.
Setup Families and Group Technology
Two products that share a large proportion of their feeders belong to the same setup family and need a shorter changeover than two that share nothing. Grouping products by the feeders they use, by the stencil they need, or by the board width, and scheduling within those families, converts many changeovers from full setups into partial ones. The grouping has to be based on data rather than on intuition, and the data is available from the bill of materials.
The grouping also has to survive the delivery schedule. A family that would save setup time but pushes a delivery date is not a saving. The practical approach is to build a sequence that respects the due dates and then, within each group of products due in the same period, order by shared feeders. The result is usually a reduction in changeover without any change to the delivery commitment, which is the point of the exercise. The work instructions should carry the family definition so that the planner and the line supervisor are working from the same list.
Feeder and Stencil Logistics
The feeder is where most of the changeover time is spent, and the state of the feeder stock decides how much of it is wasted. Feeders that are pre-loaded and labelled for the next product, stored on a cart in the order they will be fitted, reduce the changeover to the time taken to swap and verify. Feeders that are found in a corner and loaded during the changeover turn a ten-minute swap into a forty-minute exercise.
The stencil has the same character. A stencil that is stored in a rack with its part number and revision visible can be retrieved in seconds; one stored in a pile has to be searched, and searching during a changeover is the classic way to lose twenty minutes. The storage and retrieval discipline is worth more than any single piece of equipment, and its cost is a rack and a labelling rule. Where the stencil is not ready at all, the changeover cannot start, which is why the readiness check belongs in the pre-changeover sequence rather than after the machine has stopped.
Batch Size and Its Hidden Costs
A large batch amortises the setup over more units, which is why the instinct is to build big. The costs that the instinct misses are the work in process that the batch creates, the time the product spends waiting, the risk that a defect discovered at the end of the batch affects every unit in it, and the schedule flexibility that is lost while the batch runs. In a high-mix environment those costs often exceed the setup saving.
The economic batch size follows from the setup time and the holding cost, and it is usually much smaller than the batch that a full setup would suggest. Where the setup has been reduced, the batch can be reduced with it, which shortens the lead time and improves the response to a schedule change. The relationship is worth calculating for the top few products rather than in general, because the answer depends on the value of the board and on the delivery pattern.
Line Utilization Versus On-Time Delivery
Utilisation measures how much of the available time the line is producing, and it is easy to improve by running large batches of whatever is available. On-time delivery measures whether the customer gets what was promised when it was promised, and it can be damaged by exactly the same behaviour. Where the two conflict, the delivery date should win, because utilisation is an internal metric and the delivery is not.
The conflict is usually resolvable by looking at the constraint rather than the average. Where the bottleneck machine is fully loaded, utilisation there is worth protecting; where a non-constraint machine is highly utilised, the utilisation is producing inventory rather than output. Measuring utilisation per station, and comparing it with the station’s role in the constraint, avoids the situation where the line is busy and the shipments are late.
Sequencing Rules That Work in Practice
A few simple rules cover most of the benefit. Run products with the same stencil consecutively. Run products with the same board width consecutively, because the rail setting is a physical adjustment that takes time to verify. Run the product with the largest feeder count at a position in the sequence where the feeder cart can be prepared while the previous product runs. Where a product has a short run time, group it with others that share its feeders rather than running it alone.
The rules should be written down and applied by the planner, with exceptions recorded and their reasons stated. Where an exception is made for a delivery reason, the cost should be visible so that the decision is informed rather than automatic. Over time the recorded exceptions show which products are consistently difficult to schedule, and those are the candidates for a dedicated setup, a dedicated line or a design change. The maintenance schedule should also be arranged so that planned service does not land in the middle of a family run.
What to Record and Review
The record should carry, for each changeover, the product before and after, the total time, the time in each element, the waiting time and its cause, and the first article result. With a few dozen entries, the pattern is usually obvious: one product that takes twice as long as the rest, one element that dominates, or one material that is repeatedly late. None of that is visible from the output figure alone.
The review should be weekly and should produce one change to the process rather than a list. Reducing the dominant element by a defined amount and re-measuring is a controlled improvement; changing five things at once produces a result that cannot be attributed. Where the change works, it should be written into the standard changeover procedure so that it survives the next shift change, and the first pass yield for the affected products should be watched to confirm that the shorter setup has not cost quality.
FAQ
Is it worth pre-loading feeders for the next product? Almost always, where the feeder stock allows it. The pre-loading happens while the machine runs, so it converts machine-down time into operator time. The feeder count needed to do it is a capital item, and the payback is usually measured in weeks for a high-mix line.
How small can a batch be made? Small enough that the setup cost per unit is acceptable and large enough to cover the material and the first article. Where the setup has been reduced to a feeder swap and a program load, a batch of a few panels can be economic, and small batches make the schedule far more responsive.
Should the same operator always run the same product? Not necessarily, but the changeover procedure should be standardised so that any qualified operator can make the change in the same time. Operator-dependent changeover times indicate that the procedure has undocumented steps, which is a risk on any shift pattern.
