How a Schedule Accounts for Scrap and Yield Loss

If a routing loses parts to scrap, you have to start more than you ship, and a schedule accounts for that on the quantity rather than inside the routing. In EDGEBIC by User Solutions, a product carries a Yield percentage in its inventory planning fields, and replenishment planning inflates the suggested build quantity by it, so a stocked part with a known loss is sized for scrap before a planner ever sees the number. For an order you enter by hand, you size the quantity yourself. Either way the scheduler loads every operation against the quantity the order carries, so the plan reserves enough machine time and material to deliver the parts you actually promised. Get that quantity right and the plan stops shipping short; get it wrong and you either run a costly second batch or overproduce.

Yield loss compounds, and it hides

Parts are lost at every operation, to scrap and rework, so fewer good units leave the end of a routing than entered the front. The trap is that yield compounds. Each step passes a fraction, and the fractions multiply, so a routing where every step looks healthy on its own can still leak a fifth of its parts by the finish.

Consider a five-step casting routing with per-step first-pass yields of 96, 94, 92, 98, and 97 percent. Each looks fine. Multiply them together and the combined yield is about 79 percent. Start 500 and roughly 395 good castings emerge. The order was for 500. Nobody set out to lose a hundred parts; the loss hid in the compounding, and it surfaces only when the finished count comes up short.

That is why scrap has to be planned, not absorbed. The place to plan it is the quantity, before capacity is even loaded.

Where the platform does the arithmetic for you

Yield is a product setting, not a routing setting. Open a product, go to Inventory Planning (MTS / MTO), and set Yield as a percentage. Replenishment planning then divides the required quantity by the yield and rounds up, so the suggested build quantity already covers the expected loss. The documented example is exactly this: at 90 percent yield, an order for 100 starts 112. The inflated figure is what appears in the Suggested lens on the inventory calendar and on the MRP worklist, and firming that suggestion creates an order at the larger quantity.

Two boundaries are worth being precise about, because they decide where you still have work to do.

The first is that the inflation applies to a suggestion. An order you type in yourself carries the quantity you typed. The second is that there is no yield field on an individual routing step, and the scheduler does not inflate a quantity as it flows down a routing. It loads each operation against the quantity the order carries, start to finish. So for a make-to-order job, sizing for scrap is a planner decision, made once, on the order quantity.

Size the quantity, then load capacity against it

The arithmetic is the same whichever path you are on. To ship a target quantity through a routing with combined yield below one, you release more than the target: the release equals the ship quantity divided by the combined yield. For the casting example, 500 divided by 0.79 is about 634 units released to deliver 500 good ones.

Once the order carries the larger quantity, the schedule does what it always does: it loads each operation's capacity against the quantity flowing through it. Because that quantity is now 634, not 500, the plan reserves enough machine hours and material to actually produce the parts, including the ones that will scrap. The finish date and the capacity commitment both reflect reality rather than the optimistic case where nothing is lost.

That is the honest, dependable way to get scrap into the plan: put the allowance in the quantity, and let finite capacity load against it. And because more parts flow, the constraint carries more load, which ties straight into capacity vs throughput in manufacturing scheduling: the good throughput is always less than the started quantity.

The number has to be measured, not guessed

The allowance is only as good as the yield you divide by. Guess low and you ship short and run a second batch. Guess high and you overproduce, tying up the constraint on parts you never needed. So the yield figure has to come from measured performance.

That measurement comes off the floor. At the kiosk an operator taps a good count and a scrap count, and a scrap piece can never be recorded without a reason, which is what makes the scrap reports worth reading. Good pieces count toward the job's produced quantity; scrap does not, and is reported separately with its reason. The OEE report turns those punches into a Quality percentage per work center, computed as good divided by good plus scrap plus rework, and it honestly shows "n/a" where nobody has punched anything. Read that alongside the scrap reasons and you have the real loss per operation instead of folklore. As the process improves, the measured figure rises, you lower the product's Yield allowance, and you stop overproducing. Feeding real actuals back into the number is the same discipline behind why 100 percent utilization is a trap: plan against reality, not the ideal case.

A worked example

An order calls for 500 finished blades. The measured combined yield across the five-step routing is 79 percent.

Release exactly 500 and the plan loads capacity for 500, but only about 395 blades survive. The order ships short by roughly 105. Now you run an expedite second pour to make up the difference, pay for the extra setup and material, and deliver late. The shortfall showed up at the worst possible moment, at the end of the routing, with no time to react.

Size the quantity instead. Divide 500 by 0.79 to get 634, and release 634. The plan loads every operation against 634, reserving the machine hours and material for the parts that will scrap along the way. About 500 good blades come out the end, the order ships complete and on the planned date, and there is no surprise second run. Same process, same yield, entirely different outcome, decided by whether the scrap was planned into the quantity or discovered after the fact.

If the blade were a stocked part rather than a one-off customer order, you would not do that division at all: set Yield to 79 percent on the product and every replenishment suggestion for it arrives already sized.

Plan the scrap you know you will make

You cannot schedule away yield loss, but you can stop being surprised by it. The parts you will lose are, on average, a known quantity, and the only question is whether you reserve capacity for them up front or scramble for it later.

Measure your good and scrap counts at the kiosk, turn them into a real yield, set that yield on the product so stocked replenishment sizes itself, and apply the same division by hand on the make-to-order jobs you release. The schedule then reserves the right capacity, promises a date it can hit, and delivers the quantity ordered. Treat scrap as a planned allowance sized from real data, not as an emergency, and the second batch stops being a recurring line item. See how a larger quantity loads your capacity in EDGEBIC, read the wider picture in the scheduling engine guide, and for where scrap fits among your constraints, production bottleneck identification is a useful companion.