Scheduling Jewelry and Precious Metal Manufacturing Around Casting and Setters

A jewelry plant does not lose time on the bench, it loses it waiting for the next casting cycle and the one setter who can do the work. EDGEBIC by User Solutions models burnout and casting as real batch cycles, gates stone setting on certified setters with hours, and schedules the whole program backward from the date the retail order has to leave the building.

Wax growing, polishing, and packing are all schedulable in the ordinary way. Casting and setting are where a jewelry manufacturer loses weeks, and they are usually the two resources whose real capacity nobody has written down.

Burnout and Casting Are Cycles, Not Piece Rates

A twelve-hour burnout ramp costs twelve hours of kiln whether the chamber holds four flasks or twenty. That is the single most important scheduling fact in a casting shop, and most plans lose it by treating burnout as a per-flask time.

Model the kiln as a work center whose step time is the cycle. A half-empty burnout then appears in the plan as the same twelve hours a full one costs, which puts the loss where a planner can see it a week ahead. Once it is visible, filling the kiln stops being a 4 PM decision and becomes a scheduling one.

Burnout also runs through the night. Ramp and soak do not pause at the end of the shift, so carry the cycle as elapsed time and let the schedule place the pour on the first crew that is actually present. The general principle is covered in finite vs infinite capacity scheduling.

Metal Families and the Cost of a Purge

Two trees in the same alloy can follow each other with nothing between them but flask handling. A jump from 14k yellow to platinum costs a crucible change and a purge, and getting it wrong contaminates a pour you cannot recover.

Build a setup matrix on the casting work center. Rather than mapping every SKU against every other, group your work into five or six metal families (14k yellow, 14k white, 18k yellow, sterling, platinum, palladium) and record family-to-family transition hours. Six families is thirty-six cells, which is a table a scheduler maintains in an afternoon rather than a project nobody finishes. See how setup families reduce scheduling complexity and sequence-dependent setup times.

With real numbers in the matrix, the sequence optimizer groups compatible pours. The multi-run layer is guaranteed never to return a schedule worse than the baseline it started from, so running it carries no downside: compare total purges and late-job count, then accept only if it wins. For how the optimizer reports its own confidence, see how a proven optimality gap builds trust.

Setting Is a Person, Not a Bench

Stone setting is the operation that most often decides a jewelry manufacturer's real capacity, and it is the one least likely to appear in a capacity model. A shop with eight benches may have three setters who can do pave and one who can do channel work in platinum.

Record each setter's skills, put them on the shift roster, and require the matching skill on the setting operation. The schedule then books the job only where a qualified setter genuinely has hours, and a pave-heavy program does not get promised against a week where the pave setters are already committed. See how operator skills constrain a schedule and how the shift operator roster works.

The same mechanism gives you a planning tool you did not have before. Run the program with one more certified pave setter on the roster and compare the finish dates. That comparison is what turns cross-training from a good intention into a budgeted decision.

Pools of Benches, Lathes, and Tanks

Polishing lathes, tumblers, and plating tanks are usually interchangeable, and hard-naming one on the routing is how a shop ends up with a queue at station three and idle time at station five.

Put the interchangeable machines into a work center group and bind the step to the group instead of a named machine. At schedule time the engine expands the group, compares each member's projected finish against live load, and places the lot on whichever member gets it done first, carrying that member's own effective run time. If one lathe is slower on heavier castings, record that as its efficiency factor and the pool accounts for it. Lots already running stay where they are, and a lot that genuinely must go on a specific machine can be pinned so the pin survives every reschedule. See how a work center group shops a pool of machines.

Backward From the Retail Date

Jewelry ships to a season. Set the required ship date on the job and schedule backward so each operation lands as late as it can and still make the truck, which also tells you the latest honest release date for wax growing. That number is what a merchandising team needs when a program is still being finalized. See forward vs backward scheduling.

A Worked Program

A 600-piece 14k yellow pendant order with a fixed retail delivery date.

The burnout was filled with a second 14k tree rather than run at a third full, which saved a whole cycle later in the week. Casting needed no purge because the prior pour shared the metal family. Setting was placed against a specific setter's available hours, not against an assumption that any bench would do.

When a Rush Lands

A rush platinum order arriving mid-week is not a crisis, it is a reschedule with a visible cost. Enter it with its real date and let it compete for kiln, casting, and setter hours, then look at what it displaced and by how much. Because completed operations are never moved by a reschedule, everything already cast and polished stays exactly where the record says it went while only the unstarted work moves.

Heritage in High-Value, Constrained Manufacturing

User Solutions has built finite capacity scheduling since 1991, more than 35 years, for operations where a scarce shared resource and a scarce skill set the pace: US Navy, GE, BAE Systems, and Cummins across 33 locations. The lineage behind EDGEBIC, including the RMDB heritage, drove GE Railcar on-time delivery from 30 percent to 90 percent in an operation defined by shared constrained resources and long cycles. A jewelry plant losing weeks to half-empty burnouts and an overbooked setting bench is that same problem in a far more expensive material.

Where to Start

Enter your kiln, casting station, polishing pool, setting benches, and plating line as work centers with real calendars. Group your alloys into five or six metal families and record the transition hours. Record your setters' skills and put them on the roster. Then schedule backward from your season dates and see which week runs out first.

For fundamentals, what is production scheduling covers the basics. A neighboring sector with the same batch-and-skill shape is foundries and casting lot scheduling. The industry fit guide maps the rest and EDGEBIC is the product hub. Want to know what your real setting capacity is in week 42? Contact US for a demo.