Production Scheduling for Advanced Ceramics

This guide shows production planners and operations managers how to model multi-stage advanced ceramics production in Schantt — from forming and kiln firing through glazing and finishing — with parallel machines, sequence-dependent thermal-recovery changeovers, and three product classes on divergent routings across five production stages.

This guide follows a fictional composite company built from industry research on advanced ceramics; all names, parameters, and figures are illustrative.

Industry context

Advanced ceramics manufacturing combines high-temperature thermal processing with mechanical pressing, spray coating, and precision finishing. Kiln cycles dominate the timeline: bisque firing at 900–1,100 °C takes 12–18 hours per load, while glaze firing for alumina parts runs 18–30 hours at 1,300–1,500 °C and silicon carbide sintering reaches 1,700–1,900 °C in a nitrogen atmosphere over 24–36 hours. Shorter operations — press forming at 10–120 seconds per piece, spray-booth glazing in discrete loads, and diamond grinding on fired parts — must stay synchronised with these long-duration firing stages. Any misalignment between a kiln load's completion and the next stage's working hours can add days to an order's lead time.

The facility runs three product classes with different routings. Alumina structural products (roughly 60% of production by piece count) go through all five stages: forming, bisque firing, glazing, glaze firing, and finishing. Alumina simple products (about 25%) skip glazing but still pass through glaze firing — their simpler geometry does not require a glaze coating. Silicon carbide parts (the remaining 15%) use a dedicated vacuum furnace at the glaze-firing stage and skip both bisque firing and glazing, because their sintering profile at 1,700–1,900 °C in a nitrogen atmosphere is incompatible with the glaze shuttle kilns.

Changeovers add significant time pressure. Press die changes take 30–90 minutes per geometry swap. Bisque kilns require 2–4 hours to exchange kiln furniture and stabilise temperature between alumina classes. Glaze-kiln thermal recovery ranges from 2–6 hours, and the direction of the switch matters — cooling from an alumina-structural profile to an alumina-simple profile recovers faster than the reverse, and alumina-simple-to-structural takes the full 6 hours. The glazing booth needs 30–60 minutes of cleaning between glaze formulations. At the finishing stage, switching between alumina and silicon carbide classes on a grinding station requires 45–60 minutes for diamond-wheel and coolant changes, while swaps between the two alumina classes take only 20 minutes.

Vectra Technical Ceramics runs 78 people at a roughly 4,600 m² facility, making three product classes across five production stages with twelve machines, scheduled by a two-person planning team. Around 200 active SKUs are grouped into the three classes, each sharing a firing profile and routing within its class. Production runs week-round on kilns (continuous 24/7 calendar) while forming, glazing, and finishing operate on a single day shift Monday through Friday.

Process overview

flowchart LR
    F["Forming<br/>(4 presses)"]
    BF["Bisque firing<br/>(2 shuttle kilns)"]
    G["Glazing<br/>(1 spray booth)"]
    GF["Glaze firing<br/>(2 shuttle + 1 vacuum)"]
    FN["Finishing<br/>(2 grinding stations)"]

    F --> BF
    BF --> G
    G --> GF
    BF --> GF
    GF --> FN
    F --> GF

The production flow from forming through finishing — five stages, with skip-routing paths for two product classes.

Skip-routing notes: Alumina simple products move from bisque firing directly to glaze firing, skipping the glazing stage entirely. Silicon carbide products skip both bisque firing and glazing, routing from forming directly to the dedicated vacuum furnace at the glaze-firing stage. Drying between forming and bisque firing is modelled as a transfer delay, not a separate stage with machine capacity.

Scheduling challenges and how Schantt handles them

In this scenario, the schedule is driven by customer orders across three product classes with different routings, at a facility where kilns run continuously but forming, glazing, and finishing operate on a day shift. Six to ten new orders arrive each week, and the planning team queues them against a rolling horizon of roughly four weeks. If your primary driver is make-to-stock or a blend of make-to-order and make-to-stock, the same modelling applies — the scheduling engine treats every production request as work that must pass through the configured route on available machine time.

Schantt schedules forward from a start date and optimises for minimum total production time — sequencing work so that each stage completes as early as the stage before it and available machine capacity allow. Two optimisation modes are available. Auto mode runs the full scheduling algorithm against the configured constraints and produces an optimised plan. Semi-Auto mode generates a draft schedule that the planner can then adjust manually before freezing it. Manual mode lets the planner build the schedule from scratch with no algorithmic guidance.

What Schantt handles well

• Sequential multi-stage scheduling — each product moves through its ordered stages; the delays between every handoff, including drying dwell between forming and firing, are applied automatically.
• Multi-machine parallel stages — every kiln stage has two or three parallel machines; Auto and Semi-Auto modes explore machine assignments across the available fleet at each stage.
• Mixed batch-and-flow pipelines — batch firing stages coexist in one route with batch pressing and grinding, and the timing between each is computed correctly.
• Multi-product routing with stage skipping — three product classes with divergent paths interleave on shared stages without the planner having to define the skip paths separately for each order.
• Sequence-dependent changeovers — thermal-recovery times and die changeovers are set as per-pair entries; the algorithm groups similar classes to minimise kiln idle time between loads.
• Shift-aware availability — kilns run on a continuous calendar while forming, glazing, and finishing run only on day shift; the scheduler respects each machine's working window, spreads operations that cross shift boundaries correctly, and does not schedule a start on a day-shift machine outside its operating hours.

How Schantt handles each challenge

1. Thermal-recovery changeovers accumulating across class switches.

• When kiln loads of different classes follow one another, each switch consumes 2–6 hours of thermal-recovery idle time. With five or more class switches in a week, this can total 10–30 hours of lost kiln time across the two glaze kilns.
• Schantt models thermal-recovery changeovers as directional per-pair entries — switching from one class to another can take longer than the reverse, just as it does on the floor. The optimisation groups kiln loads by class where possible, reducing the total number of week-to-week cross-class transitions and the idle time they cause.

2. Firing-window alignment.

• A glaze firing cycle runs 18–30 hours, and a load that finishes just after the last inspection shift can sit idle until the next working day, adding 3–7 days to an order's completion. In current practice this missed-window event occurs roughly once every two weeks.
• Schantt projects each kiln load's completion time with calendar awareness — if a load would finish outside the day-shift window available for the next stage, the schedule makes the trade-off visible before the kiln is loaded. The planner can advance or delay the start so the handoff lands inside the working window, or let the schedule split the operation across the boundary.

3. Shift-boundary mismatch between continuous kilns and day-shift finishing.

• Kilns run 24/7, but finishing operates 08:00–17:00 Monday through Friday. Roughly six kiln loads per week complete overnight or over the weekend, building a 600–1,000 kg buffer of fired parts that cannot be ground until the next shift. A Friday-night load sits idle for about 60 hours until Monday morning.
• Schantt applies a distinct calendar to each machine, so finishing machines only accept new work during their day-shift window. The schedule shows the growing buffer as loads complete outside finishing hours, and the planner can see exactly which kiln starts will land their output inside a working finishing slot. The accumulated backlog is projected forward rather than discovered when the shift starts.

4. Quality-release hold for certifiable products.

• Roughly 30% of alumina structural output and all silicon carbide parts require destructive testing before shipment, with a hold of 5–14 days after firing. This release gate is managed outside the scheduling system.
• Schantt does not model the quality hold as a scheduling constraint — the 24-hour cool-down included in the glaze-firing cycle covers the mandatory cooling period, and quality release happens as a manual gate outside the schedule. The planner reserves capacity to account for the hold in order promising, and the schedule shows the material as completed through glaze firing when it reaches finishing.

What to model in Schantt

The Schantt configuration for this scenario mirrors the production floor with five top-level entities:

Step-by-step setup

1. Create the stages in order. Add five stages — Forming, Bisque firing, Glazing, Glaze firing, Finishing — in sequence. On each stage's detail page, set the transfer times that connect it to the next stage:
- Forming to Bisque firing: 480 minutes (drying dwell for thin-section parts; thicker parts need operator-judgment extension)
- Bisque firing to Glazing: 120 minutes (staging and quality check)
- Bisque firing to Glaze firing: 120 minutes (direct bridge for alumina simple, which skips glazing)
- Glazing to Glaze firing: 60 minutes (glaze drying and transfer)
- Glaze firing to Finishing: 1,560 minutes (mandatory cool-down — 24 hours for glaze kilns, 26 hours for the vacuum furnace)
- Forming to Glaze firing: 540 minutes (bridge for silicon carbide — drying hold plus staging before vacuum sintering)

2. Add the machines to each stage. Assign each machine to its stage and select the appropriate calendar:
- Forming: HP-1, HP-2, HP-3 (hydraulic presses), IP-1 (isostatic press) — all on day-shift calendar
- Bisque firing: BF-K1, BF-K2 (shuttle kilns) — both on continuous calendar
- Glazing: GS-1 (automated spray booth) — day-shift calendar
- Glaze firing: GF-K1, GF-K2 (glaze shuttle kilns), VK-1 (vacuum furnace) — all on continuous calendar
- Finishing: DG-1, DG-2 (diamond grinding stations) — day-shift calendar

3. Create the product classes and define each class's routing. For each product class, set the ordered stage sequence on the Product Class detail page. Enable the skip legs by defining the correct stage sequence per class:
- Alumina structural: Forming → Bisque firing → Glazing → Glaze firing → Finishing (all five stages)
- Alumina simple: Forming → Bisque firing → Glaze firing → Finishing (skips glazing)
- Silicon carbide: Forming → Glaze firing → Finishing (skips bisque firing and glazing; the glaze-firing stage routes to VK-1 only)

4. Add one product per class. Create a representative product for each class — Alumina bushing under Alumina structural, Alumina substrate under Alumina simple, and SiC seal face under Silicon carbide. Each inherits its class's routing, so roughly 200 actual SKUs share these three configuration anchors.

5. Set machine capacity parameters and changeovers. On each Machine detail page, configure the cycle duration and batch size for every product class that runs on that machine. Then add the changeover-time entries for every pair of classes that follow one another on the same machine:
- Forming presses: cycle durations 10–25 minutes per batch depending on press tonnage; die changeovers 60–90 minutes per product-class switch
- Bisque kilns (BF-K1, BF-K2): 900-minute cycle (250 kg) for alumina structural, 780-minute cycle (300 kg) for alumina simple; changeover 3–4 hours
- Glaze kilns (GF-K1, GF-K2): 1,440-minute cycle (200 kg) for alumina structural, 1,320-minute cycle (220 kg) for alumina simple; thermal-recovery changeover 4–6 hours
- Vacuum furnace (VK-1): 1,800-minute cycle (100 kg) for silicon carbide only — no changeovers needed
- Grinding stations (DG-1, DG-2): 20–60-minute cycles per batch; changeover 20–60 minutes depending on class pair

6. Configure calendars, exceptions, and downtimes. Two calendars cover all machines. The continuous calendar (default) runs 24/7 for all kilns. The day-shift calendar covers forming, glazing, and finishing — 08:00–17:00 Monday through Friday with weekends and Friday evenings as non-working time. Add calendar exceptions for non-working days: New Year's Day, International Workers' Day, and a three-day year-end shutdown. Schedule the planned downtimes — an annual kiln refractory inspection on BF-K1 and a press overhaul on HP-2 — so they block availability during those windows.

For step-by-step instructions on configuring each of these in Schantt, see the Schantt documentation.

Common mistakes

1. Using a single blanket changeover time instead of per-pair values. Glaze-kiln thermal recovery from an alumina-structural profile to alumina simple can take up to 6 hours, while the reverse direction takes only 4 hours. A symmetric value misrepresents the real time impact of each switch direction.
- Fix: Configure changeover times as directional per-pair entries on each machine — set the duration for each (from-class, to-class) pair so the schedule models the true recovery time in both directions.

2. Modelling silicon carbide as a separate product class without restricting its routing to the dedicated vacuum furnace. Without the routing restriction, the schedule may assign silicon-carbide loads to a glaze shuttle kiln, whose firing profile at 1,300–1,500 °C cannot reach the 1,700–1,900 °C and nitrogen atmosphere that silicon carbide requires.
- Fix: On the Silicon carbide product class, set the routing so it reaches only the glaze-firing stage via VK-1 — the two shuttle kilns are not included in that class's route.

3. Including cool-down in the glaze-firing cycle duration without adjusting expectations. The 24-hour passive cool-down is baked into each glaze kiln's cycle duration, which makes apparent kiln occupancy longer than active firing time alone would suggest. Comparing the reported utilisation against industry benchmarks without accounting for this will show inflated numbers.
- Fix: When reviewing kiln utilisation figures, remember that the cycle includes a full day of passive cooling. Either adjust the expected utilisation baseline downward, or create a separate cooling stage after glaze firing to distinguish active firing from cool-down.

4. Overlooking the shift-boundary mismatch when reviewing the schedule. A schedule that shows kiln loads completing at 02:00 on a Saturday looks complete on the Gantt, but those parts cannot be ground until Monday morning at 08:00. The apparent finishing-stage lead time is misleading if the finishing machines' calendar is not checked.
- Fix: Verify that every schedule detail shows the correct calendar assignment for each machine. Finishing loads should only begin within the day-shift window. Review the projected buffer between glaze firing and finishing rather than assuming immediate transfer.

5. Forgetting that single-machine stages handling multiple product classes still need changeover entries. A single-machine stage that processes more than one product class defaults to zero changeover time unless entries are configured. The bisque-firing shuttle kilns, for instance, serve both alumina classes — without directional changeover entries, the 2–4 hours needed to swap kiln furniture and stabilise temperature between runs are invisible to the schedule.
- Fix: For every machine that runs two or more product classes — regardless of whether the stage has one machine or several — add changeover entries for each (from-class, to-class) pair that actually follows one another. Even on a single-machine stage, the time penalty for switching classes must be configured explicitly.

6. Adding only one representative product when multiple products per class would improve schedule visibility. A single product per class loses the ability to distinguish orders by individual SKU attributes — specific quantities, due dates, and customer allocations stay lumped at the class level. The schedule shows class-level aggregates but not per-product detail.
- Fix: Add additional products under each class for SKUs that need separate tracking, especially those with distinct order patterns or customer-specific labelling. The class still defines routing and processing parameters; the product carries the order-level identity.

What a good schedule looks like

Before adopting Schantt, the planning team at Vectra Technical Ceramics loaded kilns by manual judgment, grouping orders by class when they could and accepting the thermal-recovery penalty when they could not. Cross-class routing conflicts were caught when the load reached the kiln, not in the plan. The kiln-to-finishing handoff was managed reactively — the buffer of overnight and weekend output was discovered at shift start each Monday.

Before (manual planning):
- Kiln loads sequenced without class-aware grouping; thermal-recovery idle time accumulated across the week
- Firing-window mismatches occurred roughly every two weeks, adding 3–7 days to order completion
- Friday-night kiln loads sat for about 60 hours until the finishing shift started Monday morning; roughly 600–1,000 kg of fired parts accumulated as backlog each week
- Divergent routings (alumina simple skipping glazing, silicon carbide using the dedicated vacuum furnace) required manual override on every order

After (Schantt Auto mode):
- Kiln loads are grouped by class automatically, with directional thermal-recovery changeovers applied per switch pair — the algorithm sequences loads to minimise total cross-class transitions on each kiln
- Each kiln load's completion is projected against the finishing calendar; loads that would finish outside working hours are visible before the kiln is loaded, so the start can be adjusted
- The buffer of overnight and weekend output is calculated in the schedule; the planner sees the backlog accumulate and can plan the Monday morning grinding load in advance
- Divergent routings render from the product-class configuration without per-order manual override — each class's route is defined once and applied to every order within that class
- The schedule is generated in minutes instead of manual hours, and the planner adjusts the optimised result in Semi-Auto mode before freezing it