Production Scheduling for Oral Solid Dose Tablets

This guide walks production planners and operations managers through setting up oral solid dose tablet scheduling in Schantt — modelling wet-granulation and direct-compression routes through eight stages, configuring sequence-dependent changeovers and shift-aware calendars, and using the scheduling algorithm to build optimised plans. All examples follow Palisade Pharma, a representative mid-market manufacturer making three product classes across a typical tablet facility.

This guide follows a fictional composite company built from industry research on oral solid dose tablets; all names, parameters, and figures are illustrative.

Industry context

Oral solid dose (OSD) tablet manufacturing is the most common solid-dosage form in global pharmaceutical production, covering immediate-release, extended-release, and fast-dissolve formulations. Production follows a multistage flow: raw materials are weighed and dispensed, then processed through granulation (where needed), dried, milled, blended with excipients, compressed into tablets, coated, and packaged. The two principal routes are wet granulation, which traverses all eight stages, and direct compression, which skips granulation, drying, and milling by blending dry powders directly. Both routes share the downstream stages — compression, coating, and packaging — making machine availability and changeover sequencing the dominant scheduling constraints across a shared line.

The industry operates on mixed processing physics: granulation, drying, blending, and coating run as batch stages with fixed cycle durations per batch, while weighing, milling, compression, and packaging run as flow stages at continuous throughput rates. Granulation batch sizes at mid-market facilities typically range from 100 to 400 kg; compression operates at 50,000 to 400,000 tablets per hour per press; and coating cycles span 60 to 240 min per batch including spray application and drying. A typical mid-market facility balances 1–2 granulation suites, 2–4 tablet presses, 2–3 coating pans, and 1–3 packaging lines, carrying 15–50 SKUs grouped into 3–6 product classes. The three machine downtimes and three calendar exceptions Palisade observes — New Year's Day, International Workers' Day, and Christmas Day — reflect internationally recognised non-working days that apply uniformly across the plant.

Most mid-market OSD plants still schedule in spreadsheets rather than dedicated production scheduling software. Around 60 % rely on manual Gantt-building in spreadsheets, roughly 30 % on ERP-based planning, and only about 10 % on dedicated advanced planning systems. A 2-person planning team managing three product classes across eight stages, ten machines, and two shift patterns faces a coordination load that spreadsheets handle poorly — particularly when changeover sequencing and parallel-machine assignment interact across the shared line.

Palisade Pharma runs approximately 85 people, making 3 product classes across 8 production stages, scheduled by a 2-person planning team.

Process overview

flowchart LR
    A["Weighing & Dispensing"]
    B["Wet Granulation"]
    C["Drying"]
    D["Milling"]
    E["Blending"]
    F["Compression"]
    G["Coating"]
    H["Packaging"]
    A --> B --> C --> D --> E --> F --> G --> H
    A -.-> E

Material flows through eight production stages from raw-material dispensing to packaged tablets; the direct-compression route (dashed arrow) skips granulation, drying, and milling.

Direct-compression (DC) products bypass granulation, drying, and milling, entering the line at the blending stage. The bridging transfer time from weighing to blending maintains the handoff delay across the skipped span.

Scheduling challenges and how Schantt handles them

The schedule at Palisade is driven by a demand plan — firm orders and forecast quantities for each product across a planning window. This guide assumes a monthly demand horizon fed into Schantt as job quantities for each product. If your plant schedules from a different trigger, such as weekly order books or Kanban signals, the same setup applies; the job list is the schedule input regardless of the demand source.

Schantt's scheduling algorithm minimises total production time — the overall completion time across all jobs — and schedules forward from a start date you choose. The practical horizon for this guide is one month, though Schantt handles shorter or longer windows with the same configuration. You run the schedule in either Auto or Semi-Auto mode. In Auto mode, the algorithm decides job sequence, machine assignments, and timing together, reordering the job list to find the fastest overall plan. In Semi-Auto mode, you fix the production order and let the algorithm optimise machine assignments within that sequence, which is useful when a campaign order is operationally determined — for example, by material availability or a committed dispatch sequence.

What Schantt handles well

• Sequential multi-stage production — ordered stages with per-class routing and forward-only transfer times chain each job through its required steps.
• Multi-machine stages with parallel lines — multiple machines at the same stage (two tablet presses, two coating pans); the algorithm assigns jobs across them.
• Mixed batch-and-flow pipelines — batch and flow stages in the same route, each timed by its own physics (batch cycle or continuous throughput), with wait-material segments where downstream stages outrun supply.
• Multi-product routing with stage skipping — direct-compression products skip granulation, drying, and milling via per-class routing, with bridging transfer times that preserve handoff delays across the skipped span.
• Sequence-dependent changeovers — directional, per-machine changeover times keyed by product-class pair, so the algorithm can favour lower-changeover sequences.
• Shift-aware availability with calendar exceptions and downtimes — working windows built from shift calendars, with holiday exceptions and planned maintenance windows that the schedule respects.

How Schantt handles each challenge

1. Granulation suite contention.

• The single high-shear granulator and fluid-bed dryer at Palisade are a coupled batch pair; together they can complete roughly 2–3 granulation batches per single shift, capping the plant's upstream throughput for wet-granulation products. A planner scheduling by hand risks overloading the granulation suite, producing a plan that cannot keep up with downstream compression demand.
• Schantt models both as batch stages with per-class cycle durations and batch sizes. Granulation is set to 20 min per batch (IR-WG) or 25 min (ER-WG), and drying to 90 min (IR-WG) or 120 min (ER-WG), with a 15 min transfer between them. The Upstream Core calendar (07:00–15:30, single shift) clamps all starts into working hours and adds wait gaps across non-working periods. The algorithm chains the 15 min transfer from granulation to drying and advances through calendar minutes only, so the schedule shows exactly how many complete granulation batches fit in a day and where the upstream bottleneck first limits downstream throughput.

2. Sequence-dependent changeover burden.

• Changeovers between product classes consume an estimated 20–30 % of available production time across the plant. Tablet-press changeovers alone range from 3 h between similar classes (IR↔DC) to 4 h between the most different pair (DC↔ER), and coating-pan changeovers from 2.5 h (IR↔DC) to 4.5 h (IR↔ER, DC↔ER). A planner working in a spreadsheet has no way to evaluate whether a different campaign order at the press would save two hours of changeover time across the week — the sequence is chosen once and committed.
• Schantt models changeovers as a directional time matrix per machine × product-class pair — the planner enters the from→to duration for every pair a machine can encounter. For a tablet press shared by all three classes, that means six directional entries covering every ordered combination. In Auto mode the algorithm explores job sequences to minimise total production time, naturally clustering similar products to reduce changeover impact; in Semi-Auto it holds the planner's fixed order and assigns machines to reduce changeover impact where the parallel capacity allows. The resulting changeover segment appears on each operation's Gantt row ahead of the processing bar, with the duration visible in the operation tooltip. Tooling — punches and dies — is treated as always available when a press runs; finite tool-set allocation is tracked manually outside the schedule. Cleaning validation between campaigns (required between certain product-class switches) is folded into the changeover duration as a single time value; the validation protocol and its documentation are managed through the facility's quality system, not Schantt.

3. Coating capacity pressure.

• Coating is 2–4× slower per kilogram than compression — an ER-WG coating cycle runs 210 min per batch (3.5 h) versus 90 min for DC — and with only two perforated pan coaters at Palisade, coating is the plant's structural downstream bottleneck.
• Schantt assigns coating jobs across both pans automatically. Each coater has its own per-class cycle duration (IR-WG: 120 min, DC: 90 min, ER-WG: 210 min), directional changeovers, and the Downstream Core calendar (06:00–22:00, two shifts). The algorithm balances the coating load across the two machines while respecting the compression buffer (a 3 h transfer time from compression to coating that represents drum staging). The schedule shows exactly which pan runs which product and when each batch finishes.

4. Mixed batch-and-flow rhythm across a single route.

• Wet-granulation products cross both batch and flow stages — upstream stages run batch cycles (granulation, drying), while compression and packaging downstream run at continuous rates. A spreadsheeted plan that treats all stages as either uniform cycle times or uniform throughput produces mismatched timing and hidden idle gaps.
• Schantt's stage type (batch or flow) determines the duration physics per stage. Batch stages compute duration from the job quantity, the batch capacity, and the cycle time per batch; flow stages compute duration from the job quantity and the machine's throughput rate per hour. The simulation chains each downstream stage from its upstream completions, emitting wait-material segments when a downstream flow stage runs ahead of the batch supply. The planner sees processing and wait intervals on the Gantt, with tooltips explaining each pause.

5. Direct-compression skip routing and shared machine coordination.

• DC products skip three upstream stages but blend in the same blender, compress on the same presses, coat in the same pans, and pack on the same line as wet-granulation products. Getting the skip route right — including the bridging handoff from weighing directly to blending — is essential, or the schedule either invents empty operations for skipped stages or ignores the material handoff time.
• Schantt's per-class routing defines exactly which stages each product class visits. DC's routing skips granulation, drying, and milling; a bridging transfer time of 45 min from weighing to blending accounts for the material handoff across the skipped span. The resulting schedule shows DC jobs flowing weighing → blending → compression → coating → packaging with no ghost rows for skipped stages, and the bridging time prevents them from arriving too early at blending relative to the upstream weighing finish.

What to model in Schantt

Palisade's scheduling model in Schantt starts with five entity types:

Step-by-step setup

1. Create the stages and set transfer times. Add the eight stages in production order (position 1–8), setting each stage's type to batch or flow as listed in the entity table above — weighing is flow, granulation is batch, drying is batch, milling is flow, blending is batch, compression is flow, coating is batch, packaging is flow. On each stage's detail page, configure the forward-only transfer times from that stage to its successor. The eight transfer-time entries cover every stage-to-stage handoff, including the bridging transfer from weighing to blending (45 min) needed for the DC skip route. Key transfer times include:
- Weighing → granulation: 30 min (IBC transfer to granulation suite)
- Granulation → drying: 15 min (direct chute discharge in coupled suite)
- Drying → milling: 120 min (per-class equilibration hold, conservative for ER-WG)
- Weighing → blending (DC bridge): 45 min
- Blending → compression: 45 min (IBC transfer to press hopper)
- Compression → coating: 180 min (drum-staging buffer)
- Coating → packaging: 120 min

2. Add the machines to each stage. Assign each of the 10 machines to its owning stage — one dispensing booth to weighing, one high-shear granulator to granulation, one fluid-bed dryer to drying, one cone mill to milling, one bin blender to blending, two rotary tablet presses to compression, two perforated pan coaters to coating, one blister packaging line to packaging. Machines at multi-machine stages (compression, coating) will receive concurrent job assignments from the algorithm.

3. Create product classes and define routings. Create the three product classes: IR-WG (unit: tablet), DC (unit: tablet), ER-WG (unit: tablet). On each class's detail page, define its per-class routing — the ordered stages the class traverses. IR-WG and ER-WG route through all eight stages; DC routes through weighing, blending, compression, coating, and packaging only. On each class's routing page, enable the partial-transfer toggle at the blending stage for all classes (transfer quantity: 50 kg), allowing blending to begin pushing material to the press hopper before the full blend batch is complete.

4. Add the products. Create one representative product per class: Metformin IR 500 mg (IR-WG), Ibuprofen DC 200 mg (DC), Metformin XR 750 mg (ER-WG). Each product inherits its class's routing and machine configuration automatically. Assign a distinct display colour to each product for Gantt readability.

5. Set machine capacity parameters and changeovers. On each machine's detail page, enter the per-class processing parameters and the directional changeover times between classes the machine can process. The parameters differ by stage type:
- Batch stages (granulation, drying, blending, coating): batch size (200 kg for all) and cycle duration per class — granulation 20 min (IR-WG) / 25 min (ER-WG); drying 90 min (IR-WG) / 120 min (ER-WG); blending 20 min (all three classes); coating 120 min (IR-WG) / 90 min (DC) / 210 min (ER-WG) on both pans.
- Flow stages (weighing, milling, compression, packaging): throughput per class — weighing 200 kg/h (all classes); milling 400 kg/h (IR-WG, ER-WG); compression 100,000 tablets/h (IR-WG) / 120,000 tablets/h (DC) / 80,000 tablets/h (ER-WG) per press; packaging 60,000 tablets/h (all classes).

For changeovers, enter the directional durations for every machine shared by two or more classes — 48 directional pairs in total across the plant. The granulator and fluid-bed dryer each have one relevant pair (IR↔ER) at 270 min and 180 min respectively. The dispensing booth has 15 min across all pairs. The bin blender has 60 min across all pairs. Each tablet press has six directional entries ranging from 180 min (IR↔DC) to 240 min (DC↔ER). Each coating pan has six entries from 150 min (IR↔DC) to 270 min (IR↔ER, DC↔ER). The blister line has 120 min across all pairs.

6. Configure calendars, exceptions, and downtimes. Create the two weekly calendars: Upstream Core (single shift, Mon–Fri, 07:00–15:30) assigned to weighing, granulation, drying, and milling; Downstream Core (two shifts, Mon–Fri, 06:00–22:00, set as default) assigned to blending, compression, coating, and packaging. Add the three geography-neutral calendar exceptions: New Year's Day (1 January, non-working), International Workers' Day (1 May, non-working), Christmas Day (25 December, non-working). Add the three machine-downtime windows: high-shear granulator and fluid-bed dryer semi-annual maintenance (1–7 July), and factory-wide instrument calibration (17–18 March).

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

Common mistakes

1. A single blanket changeover instead of directional per-pair times. Using a single changeover duration for all product-class pairs hides the 60–90 min difference between an easy swap (IR↔DC on a press, 180 min) and a hard one (DC↔ER, 240 min). the algorithm cannot favour lower-changeover sequences if it sees uniform durations. Fix: Enter every directional pair that a machine can encounter — 48 entries across the plant — using the real industry-advised durations for each from→to combination.

2. One product class covering both wet-granulation and direct-compression routes. A single class that routes through all eight stages forces DC products through granulation, drying, and milling operations they never perform on the plant floor, creating phantom operations and inflating the schedule. Fix: Create separate classes — IR-WG and ER-WG for wet-granulation products, DC for direct-compression products — each with its own routing that reflects reality.

3. Missing the bridging transfer time for the skip route. When DC products skip three stages, the handoff from weighing to blending lacks a defined delay — the schedule treats the transfer as zero minutes unless explicitly configured. Fix: Add the bridging transfer time from weighing to blending (45 min) so the gap across the skipped span has a realistic delay.

4. Machine count that does not match the floor. Modelling a single tablet press or single coating pan when the plant runs two of each produces a schedule that bottlenecks at stages that are not actually constrained — and hides the real machine-assignment decisions the algorithm should be making. Fix: Add both rotary tablet presses and both perforated pan coaters so the algorithm can distribute work across the available parallel capacity.

5. Calendar mismatch between upstream and downstream shifts. Applying a single calendar across all eight stages ignores the real working-time difference — single shift (07:00–15:30) for upstream and two shifts (06:00–22:00) for downstream. The schedule then allows granulation to keep running into the evening, or clamps compression to a single shift's hours, producing timings that cannot be executed. Fix: Create separate calendars for the two shift patterns and assign the correct one to each machine.

What a good schedule looks like

A well-configured Schantt schedule replaces a manually maintained spreadsheet with a live, optimised plan that reflects real machine capacity, shift patterns, and changeover logic.

Before (spreadsheet baseline): Palisade's planning team of two builds a weekly Gantt by hand in a spreadsheet. Changeover sequences rely on tribal knowledge passed between planners; testing a different campaign order means re-entering rows for half the jobs, and the team rarely has time to explore more than one scenario. The schedule takes hours to update — a single job change cascades through every downstream row — and what-if analysis is impractical. Hidden bottlenecks, such as coating capacity on ER-WG campaigns or granulation throughput on back-to-back IR-WG batches, surface only when actual production falls behind the plan, by which point the schedule is already out of date.

After (Schantt Auto mode): The planner enters the month's jobs — three products, each with a production quantity — sets the start date, and runs the schedule in Auto mode. The algorithm sequences the jobs, assigns each operation to the best machine (which press, which coating pan), and times every step against the correct calendar and shift pattern. The optimisation minimises total production time by clustering similar changeovers to reduce cleaning time and by balancing load across the two presses and two coating pans. What-if analysis becomes a matter of minutes: adjust a quantity, swap the campaign order to test a different sequence, add a new product, and re-run. The Gantt shows every operation with its changeover, processing, and wait-material segments, with shaded calendar overlays for non-working time, holiday exceptions, and planned maintenance — so the planner can see at a glance why a job paused between shifts or why an operation skipped the maintenance week.