Production Scheduling for Tablet Supplements

Tablet supplement manufacturers face a scheduling challenge that blends batch chemistry with high-speed flow production: wet-granulation and direct-compression products share downstream equipment while following different physical paths upstream. This guide shows how to model that dual-route reality in Schantt — configuring stages, machines, directional changeovers, and split-shift calendars so your schedule reflects what actually happens on the plant floor.

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

Industry context

Tablet supplement production is an eight-stage hybrid flowshop that combines batch and flow processing in a single routed sequence. The upstream stages — weighing, blending, granulation, drying, milling — handle powder processing in batches, working from raw excipients and active ingredients into a uniform granulated blend. The downstream stages — compression, coating, packaging — run on a continuous or sustained-cycle basis and represent a different operating rhythm with higher throughput potential.

Not all tablet products follow the same path. Wet-granulation products (immediate-release and extended-release) pass through all eight stages: active ingredients are wet-massed, dried, and milled into a free-flowing granulation before compression. Direct-compression products skip granulation, drying, and milling entirely — the dry-blended powder feeds straight from blending to the tablet press. This dual-route structure means the scheduling system must handle divergent routings on shared equipment, with different processing physics for batch and flow stages.

Changeover times are industry-defining. Transitions between potency tiers on tablet presses and between coating types on pan coaters are directional — an IR→ER potency change on a press takes 240 minutes while the reverse takes 120 minutes, because extended-release residues require more thorough cleaning. Coating pan transitions follow a similar pattern: switching from a light to a dark colour coating takes less time than the reverse. These asymmetries compound when multiple product classes share the same downstream machines, and they make sequence planning the difference between an efficient week and one consumed by cleaning.

NovaVita Nutraceuticals runs approximately 70 people across three product classes and eight production stages at a single facility, scheduled by a two-person planning team.

Process overview

flowchart LR
    W["Weighing & Dispensing"]
    BL["Blending"]
    G["Granulation"]
    D["Drying"]
    M["Milling"]
    C["Compression"]
    CO["Coating"]
    P["Packaging"]

    W --> BL
    BL --> G --> D --> M --> C
    BL -.->|DC skip| C
    C --> CO --> P

The tablet supplement process: eight stages across two operating zones, with direct-compression products skipping the granulation-drying-milling block.

Direct-compression products skip granulation, drying, and milling entirely. Their route runs weighing → blending → compression → coating → packaging, with a bridging transfer time configured to account for the skipped stages.

Scheduling challenges and how Schantt handles them

The production schedule at a tablet supplements plant is driven by the weekly demand plan — typically, specific batch quantities for each product class that must ship by week-end. The optimizer minimizes total production time across all jobs, scheduling forward from a start date. For this scenario, the practical horizon is one to two weeks, covering the full batch-to-packaged-product cycle.

Schantt provides two scheduling modes. In Auto mode, the algorithm determines job sequence, machine assignments, and timing from a list of products and quantities. In Semi-Auto mode, you supply the production sequence in a fixed order, and the algorithm optimises machine assignments within it — useful when regulatory or material-flow constraints lock the sequence.

What Schantt handles well

• Sequence-dependent changeovers with directional asymmetry — Schantt's changeover-time configuration lets you set per-machine, per-pair durations that reflect real-world asymmetry: IR→ER on a tablet press costs 240 minutes while ER→IR costs 120, because ER residues demand more cleanup. The optimizer favours the shorter direction naturally.

• Dual-route per-class routing with stage skipping — Direct-compression products skip granulation, drying, and milling while sharing downstream equipment with wet-granulation products. Each product class defines its own set of stages; absent stages produce no operation and no Gantt row.

• Multi-machine stages with automatic assignment — With two tablet presses and two coating pans, Schantt explores machine assignments across each stage's parallel resources and picks the combination that minimises total production time. No manual press or pan allocation needed.

• Split-shift calendars (upstream single, downstream double) — Upstream stages (weighing through milling) run one shift; downstream stages (compression through packaging) run two. Separate calendars per machine group ensure operations advance only during working hours for that machine.

• Mixed batch-and-flow pipelines with partial transfer — Batch stages (blending, granulation, drying, coating) and flow stages (weighing, milling, compression, packaging) interleave in the same route. At the blending-to-compression handoff, the press can start on the first 50 kg without waiting for the full blend.

• Calendar exceptions and planned downtimes — Holidays, weekend shutdowns, and maintenance windows are modelled as exceptions and downtimes that the algorithm respects when sequencing.

How Schantt handles each challenge

1. Asymmetric changeover chains across shared downstream equipment.

• The two tablet presses and two coating pans serve all three product classes, each with a six-directional changeover matrix. An IR→ER potency transition on a press takes 240 minutes; ER→IR takes 120. Coating pan changeovers are similarly asymmetric: light→dark at 120 minutes, dark→light at 240, and an ER→IR coating transition takes 270 minutes while IR→ER takes 150. On the floor, sequence choice directly determines whether changeover time consumes 20–30% of available production hours or is contained to a smaller fraction. Grouping same-class runs and sequencing from shortest-to-longest changeover is the planner's primary lever, but with three classes and ten shared machines, manual optimisation is impractical.
• Schantt models each changeover as a directional duration per machine and product-class pair. The algorithm evaluates all feasible sequences during optimisation, naturally preferring sequences with shorter aggregate changeover time. The planner enters the directional durations once — drawing on their own QA and cleaning-policy knowledge — and the schedule accounts for every transition on every shared machine without manual per-run adjustment.

2. Divergent routings and the DC skip route.

• Direct-compression products bypass granulation, drying, and milling — three stages that wet-granulation products must traverse. This creates a scheduling asymmetry where DC jobs arrive at the press sooner but must also compete with WG jobs for the same compression capacity. Without proper routing configuration, a DC product scheduled between two WG runs can produce silent timing errors — the system either inserts zero-duration gaps for the skipped stages or fails to account for the material handoff time between blending and compression.
• Schantt's per-class routing means each product class defines exactly the stages it traverses. The DC class omits granulation, drying, and milling from its routing entirely. A bridging transfer time of 45 minutes is configured from blending directly to compression, representing the real material-handoff interval across the skipped span. The result: DC jobs move from blending to compression correctly timed, and the Gantt shows no phantom operations.

3. Upstream single-shift bottleneck constraining downstream two-shift throughput.

• Upstream stages run one shift (07:00–15:30), capping throughput at roughly three blending batches per day (600 kg at 200 kg per batch). Downstream compression runs two shifts (07:00–23:00) on two presses — a per-machine throughput of 80,000–120,000 tablets per hour. This creates roughly an 8:1 throughput ratio between compression and blending capacity per machine. The plant runs a decoupling buffer — a 180-minute drum staging hold between compression and coating — to smooth the flow. But upstream starvation of the downstream line is a recurring risk: if blending finishes late on a Friday, the presses may sit idle Monday morning waiting for milled granules.
• Schantt assigns separate calendars to upstream and downstream machine groups: the dispensing booth, bin blender, granulator, dryer, and cone mill run on the single-shift calendar; both tablet presses, both coating pans, and the bottle line run on the two-shift calendar. Simulated material flow propagates upstream scarcity downstream automatically. When a WG batch starts late, the algorithm extends compression wait times without manual intervention. The planner sees the material-starvation gaps on the Gantt and can adjust the upstream schedule or sequence to compensate.

4. Variable coating-cycle durations by product class.

• Coating cycle times differ significantly by product class: an ER-WG batch takes 210 minutes in a coating pan while IR-WG takes 105 minutes and DC takes 95 minutes. The ER-WG product requires a functional coating for extended-release properties, nearly doubling the pan residence time compared with the cosmetic film coatings of the other classes. Scheduling a long ER-WG coating run between two shorter IR-WG or DC runs extends the total time the coating pans are occupied, which shifts drum staging and can delay packaging downstream. A planner managing these durations manually risks under-utilising the second pan or committing to a sequence that finishes late.
• Each product class on each coating pan carries its own processing-time configuration. Schantt uses the per-class duration to compute each operation's true calendar time, and with two pans available, it explores assignments that balance coating load across pans. The algorithm can assign the long ER-WG run to one pan while the other handles shorter IR-WG and DC runs concurrently, reducing overall coating-stage completion time.

5. Format-dependent packaging changeovers.

• The bottle packing line serves all three product classes, and format changes between bottle sizes (120-count, 100-count, 90-count) take 60–120 minutes depending on the change complexity. Same-class format changes take roughly 60 minutes; cross-class changes take about 90 minutes. Packaging is the final stage — any changeover delay here pushes the ship-date. When packaging idle time compounds upstream delays, the schedule can slip by a full shift even when earlier stages ran on time.
• The packaging line's changeover matrix includes all six product-class pairs. Schantt sequences packaging operations to minimise the aggregate changeover time across the schedule, and because packaging runs on the two-shift downstream calendar, it recovers time faster than upstream stages. The result: packaging changeover time is optimised alongside every other stage, not as an afterthought.

What to model in Schantt

The entity counts below represent the full configuration surface for this scenario.

Step-by-step setup

1. Create the stages in production order. Define weighing (flow), blending (batch), granulation (batch), drying (batch), milling (flow), compression (flow), coating (batch), and packaging (flow) — each with its correct production type. On each stage's detail page, configure the transfer times between successor stages:

• Weighing → blending: 30 minutes (IBC transfer)
• Blending → granulation: 15 minutes (IBC discharge)
• Granulation → drying: 15 minutes (direct chute)
• Drying → milling: 120 minutes (conservative equilibration baseline)
• Milling → compression: 30 minutes (vacuum transfer)
• Blending → compression: 45 minutes (bridging transfer for DC skip route)
• Compression → coating: 180 minutes (drum staging buffer)
• Coating → packaging: 60 minutes (drum transfer)

2. Add the machines to each stage. Assign the dispensing booth to weighing, the bin blender to blending, the high-shear granulator to granulation, the fluid-bed dryer to drying, the cone mill to milling, two rotary tablet presses to compression, two perforated pan coaters to coating, and the bottle packaging line to packaging.

3. Create the product classes and define their routings. Create IR-WG, DC, and ER-WG. For each class, define the stages it traverses:

• IR-WG: all eight stages (weighing through packaging)
• DC: five stages — weighing, blending, compression, coating, packaging — omitting granulation, drying, and milling
• ER-WG: all eight stages

Enable partial transfer on blending for all three classes, set to 50 kg. This allows the downstream press to start before the full 200 kg blend completes.

4. Add one representative product per class. Create Calcium + Vitamin D3 120-count in IR-WG, Vitamin C 500 mg 100-count in DC, and Time-Release Niacin 500 mg 90-count in ER-WG. Each inherits its class routing automatically.

5. Configure machine capacity parameters and changeovers. On each machine's detail page, set the batch-stage parameters (cycle duration and batch size) and flow-stage parameters (throughput per hour) for the product classes that machine handles. Key values:

• Blending (bin blender): 200 kg batch, 20 min cycle — all classes
• Granulation (high-shear granulator): 200 kg batch, 25 min cycle — IR-WG and ER-WG only
• Drying (fluid-bed dryer): 200 kg batch, 90 min (IR-WG) / 120 min (ER-WG)
• Coating (pan-1 and pan-2, same values): 200 kg batch, 105 min (IR-WG) / 95 min (DC) / 210 min (ER-WG)
• Compression (press-1 and press-2): 100,000 tbl/hr (IR-WG) / 120,000 tbl/hr (DC) / 80,000 tbl/hr (ER-WG)
• Weighing, milling, packaging: throughput as flow stages

Then enter the changeover matrix for each shared machine. Each machine shared by two or more classes needs all six directional pairs:

• Dispensing booth: 15 min for all pairs
• Bin blender: 30 min for IR-WG↔DC, 45 min for IR-WG↔ER-WG and DC↔ER-WG
• Granulator: 270 min for IR-WG↔ER-WG
• Dryer: 180 min for IR-WG↔ER-WG
• Cone mill: 30 min for IR-WG↔ER-WG
• Press-1 and press-2: 180 min IR↔DC, 240 min IR→ER / 120 min ER→IR, 240 min DC→ER / 120 min ER→DC
• Pan-1 and pan-2: 120 min IR→DC / 240 min DC→IR, 150 min IR→ER / 270 min ER→IR, 150 min DC→ER / 270 min ER→DC
• Bottle line: 90 min for all pairs

6. Configure calendars, exceptions, and downtimes. Create the upstream single-shift calendar (Monday–Friday 07:00–15:30) and assign it to the dispensing booth, bin blender, granulator, dryer, and cone mill. Create the downstream two-shift calendar (Monday–Friday 07:00–23:00), set it as the team default, and assign it to both presses, both coating pans, and the bottle line.

Add three calendar exceptions: New Year's Day (Jan 1), International Workers' Day (May 1), and Christmas Day (Dec 25) — all non-working across all calendars. Add two machine downtimes: semi-annual preventative maintenance on the granulator (July 1–3) and the year-end plant shutdown (December 24–31, factory-wide).

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

Common mistakes

1. Using a single blanket changeover duration per machine instead of directional per-pair values. A single 180-minute changeover on the tablet press ignores the real asymmetry — IR→ER needs 240 minutes while ER→IR needs 120. The optimizer cannot favour the shorter direction and may sequence an unfavourable transition, adding hours to the schedule. Fix: configure all six directional pairs for every shared machine, using durations that reflect the actual cleaning protocol per transition.

2. Assigning the same calendar to upstream and downstream stages. When upstream equipment runs one shift and downstream runs two, a single calendar overrides the actual operating hours of one group. Upstream machines appear available when they are not, or the press schedule assumes fewer hours than the two-shift reality allows. Fix: create two calendars — one single-shift, one two-shift — and assign each machine to its correct calendar.

3. Omitting the bridging transfer time for the DC skip route. Direct-compression products skip granulation, drying, and milling. Without a bridging transfer time from blending directly to compression, the handoff across the skipped stages has no defined duration — material movement is effectively instantaneous in the model. Fix: configure a 45-minute bridging transfer time from blending to compression on the Stage detail page.

4. Configuring coating pan cycle durations as identical across all product classes. ER-WG requires a functional coating that takes 210 minutes per batch — roughly double the 95–105 minutes needed for cosmetic film coatings on IR-WG and DC. A single blanket duration over-runs the schedule for short-cycle products or under-runs the long-cycle product, producing a schedule the floor cannot follow. Fix: set per-class processing times on each coating pan, matching the actual cycle for each product.

5. Treating the year-end shutdown as a calendar exception rather than a multi-day downtime. A calendar exception handles single-date holidays. A multi-day shutdown (Dec 24–31) needs a machine downtime entry to subtract all those days from working capacity. A calendar exception on Dec 25 alone leaves Dec 24, 26, 27, 28, 29, 30, and 31 as working days, incorrectly showing capacity that does not exist. Fix: create a factory-wide machine downtime spanning the full shutdown period.

What a good schedule looks like

Before (manual sequencing): Production orders are scheduled by due-date with no systematic changeover grouping. The planner sequences by customer request, unaware of cumulative cleaning time:

• Tablet press changeovers consume an estimated 20–30% of available production hours — the press runs one IR batch, then one ER, then DC, each transition incurring the full directional penalty without grouping same-class runs.
• Coating pan sequence ignores colour-depth ordering: a dark-coated ER run before a light-coated IR run imposes the full 240-minute dark→light changeover instead of the 120-minute light→dark transition.
• The decoupling buffer between compression and coating is frequently consumed or overshot, causing packaging to idle while waiting for coated tablets.
• One week's schedule often spills into the next, particularly when upstream single-shift capacity is the constraint.

After (Schantt Auto mode): The algorithm evaluates all feasible sequences and machine assignments, favouring directions with shorter changeover time:

• Grouped same-class runs reduce press changeover frequency — IR batches sequence together, then ER, then DC, with the 120-minute (ER→IR) transition occurring only once per week instead of every shift.
• Coating pan assignments balance load: the long ER-WG coating cycle (210 minutes) runs on one pan while the other handles shorter IR-WG and DC cycles concurrently.
• Upstream material starvation is visible on the Gantt as wait-material segments, and the schedule tightens — press idle time on Monday morning is reduced because the simulation propagates blending completion times forward.
• Total production time for a representative week's three-product run drops measurably (hours saved in changeover consolidation alone), and the schedule fits within five working days on the downstream calendar.