Production Scheduling for Protein Bars / Functional Food Bars

Schantt models the full protein bar production flow — blending, forming, baking, enrobing, and packaging — as a hybrid batch-and-flow pipeline with parallel machines and sequence-dependent changeovers. This guide walks through configuring a five-stage protein bar facility in Schantt and creating optimized schedules that cluster coating runs and balance line assignments.

This guide follows a fictional composite company built from industry research on protein bars / functional food bars; all names, parameters, and figures are illustrative.

Industry context

Protein bars and functional food bars are formulated from protein powders, grains, binders, sweeteners, and functional ingredients such as vitamins or fibre. Production typically starts with dry- and wet-blending a dough mass in batch mixers, then forming bars via cold extrusion, followed by baking (for baked varieties), enrobing with chocolate coating (for coated varieties), and finally flow-wrapping. The combination of dairy and plant-based ingredients creates a strict allergen-crossing problem: changeover cleaning between whey and plant-based runs is significantly longer than transitions within the same allergen category. Coating changeovers add further time as enrober troughs, curtains, and temper units must be emptied, cleaned, and re-tempered between coating types. This dual changeover dynamic — allergen-based at the forming stage and coating-based at the enrobing stage — makes protein bar scheduling a sequence-dependent combinatorial problem that spreadsheets struggle to optimise.

Manufacturers in this segment typically manage 10 to 60 SKUs across two or three product families with divergent routings — some baked, some cold-formed, some enrobed, some uncoated — on shared parallel lines. The scheduling challenge is to sequence production in campaign blocks that minimise allergen and coating changeover penalties while keeping all lines fed and avoiding starvation at the packaging stage.

Crest Nutrition Company runs approximately 55 people at a 2,500 m² facility, making three product classes across five production stages, scheduled by a two-person planning team. Crest produces Chocolate Whey Crunch (whey-based, chocolate-coated, cold-formed), Almond Oat Bake (plant-based, uncoated, baked), and Dark Cocoa Plant Bar (plant-based, chocolate-coated, cold-formed).

Process overview

flowchart LR
    A["Blending<br/>(Batch)"]
    B["Forming / Extrusion<br/>(Flow)"]
    C["Baking<br/>(Flow)"]
    D["Enrobing / Coating<br/>(Flow)"]
    E["Packaging<br/>(Flow)"]

    A -->|"5 min tote<br/>transfer"| B
    B -->|"15 min cool<br/>Baked class"| C
    B -->|"15 min cool<br/>Coated classes"| D
    C -->|"15 min cool<br/>Uncoated class"| E
    D -->|"5 min"| E

Figure 1: Crest Nutrition Company's production flow. Blending is a batch stage; all subsequent stages are flow stages. Product classes skip stages per their individual routings.

Crest's three product classes follow three distinct routes through these five stages. Chocolate Whey Crunch and Dark Cocoa Plant Bar (both cold-formed, coated) skip baking entirely — bars travel from forming directly to pre-coat cooling. Almond Oat Bake (plant-based, uncoated, baked) skips enrobing — bars travel from the baking cooling conveyor directly to packaging.

Scheduling challenges and how Schantt handles them

Crest's planning team works from a rolling weekly demand forecast and confirmed customer orders entered as a list of product quantities for the coming week — the schedule is driven by demand quantity, not by raw-material availability. (If your primary constraint differs, the modeling approach still applies; adjust the input list to match your order book or production plan.) Schantt's algorithm minimises total production time — the overall completion time across all jobs — by optimising job sequence, machine assignments, and timing, scheduling forward from a chosen start date. This guide assumes a one-week rolling horizon (Monday to Friday, two shifts). Schantt offers two optimisation modes: Auto mode, where the algorithm determines job sequence and machine assignments from scratch, and Semi-Auto mode, where the planner fixes the production order and the algorithm optimises machine assignments within that fixed sequence.

What Schantt handles well

1. Multi-stage sequential routing with mixed batch-and-flow physics — model blending as a batch stage (500 kg, 12-minute cycle) and forming, baking, enrobing, and packaging as flow stages (throughput in bars per hour), all in one route per product class.

2. Per-class routing with stage skipping — each of Crest's three product classes follows its own path: the baked class skips enrobing, the coated classes skip baking. Schantt models each class's route independently, producing no operations for skipped stages.

3. Directional sequence-dependent changeovers — the enrober changeover of 65 to 75 minutes between dairy and plant-based coatings and the forming line changeover of the same duration across allergen categories are modeled as directional time penalties. The algorithm clusters similar runs to reduce total changeover time.

4. Multi-machine parallel stages — with two forming lines, two enrobers, and two packaging lines, Schantt explores machine assignments across each stage's parallel pool in Auto and Semi-Auto modes, choosing the combination that minimises total production time.

5. Shift-aware calendars with exceptions and downtimes — the two-shift Monday-to-Friday 06:00-to-22:00 pattern, three annual holidays, and planned maintenance windows (quarterly clean-in-place deep-clean, annual enrober overhaul) are all modeled and rendered as Gantt overlays.

How Schantt handles each challenge

1. Coating changeover campaign sequencing.

• The enrober changeover between whey-chocolate and plant-based-chocolate coatings at Crest takes 65 to 75 minutes — a full hour or more lost per coating switch across the two enrobers. With weekly coating-type rotation, Crest was spending 6 to 10 hours per week on enrober changeovers alone. Coating changeovers require emptying the enrober trough, cleaning the curtain and temper unit, and re-stabilising the coating at the correct temperature (29 to 32 °C for chocolate). Each switch consumes a directional time penalty that depends on which coating type is being removed and which is being introduced.
• Schantt handles it through directional, per-machine changeover entries on each enrober. The changeover time from whey-chocolate to plant-based-chocolate (75 minutes) and from plant-based-chocolate to whey-chocolate (65 minutes) are both stored, so the algorithm's total-production-time calculation penalises plans that interleave coating types. In Auto mode the algorithm clusters coating-homogeneous campaigns; in Semi-Auto mode the planner sets the coating order and the algorithm assigns machines to respect it. The changeover segment appears as a labelled band before the next operation's processing bar on the Gantt.

2. Allergen changeover sequencing on forming lines.

• Crest's forming lines share the same allergen-crossing penalty as the enrobers: switching from a whey-based formulation to a plant-based formulation requires 75 minutes of cleaning, while the reverse direction takes 65 minutes. Transitions between two plant-based formulations (baked versus coated) take only 20 minutes. The forming lines are the first point where allergen categories diverge. A whey-to-plant-based run order on either extrusion line consumes over an hour of productive time for cleaning and line verification. Without a sequencing strategy, the schedule drifts into high-changeover patterns that erode each shift's effective throughput.
• Schantt handles it with the same directional changeover mechanism applied to each extrusion line. The changeover matrix captures all six product-class-pair directions on both forming lines. Because the algorithm sees the time penalty of each transition, it favours sequences that group runs by allergen category. In Semi-Auto mode the planner locks the dairy-to-plant-based campaign order, and Schantt assigns jobs to forming lines within that fixed sequence.

3. Parallel line assignment and bottleneck balancing.

• With two extrusion lines, two enrobers, and two packaging lines, machine assignment directly determines whether work-in-progress queues form or lines run smoothly. Crest's team manually rebalanced every two to three campaign blocks as bottlenecks drifted between forming and packaging. No single machine is the permanent bottleneck — the constraint shifts depending on which product class is running. A heavy coating campaign saturates both enrobers while forming and packaging have available capacity; a baked-uncoated campaign shifts the bottleneck to the tunnel oven and packaging lines. Manual rebalancing requires the planner to reassign jobs one by one and estimate the effect on completion time.
• Schantt handles it by treating machine assignment as an optimisation variable across every stage's parallel pool. In both Auto and Semi-Auto modes, the simulation evaluates how each assignment affects upstream material supply and downstream wait time, then selects the assignment that minimises total production time. On the Gantt, the planner sees which machine each operation runs on, plus wait-material segments where a downstream stage is starved — making the bottleneck visible without manual calculation.

4. Packaging line changeovers.

• Crest's two HFFS (horizontal form-fill-seal) flow-wrap lines run all three product classes, with 20-minute changeovers between any two class pairs — a symmetric format-change penalty for adjusting film, sealing temperature, and package dimensions. At a packaging throughput of 6,000 bars per hour per line, each 20-minute changeover costs 2,000 bars of lost output per line. With weekly product rotations across three classes, the packaging schedule can accumulate 6 to 10 changeover events per week, losing 2 to 4 hours of packaging time.
• Schantt handles it through the same directional changeover mechanism. The 20-minute changeover is symmetric and applies to both packaging lines across all three product-class pairs. In Auto mode, the algorithm sequences jobs to reduce the number of packaging format switches where doing so shortens total production time. In Semi-Auto mode, the planner's fixed order determines the packaging sequence; the algorithm assigns each packaging operation to one of the two HFFS lines to balance the workload. The Gantt shows each changeover segment on the packaging line's row, making the total changeover time per line visible at a glance.

What to model in Schantt

Crest's configuration requires five top-level entities with the following counts:

The per-class routings (three routes, each visiting four of the five stages), directional changeover matrices (28 directional pairs across six machines), five stage-to-stage transfer times, three calendar exceptions, and two machine downtimes are configured on the detail pages of their parent entities.

Step-by-step setup

1. Create the five stages in production order. Set Blending as a batch stage and Forming, Baking, Enrobing, and Packaging as flow stages. On each stage's detail page, add the relevant transfer times between consecutive stages:

Transfer times:
- Blending to Forming: 5 minutes
- Forming to Baking: 5 minutes
- Forming to Enrobing: 15 minutes (pre-coat cooling bridge for coated classes)
- Baking to Packaging: 15 minutes (cooling conveyor bridge for baked class)
- Enrobing to Packaging: 5 minutes

2. Add the eight machines to their stages. Assign each machine to its stage:

Blending: Ribbon Blender / Jacketed Kettle. Forming: Extrusion Line 1, Extrusion Line 2. Baking: Multi-Zone Tunnel Oven. Enrobing: Universal Enrober 1, Universal Enrober 2. Packaging: HFFS Flow-Wrap Line 1, HFFS Flow-Wrap Line 2.

3. Create the three product classes and define each class's routing. For each product class, add the stages it visits — the skipped stages are simply left out of the class's routing. The stage order in each class's routing must follow the production flow: forming always comes after blending, packaging always comes last. Enable partial transfers on any handoff where the downstream stage can begin before the upstream stage finishes (Crest's cold-formed bars can be transferred promptly, so the default full-batch handoff applies). The transfer-time settings on each stage page link the routing stages and define how long bars spend in transit or in cooling dwell between successive stages.

Routing details:
- Whey Chocolate-Coated, Cold-Formed: Blending → Forming → Enrobing → Packaging (skips Baking)
- Plant-Based Uncoated, Baked: Blending → Forming → Baking → Packaging (skips Enrobing)
- Plant-Based Chocolate-Coated, Cold-Formed: Blending → Forming → Enrobing → Packaging (skips Baking)

4. Add one product per class. Create Chocolate Whey Crunch under the whey chocolate-coated class, Almond Oat Bake under the plant-based uncoated class, and Dark Cocoa Plant Bar under the plant-based chocolate-coated class.

5. Set each machine's capacity parameters and changeovers. On each machine's detail page, configure the batch parameters (for Blending) or throughput (for all flow stages) and the directional changeover matrix. This step depends on the product classes existing, so it comes after step 3.

Batch parameters (Blending):
- Batch size: 500 kg
- Cycle duration: 12 minutes

Throughputs (flow stages):
- Extrusion lines: 12,000 bars per hour (whey and plant-based chocolate-coated), 11,000 bars per hour (plant-based uncoated)
- Tunnel oven: 9,000 bars per hour (plant-based uncoated only)
- Enrobers: 12,000 bars per hour (both coated classes)
- HFFS packaging lines: 6,000 bars per hour (all three classes)

Changeover durations (directional):
- Forming lines: 75 minutes (whey to any plant-based), 65 minutes (plant-based to whey), 20 minutes (plant-based uncoated to plant-based chocolate-coated and reverse)
- Enrobers: 75 minutes (whey-chocolate to plant-based-chocolate), 65 minutes (reverse)
- Packaging lines: 20 minutes (symmetric across all class pairs)
- Blending: 0 minutes (all class pairs — identical batch parameters)

6. Configure the calendar, exceptions, and downtimes. Set the default calendar to Monday to Friday, two shifts, 06:00 to 22:00. Add the three annual exceptions (New Year's Day, International Workers' Day, year-end shutdown on 31 December) and the two maintenance windows (quarterly full-line clean-in-place deep-clean, 6 hours; annual enrober 1 overhaul, 8 hours).

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

Common mistakes

1. Using a single blanket changeover value per machine instead of directional pairs. A symmetric changeover (for example, 70 minutes both ways) loses the real asymmetry between whey-to-plant-based (75 minutes) and plant-based-to-whey (65 minutes) transitions. Fix: Enter each directional pair separately — the algorithm can then favour the shorter direction when it shortens total production time, and the directional Gantt segment shows the exact duration per transition.

2. Grouping all coated products into one product class. If chocolate-coated whey bars and chocolate-coated plant-based bars share a single class, they inherit the same routing, changeover structure, and machine parameters — but their enrober changeover times differ, and the allergen-driven forming changeover cannot be represented. Fix: Split by allergen category and coating type into separate product classes, each with its own routing and changeover matrix.

3. Configuring fewer machines than the actual floor layout. Crest runs two extrusion lines, two enrobers, and two packaging lines. If only one machine per parallel stage is configured, the algorithm cannot distribute workload across the second line, and the schedule will show unrealistic wait times at the single bottleneck. Fix: Add every production machine that a product class can route through, even if it is a standby or shared line.

4. Modeling a batch rack oven as a flow stage. A batch oven that loads trays and unloads after a fixed bake time has batch physics — a full-oven cycle, not a continuous throughput rate. If configured as a flow stage with a bars-per-hour rate, the timing of individual batches and their thermal recovery between loads is not represented correctly. Fix: If your facility uses a batch rack oven, set it as a batch stage with an appropriate batch size (trays per load) and cycle duration (bake time plus load and unload).

What a good schedule looks like

A well-configured Schantt schedule replaces Crest's weekly scramble to sequence coating runs and balance lines with a repeatable, optimised weekly plan. The schedule shows every production run on its assigned machine, the full coating-changeover sequence, and the resulting workload distribution across Crest's parallel lines at each stage.

Before (manual planning in spreadsheets): The planning team spent 6 to 10 hours per week managing enrober changeovers alone, losing 2 to 4 additional hours to packaging format changes. Bottleneck rebalancing happened reactively every two to three campaign blocks as the constraint drifted between forming and packaging without clear visibility. Coating-type interleaving was the default — the planning team lacked the time to manually sequence every changeover pair for minimal penalty. The weekly schedule was a best-effort compromise that left capacity on the table and required mid-week rework when actual production rates diverged from the plan.

• Enrober changeover time: 6 to 10 hours lost per week
• Packaging changeover time: 2 to 4 hours lost per week
• Bottleneck rebalancing: manual, every two to three campaign blocks
• Coating runs: interleaved, incurring 65-to-75-minute penalties multiple times per week

After (Schantt Semi-Auto mode with campaign ordering): Using Semi-Auto mode, Crest's planner sets the weekly campaign order — whey-coated first, then plant-based uncoated baked, then plant-based coated — and lets the algorithm assign machines and timing. Coating changeovers are clustered: two to three coating switches per week instead of six or more, saving hours of changeover time. The Gantt clearly shows each machine's operations with changeover segments, wait-material pauses, and working-hour boundaries, giving the planning team a schedule they can release to production with confidence.

• Enrober changeover time: reduced by clustering coating-homogeneous campaigns; each coating switch is visible as a labelled Gantt segment
• Packaging changeover time: reduced through campaign grouping — fewer format changes within each coating block
• Bottleneck visibility: wait-material segments on the Gantt show exactly where forming outruns packaging or packaging starves, without manual calculation
• Machine balance: both extrusion lines, both enrobers, and both packaging lines are assigned automatically, spreading workload across Crest's full capacity
• Schedule confidence: the planner inspects the Gantt once to confirm dough-to-oven timing and allergen sequencing, then releases the week's plan