Production Scheduling for Water-based / Latex Adhesives

This guide shows production planners and plant managers how to schedule water-based and latex adhesive production across multiple reactors, blending tanks, and filling lines in Schantt — with sequence-dependent changeovers, cure hold windows, and mixed-shift calendars. You will learn how to model a hybrid-flowshop adhesive facility, configure the entity hierarchy, and produce optimised schedules that respect every constraint.

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

Industry context

Water-based and latex adhesives are formulated through emulsion polymerisation, then blended with additives and fillers, and finally filled into pails, drums, or totes for shipment. The chemistry is demanding: reaction temperature, monomer feed rate, and agitation must be tightly controlled during polymerisation, and many formulations require a cure hold after blending before the batch is ready for filling. The result is a hybrid-flowshop process where batch reactors and continuous fill lines operate under different calendar rules and different changeover penalties. Scheduling this flow means sequencing batches across four parallel reactors of different sizes, managing cleaning transitions between polymer families, and ensuring that filled product leaves the plant on time — all while a shared clean-in-place circuit and a single-shift quality lab add external gates the schedule must respect.

The typical plant runs several product families — pressure-sensitive adhesives (PSA) for labels and tapes, construction adhesives for flooring and tile, and woodworking adhesives such as PVA glue. Each family follows a different route through the same physical stages, and each imposes its own cleaning requirements on shared equipment. The cure hold between blending and filling is the critical timing constraint: too short and the adhesive has not reached specification; too long and the batch may drift out of quality tolerance and must be scrapped.

A representative SMB adhesive facility produces approximately 15,000 tonnes annually across roughly 200 active SKUs, with about 25 SKUs scheduled in a typical week. The reactor bank comprises four vessels (5, 10, 10, and 20 MT capacity), served by two 8 MT blending tanks and three filling lines dedicated to pails (80 units per hour), drums (30 per hour), and totes (6 per hour). Polymerisation cycles range from 4 to 10 hours per batch depending on the formulation, and blending adds 2 to 4 hours. The plant runs a shared clean-in-place circuit that serves three of the four reactors, so cleaning contention is a real constraint the schedule must account for. A one-shift quality-control lab releases batches on weekdays.

Valley Adhesives Ltd. runs approximately 85 people at a single facility, making three product classes across three production stages, scheduled by a two-person planning team.

Process overview

flowchart LR
    POLY["Polymerisation<br/>(Batch Stage)"]
    BLEND["Blending<br/>(Batch Stage)"]
    FILL["Filling<br/>(Flow Stage)"]

    POLY -- "PSA, Construction" --> BLEND
    POLY -- "Woodworking: skip" --> FILL
    BLEND -- "PSA: 12 h cure hold" --> FILL
    BLEND -- "Construction: direct" --> FILL

Process flow for water-based and latex adhesives at Valley Adhesives Ltd. PSA passes through all three stages with a 12-hour cure hold; Construction adhesives skip the cure hold; Woodworking adhesives skip Blending entirely.

PSA, Construction, and Woodworking adhesives follow different paths through the same stages. PSA passes through all three stages; Construction adhesives skip the cure hold and proceed directly to filling; Woodworking adhesives skip Blending and have a minimal cure hold.

Scheduling challenges and how Schantt handles them

In this scenario the schedule is driven by weekly customer demand — the planning team pulls the coming week's orders and builds a job list from them. If your plant is driven by make-to-stock replenishment or a seasonal sales forecast instead, the same model and constraints apply; only the job-list source changes.

Schantt optimises total production time, scheduling jobs forward from a start date. This guide assumes a practical horizon of one to two weeks, which matches the typical SMB adhesive plant's planning cycle. Longer horizons are supported but are less common in practice.

Schantt offers two scheduling modes for this kind of work. In Auto mode, the algorithm determines both the sequence of jobs and the machine assignment on each stage, exploring the full solution space to find the shortest total production time. In Semi-Auto mode, the planner locks the job order and lets the algorithm choose the best machine for each operation — useful when the planner has a preferred sequence based on customer relationships or raw-material availability.

What Schantt handles well

• Multi-machine batch scheduling with variable-size reactors — Schantt models each reactor as a machine on a batch stage with its own batch capacity and cycle time per product class. In Auto and Semi-Auto modes, the algorithm evaluates every eligible reactor for each job and assigns it to the one that minimises total production time.

• Sequence-dependent cleaning and changeovers — Directional changeover matrix per machine (class A to B can differ from B to A). The algorithm folds changeover time into each operation's start, favouring sequences that cluster similar classes. Each cleaning segment appears as its own labelled block on the Gantt.

• Mixed batch-and-flow pipelines — Batch reactors (fixed cycle, full load) and continuous fill lines (units per hour) in one route. The simulation chains downstream stages after upstream supply and emits wait-material segments when a fill line is starved.

• Shift-aware calendars per machine group — Reactors and fill lines get independent calendars (24/6 versus two-shift Monday to Friday). Work advances only in working windows; non-working stretches render as shaded Gantt overlays.

• Multi-product routing with stage skipping — Each product class has its own per-class routing. Stages absent from a class's routing produce no operation or Gantt row. Transfer times bridge across skipped spans.

• Calendar exceptions and downtime — Holidays, shutdowns, and maintenance windows modelled as team-wide exceptions or per-machine downtime. Both subtract from capacity and render as Gantt overlays.

How Schantt handles each challenge

1. Manual reactor assignment across variable-size vessels.

• With four reactors ranging from 5 to 20 MT capacity and three product classes each with different cycle times and eligibility rules, the planner spends two to three hours every week bin-packing jobs by hand. A mis-assigned batch — a large PSA order on an undersized reactor, for example — pushes the promised fill date out by the reactor's full cycle time. Woodworking adhesives add another layer: they cannot run on the largest reactor (R4, 20 MT), so the planner must keep track of which vessels are eligible for which jobs across every shift.

• Schantt models each reactor as a machine on the polymerisation stage, with its own batch capacity and cycle duration per product class. In Auto and Semi-Auto modes, the algorithm evaluates every eligible reactor for each job and assigns it to the one that minimises total production time. The planner reviews the assignment on the Gantt and, where needed, overrides it in Semi-Auto for edge cases such as reserved reactor capacity for a rush order.

2. Sequence-dependent cleaning across polymer classes.

• Switching between PSA, Construction, and Woodworking on a reactor requires a full clean-in-place cycle lasting four to six hours for cross-polymer transitions, with a lighter 15-to-30-minute rinse for same-class repeats. With three classes and six directional pairs per reactor, a single ill-sequenced transition — a PSA batch followed by a full clean — consumes an entire shift's worth of reactor time. Because the changeover from Woodworking to PSA (6 hours) is twice as long as the reverse (3 hours), sequence direction matters as much as the classes involved. The fill lines add their own format-change penalties: 15 to 60 minutes per line when switching between pail, drum, and tote configurations, and each line serves a different subset of classes.

• Schantt's directional changeover matrix captures each pair's specific duration — PSA-to-Construction differs from Construction-to-PSA on the same reactor. The algorithm folds every changeover into the operation's start time and favours sequences that cluster similar classes, turning what was a manual optimisation into an automatic decision. On the Gantt, each cleaning segment appears as its own labelled block with the changeover duration clearly visible. The planner manually staggers cleaning windows on reactors sharing a CIP circuit, since Schantt does not detect overlapping cleans on shared skids.

3. Cure hold windows between blending and filling.

• PSA requires a minimum 12-hour cure hold after blending, with a maximum window of 24 hours before the batch must reach the fill line. Woodworking adhesives tolerate a looser 0-to-72-hour hold. The weekend idle gap — reactors run around the clock but filling stops at 18:00 Saturday through 06:00 Monday — can push a Friday-finished PSA batch past its 24-hour maximum if not carefully timed. Valley Adhesives estimates roughly half of its approximately 300 tonnes of annual scrap comes from batches that missed their cure window.

• Schantt models the cure hold as a scheduled transfer time between stages — a minimum delay that elapses in continuous wall-clock time, 24 hours a day, regardless of shift boundaries. The algorithm respects this minimum hold when chaining the blend operation to its downstream fill operation. For the maximum window, the planner inspects each PSA batch on the Gantt to verify that the fill start falls within 24 hours from cure completion. Every operation's start and end times are visible, and batches approaching the limit are straightforward to spot against the deadline.

4. Weekend fill-line starvation and Monday morning backlog.

• The reactor calendar (24/6 extended coverage) and the fill-line calendar (two shifts, Monday to Friday) create a recurring weekend mismatch. Reactors keep producing through Saturday, but at 18:00 Saturday the blending and filling hall goes dark until Monday morning. The cure holds continue elapsing during the weekend — they run in wall-clock time — so by Monday some batches have already passed their 24-hour maximum window and are at risk of scrap before the fill hall even opens. Come Monday, four to seven batches are queued at the fill hall entrance, representing 25 to 40 hours of fill work. The supervisor spends roughly three hours every Monday morning sequencing this queue by hand, with limited visibility into which batches are urgent and which can wait.

• Schantt's separate calendars per machine group make this mismatch explicit in every schedule. When a fill line is starved because the upstream batch finished outside its working hours, the Gantt renders a wait-material segment that pinpoints exactly where and why the delay occurs. In Semi-Auto mode, the planner preserves Monday's fill order while letting the algorithm optimise machine assignments. The Gantt's shaded non-working overlays and wait-material segments turn Monday's manual puzzle into a visual queue that the planner can resolve in minutes rather than hours.

What to model in Schantt

These are the first-class entities you create to represent the adhesive facility in Schantt:

Step-by-step setup

Follow these steps in order. Each step builds on the previous one, and the product classes must exist before you configure machine-level parameters and changeovers.

1. Create the stages and set transfer times. Create Polymerisation (batch), Blending (batch), and Filling (flow) in that order. On each stage's detail page, set the transfer times between consecutive stages:

• Polymerisation to Blending: 30 minutes (pump transfer from reactor to blend tank)
• Blending to Filling: 720 minutes for PSA (12-hour cure hold); 0 minutes for Construction (direct, no hold)
• Polymerisation to Filling: 30 minutes (skip bridge for Woodworking, which bypasses Blending)

2. Add the machines to each stage. Assign the four reactors (R1, R2, R3, R4) to Polymerisation, the two blending tanks (B1, B2) to Blending, and the three fill lines (L1, L2, L3) to Filling.

3. Create product classes and define per-class routing. Create three product classes: PSA, Construction Adhesives, and Woodworking Adhesives. On each class's detail page, set the routing — the ordered sequence of stages the class passes through — and disable partial transfer on every routing leg (adhesive batches remain physically continuous through curing and blending; no staggered handoff).

• PSA: Polymerisation → Blending → Filling (all three stages)
• Construction: Polymerisation → Blending → Filling (all three stages, but cure hold is 0 minutes)
• Woodworking: Polymerisation → Filling (skips Blending; uses the bridge transfer)

4. Add one representative product per class. Create three products:

• Water-based acrylic label adhesive (PSA class)
• Acrylic latex flooring adhesive (Construction class)
• PVA wood glue (Woodworking class)

5. Set machine capacity parameters and changeovers. On each machine's detail page, enter the batch capacity and cycle duration for batch machines, or the line speed (throughput) for flow machines, per product class. Then configure the directional changeovers for every pair of classes that the machine processes.

Batch capacities and cycles (per eligible class × machine pair):

• Reactors R1–R3: 5 to 10 MT batch capacity, cycle times from 4 to 7 hours depending on class
• Reactor R4: 20 MT capacity, cycles from 7 to 8 hours; processes PSA and Construction only (Woodworking is incompatible)
• Blending tanks B1–B2: 8 MT capacity, cycles of 2.5 hours (PSA) or 3.5 hours (Construction)

Line speeds (per eligible class × machine pair):

• L1 (pail line): 80 units per hour for Construction and Woodworking
• L2 (drum line): 30 units per hour for PSA and Woodworking
• L3 (tote line): 6 units per hour for PSA and Construction

Changeover durations (directional, per machine):

• Reactors R1–R3: six directional class pairs each, ranging from 3 hours (PSA to Woodworking) to 6 hours (Woodworking to PSA)
• Reactor R4: two directional pairs (PSA to Construction and reverse), 4 to 5 hours
• Blending tanks B1–B2: two directional pairs, 45 to 60 minutes
• Fill lines L1–L3: two directional pairs each, 30 to 60 minutes per format switch

6. Configure calendars, exceptions, and downtimes. Create two calendars. The reactor calendar runs 24 hours Monday to Friday and 06:00 to 18:00 Saturday and Sunday (24/6). The fill lines calendar runs two shifts, 06:00 to 22:00, Monday to Friday only. Add two calendar exceptions — New Year's Day (1 January) and International Workers' Day (1 May) — as non-working days for the entire facility. Enter two machine downtimes: the year-end maintenance shutdown affecting all equipment (24 December through 1 January) and R4's quarterly boil-out (15 to 16 March).

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

Common mistakes

1. A single blanket changeover time instead of directional per-pair values. The cleaning duration between PSA and Construction differs from Construction to PSA on the same reactor, and Woodworking transitions are different still. A single average figure produces inaccurate sequencing and misleads the algorithm into accepting a suboptimal job order.
Fix: Configure the full directional matrix for every shared machine — each from-class-to-class pair that routes through it. Use the per-pair durations from your plant data.

2. Modelling the cure hold as a separate stage. Creating a dedicated cure stage between Blending and Filling adds an unnecessary entity and a phantom row on the Gantt. Cure is a time delay, not a process step with its own machine.
Fix: Represent the cure hold as a transfer time between the Blending and Filling stages. Set the minimum delay in minutes on the Stage detail page; check the maximum window manually on the Gantt.

3. One product class for all adhesives. If PSA, Construction, and Woodworking share a single class, the divergent routings cannot be modelled — Woodworking's Blending skip and PSA's cure hold become invisible to the algorithm.
Fix: Split into separate product classes per divergent routing path. The three classes capture the real flow differences.

4. Assigning the same calendar to all machines. Reactors running a 24/6 schedule and fill lines on two-shift weekdays produce very different availability patterns. A single calendar assigns unrealistic working windows and produces schedules that fail on the plant floor.
Fix: Create two calendars — one for the reactor bank and one for the blending-and-filling hall. Assign each machine to its correct calendar.

5. Omitting partial-transfer constraints on routing legs. With partial transfer enabled, the algorithm could start filling before the entire batch has cured, which is physically unrealistic for adhesive batches.
Fix: On each product class's detail page, disable partial transfer on every routing leg. The dataset's explicit false values guarantee that every batch stays physically continuous through all stages.

What a good schedule looks like

The difference between manual scheduling and a Schantt-powered schedule shows up across the entire planning week.

Before (manual bin-packing and Monday sequencing):

• Reactor assignment consumes two to three hours per week of hand-done bin-packing, with no algorithmic search for better machine fit.
• Cross-class cleaning transitions are unplanned; the fill hall discovers the sequence only when the first Monday batch arrives. A poorly clustered Monday can consume four to eight hours of cleaning within the first day.
• Monday morning requires roughly three hours of supervisory sequencing to untangle the queue of four to seven waiting batches, representing 25 to 40 hours of fill work.
• Fill line starvation is a recurring weekend phenomenon: reactors keep producing through Saturday while the fill hall sits idle, then the backlog materialises Monday morning.
• The cure-window scrap risk is invisible until a batch fails quality control; roughly half of the annual 300 tonnes of scrap comes from missed maximum windows.

After (Schantt Auto or Semi-Auto mode):

• The algorithm assigns every job to its best-fit reactor in seconds. The planner reviews the Gantt and overrides in Semi-Auto only when a specific machine must be reserved.
• Changeover time drops because the algorithm clusters similar classes into contiguous runs, grouping PSA batches together before switching to Construction or Woodworking. A Monday that once began with four hours of cross-class cleaning now opens with same-class sequences and short rinses.
• Monday's fill queue is visible on the Gantt before the weekend begins. The planner walks in to a ready-made schedule: Semi-Auto preserves the preferred fill order while Auto optimises the full sequence automatically.
• Wait-material segments flag every starved fill line by location and cause. The planner sees the weekend gap and can adjust the Friday schedule to pre-position the right batches.
• Every PSA batch's cure-then-fill chain is visible on the Gantt. The planner checks that each fill start falls within the 24-hour maximum, catching at-risk batches before they become scrap. The annual scrap burden from missed windows falls sharply.