What production scheduling looks like in a specialty chemicals plant: before and after software
Scheduling batch runs in a specialty chemicals plant is one of the hardest operational problems in the industry. Here is why, and what changes when you have software that actually understands how a campaign-based chemical plant works.
What production scheduling looks like in a specialty chemicals plant: before and after software
Scheduling batch runs in a specialty chemicals plant is one of the hardest operational problems in the industry. Here is why, and what changes when you have software that actually understands how a campaign-based chemical plant works.
What makes specialty chemicals scheduling hard
Specialty chemicals is not discrete manufacturing. You are not scheduling units down an assembly line. You are scheduling campaigns: runs of related products that share equipment, raw materials and cleaning regimes, sequenced to minimise changeover loss and meet delivery commitments across dozens of customer accounts.
Every campaign decision is bounded by hard physical constraints. Reactors and tanks have fixed volumes, fixed materials of construction, and fixed allowable chemistries. A reactor that ran a fluorinated intermediate yesterday cannot run an amine product today without a documented cleaning cycle, and that cycle takes hours, sometimes shifts.
Changeover cleaning is itself a scheduling problem. The validated cleaning between product A and product B may be different from the cleaning between A and C. Some sequences are cheap, others wipe out a day. The planner who can see the sequence-dependent setup matrix in their head is the planner whose plant runs efficiently. Most of the time, that person is one experienced individual.
Raw material availability layers on top. A specialty intermediate with a 14-week lead time arrives in a single drum, and the campaign that consumes it has to land in the window before the next drum is due. Miss the window, the reactor sits idle. Schedule too aggressively, you run out mid-batch.
Then there is shelf life. Some intermediates are stable for months, others for days. A bulk that should have been filled on Tuesday is scrap by Friday. The schedule has to know.
The before state
In most specialty chemicals plants, the schedule lives in spreadsheets, whiteboards and the head of the most experienced planner on shift.
The spreadsheet captures the next two weeks of planned campaigns. It is rebuilt every Monday morning. It is out of date by Tuesday lunchtime because two customer orders moved, a raw material delivery slipped and reactor 3 went offline for an unplanned maintenance window.
The whiteboard in the planning office holds the live picture. Coloured magnets, hand-drawn arrows, sticky notes for changeovers. The whiteboard is the truth. The spreadsheet is what gets emailed to commercial when they ask for an update.
The planner holds everything else: which sequences burn cleaning time, which customer will accept a one-week slip and which will not, which reactor the analytical team prefers for the high-purity grade because the historical impurity profile is cleanest.
What goes wrong is predictable. The planner takes a week off and the schedule degrades. A new product is introduced and the cleaning matrix is incomplete, so the first campaign is sequenced into a six-hour wash. Commercial promises a delivery to a strategic account that requires a cancelled campaign somewhere else, and nobody can see the knock-on until the substitution has already burned a tank. Margin per campaign is calculated quarterly from the financial close, by which point the bad decisions have already been made.
What software needs to understand about chemicals scheduling
Generic ERP scheduling modules were designed for discrete manufacturing. They schedule work orders down work centres on a calendar. They do not understand campaigns, they do not understand sequence-dependent changeovers, and they do not understand the difference between a reactor and a fill line.
Software that works in a specialty chemicals plant has to model the plant the way the plant actually runs:
- Campaign logic. A campaign is a sequence of related batches sharing equipment and a cleaning state, not a single work order. The system needs to schedule campaigns as first-class objects, not approximate them with linked orders.
- Constraint-based sequencing. The scheduler needs the sequence-dependent setup matrix in data, not in the planner's head. It needs to know that A → B is two hours and A → C is twelve, and to sequence accordingly.
- Integration with procurement. When raw material X is due on the 14th, the campaign that consumes X cannot start on the 12th. The scheduler needs the live PO horizon, not yesterday's purchasing report.
- Integration with inventory. Bulk produced today consumes shelf life starting today. The scheduler needs to plan the fill against the bulk, not against an aspirational stock figure.
- Visibility of margin per campaign. The planner trading off two candidate sequences needs to see the margin implication of each, in the moment, not at quarter close.
What good scheduling software looks like in a chemical plant
In a chemical plant, good scheduling software is unglamorous. The planner opens it in the morning, sees the live state of every reactor, tank and fill line, sees the next four weeks of planned campaigns against constraints, and sees where commercial commitments collide with capacity, raw material or shelf life.
When a customer order moves, the impact ripples through the plan automatically and the planner sees which campaigns are at risk, in which order to address them, and what the margin cost of each option is. When a reactor goes down, the affected campaigns surface immediately with feasible resequencing options that respect cleaning constraints and material availability.
The scheduler is not a black box that produces a plan and asks the planner to trust it. It is a tool that gives the planner constraints, options and trade-offs, and lets the planner make the call. The judgement still sits with the human who knows the plant; the bookkeeping that previously consumed that judgement now sits with the software.
When the experienced planner takes a week off, the plant does not degrade. The picture is in the system, not in their head.
How Worldover handles it
Worldover models the specialty chemicals plant the way the plant runs: campaigns as first-class objects, sequence-dependent changeover matrices in data, reactor and tank constraints respected by default, and raw material and shelf-life horizons connected live to procurement and inventory.
When commercial moves an order, the affected campaigns and customers surface in one view, with resequencing options ranked by margin and feasibility. When a unit goes down, the same view shows what can absorb the work and what cannot. Willow drafts the resequencing actions for the planner to approve, rather than the planner rebuilding the spreadsheet from scratch.
Per-campaign margin updates every shift, because the cost data sits on the same record as the production data. The picture lives in the system, the judgement stays with your planner, and the plant runs without depending on one person's memory.