Chemical Processing: Scheduling Continuous and Batch in One Plan

Chemical processing scheduling has to cover two worlds at once: continuous reactors and lines where output is a stream, and discrete filling and packaging where output is a count of units. The overlap model that fits one is wrong for the other, and most chemical products pass through both. EDGEBIC by User Solutions sets the model per station: an hours-based start-to-start lag for continuous equipment, a piece-count transfer batch for discrete steps, so one plan matches how the product physically runs end to end.

For the continuous-process capability, see scheduling ovens, baths and other continuous-process equipment. For the overlap models in full, see lot streaming explained and what a transfer batch is. For the sector view, see chemical manufacturing scheduling. Pair this with scheduling battery continuous lines and foundry lot scheduling. The full map is at how different industries use EDGEBIC.

Two worlds in one product

A typical chemical product passes through equipment that behaves in fundamentally different ways. The front end is continuous: reactors, blending vessels, extruders, columns, where the output is a stream and the concept of a discrete piece does not apply. The back end is discrete: filling, capping, labeling and packaging, where output is a count of units you can tally.

A schedule that uses one overlap model for the whole plant gets one end wrong. Apply a piece-count model to a reactor and it means nothing, because there are no pieces to count on a stream. Apply a wait-for-the-whole-batch model to a filling line and it serializes steps that physically overlap. And a schedule that ignores overlap entirely, making every step wait for the previous to finish the whole lot, is too long on both ends.

The plant needs a schedule that chooses the right overlap model per station: continuous where the output streams, discrete where it counts.

Continuous stations: the hours-based lag

For continuous equipment, EDGEBIC uses a start-to-start lag measured in hours. A downstream step may begin a set number of hours after the upstream step starts, independent of quantity, run time or setup. A lag of one hour on a reaction step means blending can begin one hour after reaction starts, overlapping the two rather than waiting for reaction to finish.

You steer the engine to this model by marking the station as continuous process. That flag tells the engine the piece-count model does not apply to steps routed there, so it uses the hours-based lag instead. Reactors, blending vessels, columns and extruders all belong in this category, because their output is a continuous stream.

The flag also protects against a silent misconfiguration. If a piece count is entered on a step running on a continuous station, it has no effect, and a report flags exactly that so you are not misled into thinking piece-count overlap is active. When the lag fires, it sets the downstream start directly and replaces any queue buffer on that step; if you need a physical hold on top, such as a settling or degassing gap, you add it as a transfer delay in hours so it composes with the lag rather than being discarded.

Discrete stations: the piece-count transfer batch

For the discrete back end, EDGEBIC uses a piece-count transfer batch. Once the product is filled into discrete units, packaging can begin as soon as a set number of units has accumulated, rather than after the whole batch is filled. You set that number as the transfer batch on the filling step, and packaging starts when the units are ready.

This is the right model where you can count output. Filling produces units; a case-pack needs a certain number of them before it can run; the transfer batch expresses exactly that threshold. A physical handling delay can be added on top in hours where the units need to settle or cure before packing. And a sensible guard applies: if the transfer batch is set larger than the order quantity, the engine caps it at the order size so a small run does not wait for units that will never exist, and a report flags that case.

A worked continuous-and-batch product

Consider a product that reacts, blends, fills and packs.

• The reactor is marked continuous process. The reaction runs for several hours.
• Blending is marked continuous process, set with a start-to-start lag so it begins a set number of hours after reaction starts, overlapping the two.
• Filling is a discrete station. It produces units at a known rate.
• Packaging is set with a piece-count transfer batch, so it begins once enough filled units have accumulated.

The plan then flows the way the product physically runs. Blending overlaps reaction by its lag rather than waiting for the reactor to finish. Packaging overlaps filling by piece count rather than waiting for the whole batch to be filled. The continuous front end and the discrete back end each compress, and the total time for the product is materially shorter than a plan that serialized every step. Crucially, the model is chosen per station, so the reactor's stream and the filling line's units are each handled correctly in the same schedule.

Loading the plant against real capacity

The overlap models sit inside a finite-capacity plan that loads every station against its real hours. Reactors, blenders, filling lines and packaging lines each have their own capacity, and the finite-capacity engine will not stack work beyond what a station can hold. When a reactor or a shared blending vessel is the constraint, its capacity sets the beat for the whole plan, and you can identify and schedule around it as the bottleneck. Products that share a reactor or a filling line compete for it honestly, so the plan shows the real contention rather than assuming infinite equipment.

For products committed to a ship date, the plan can be pulled to finish just in time; for continuous production feeding stock, the goal is usually to keep the reactors loaded and the product flowing. Either way, the schedule reflects the mix of continuous and discrete equipment rather than forcing one model on all of it.

Shared equipment is where the finite-capacity discipline pays off most in a chemical plant. A reactor or a blending vessel that several products pass through is a genuine constraint, and a schedule that assumes it can serve everyone at once produces a plan the plant cannot run. Because the engine loads that reactor against its real hours and lets only one job hold it at a time, the plan shows exactly when two products would collide on it, days before they do. You can then resequence, move one product to an alternate vessel if you have one, or accept the wait with eyes open. The same is true of a filling line shared across packaged goods: the plan surfaces the contention rather than hiding it, which is what turns a schedule from a wish list into something the plant floor can actually follow.

Common mistakes in chemical scheduling

A reactor left unmarked. If a continuous station is not flagged, the engine may try to apply a piece count or default to waiting for the full batch. Mark every reactor, blender, column and extruder as continuous process.

A piece count on a continuous station. It has no effect and only creates confusion; the engine ignores it and a report flags it. Put piece counts on the discrete filling and packaging steps, and use the hours-based lag on continuous stations.

A queue buffer expected to survive the lag. When the lag fires it replaces the queue-adjusted end for that step, so a settling queue set alongside a lag is discarded. Use a transfer delay in hours for a physical hold on top of the overlap.

A transfer batch larger than the order on a filling step. It cancels the overlap and behaves like waiting for the whole batch. The engine caps it and flags it; lower the batch for small runs to get real overlap.

Rolling it out

1. Classify each station by whether its output is a continuous stream or a count of discrete units.
2. Mark the continuous stations as continuous process, and set an hours-based lag on each continuous handoff from how the equipment physically feeds.
3. Set piece-count transfer batches on the discrete filling and packaging steps, with handling delays in hours where units need to settle.
4. Confirm every station carries its real capacity, so the whole plant loads honestly.
5. Run the schedule and read a full product, confirming the continuous steps overlap by their lags and the discrete steps overlap by piece count.
6. Check the report for any piece count sitting on a continuous station, and clear it.

Fermenters and stills are the same shape of problem on a longer clock: tank scheduling for craft brewing and distilling models each vessel as a single load held for its full cycle.

Bring one product routing that spans reactors and packaging to a demo of manufacturing scheduling software, and we will build the continuous-and-batch plan with you.