Reading Parallel and Alternate Branches on the EDGEBIC Canvas

On the EDGEBIC routing canvas, the arrow between two nodes is not decoration: its type is the difference between "then do this," "do this at the same time," and "do this instead." Three connector types carry three completely different scheduling behaviors, and reading them correctly is what separates a diagram you can trust from a picture that merely looks organized.

This post covers how each branch type is drawn, how to confirm what you are looking at, and what the engine does with it. For the plain chain-and-dependency reading that comes first, start with reading dependencies on the design canvas in EDGEBIC by User Solutions.

Three Connector Types, Three Meanings

Right-click any connector in the designer and the menu offers Connector Type with three choices, plus Delete Connector.

Routing links stay Normal. The other two attach an extra machine to an existing step, which is why the menu refuses to convert some links: if the target is an end product node, a product node, or a work center already sitting in a sequence chain, the link must remain sequential. That refusal is a shape warning, not a missing feature.

Reading a True Alternative Branch

A true alternative is a substitute. "Mill it on Mill-1, but if Mill-1 is booked solid, Mill-2 is acceptable." The scheduler evaluates the candidates and books exactly one machine. It is not extra capacity running alongside.

On the canvas, an alternative is a node joined with a True Alternative connector, and selecting it shows a True Alternative Work Center banner in the Properties panel. In the data grid the same thing appears as a child row under the step, marked with a circular-arrow glyph. Two views, one routing, as always.

The critical detail is that an alternative carries its own setup and hours, not a scaled copy of the primary's. An older, slower mill gets larger values, and the scheduler uses them directly. That is what makes the selection meaningful:

Mill-1 booked 3 days   Mill step waits          starts Thu, 51.0 h
Mill-2 free tomorrow   1.0 setup + 0.60 x 100   starts Wed, 61.0 h
                       Mill-2 finishes about 1.5 days EARLIER despite being slower

When a schedule run touches steps with alternatives, EDGEBIC asks how to decide: Auto with a selection strategy, or Manual with a per-step choice. The three strategies are Earliest Completion (fastest outcome wins), Primary First (stay on the primary while it has same-day capacity), and Earliest Start (whichever machine can begin soonest). On a reschedule, a step that already has actuals stays locked to the machine that actually ran it.

Reading a Parallel Branch

A parallel machine runs at the same time as the primary on the same step, and there are two fundamentally different kinds hiding behind one dashed connector. The Properties panel is where you tell them apart, through the Finite Capacity checkbox.

Independent Parallel (Finite Capacity), ticked. Two machines split one batch. All members must start together, each then runs its own capacity-checked share, and the wall-clock time drops. Fifty-one hours of milling on one mill becomes roughly 25.5 hours on each of two. The group waits for the first moment every member is simultaneously free, so it will not start one mill alone.

Dependent Parallel (Factor Based), unticked. A synchronized mirror. The extra machine copies the primary's exact start and end, and books the primary's hours multiplied by the Parallel Factor (1.0 for equal work, 0.5 for half). It does not shorten anything. It records that a second machine is physically committed to the same window, which is the right model for multi-spindle drilling, synchronized welding fixtures, and paired presses.

One consequence to know before you use mirrors: a dependent mirror is booked without checking that machine's own load. If a mirror machine also carries independent work, it can end up committed twice in the same window, and the resource load view will show it above 100 percent. Keep mirror machines dedicated in the affected periods and watch the scheduler anomalies report.

A parallel child's Properties panel edits only Parallel Factor and Finite Capacity; everything else inherits from the parent step, and the banner says so. If you find yourself unable to change hours on a node, that banner is the reason: edit the parent.

Hover Before You Edit

The node boxes on the canvas deliberately show a single title so large diagrams stay readable. Everything else lives in the tooltip and the Properties panel: sequence, setup time, hours required, queue time, flow step, instance counts, and the node's role as a primary step, a parallel copy, a true alternative, or a sub-assembly.

Editing the wrong node is the classic mistake in a diagram with branches, and one hover prevents it. The full tooltip vocabulary is in what a routing node tooltip tells you.

Alternates, Parallels, or a Pool?

Three tools sound similar and are not:

• True alternative when the substitution is specific to one operation. "This mill step may run on Mill-2 instead."
• Independent parallel when you want throughput on one operation. "Both mills work this step at once."
• A work center group when the substitution is general. "Any mill in the pool may take this step, plant-wide," which is better modeled once as a group than hand-configured on every routing step.

Choosing wrongly is not fatal, but it shows up as maintenance: hand-configured alternates on forty steps is the sign a group was the right answer.

What Happens on the Next Run

Nothing on the Gantt moves when you save the diagram. These are routing fields, and the engine reads them on the next scheduling run. Then:

• A new alternative brings up the selection dialog, and steps may book onto the alternate.
• A new independent parallel shortens the operation's wall clock and makes the group wait for a common free slot.
• A new dependent parallel adds mirror rows with identical start and end on the extra machines, leaving the primary's dates unchanged.

Bars that then overlap on the chart are usually correct rather than broken, especially where lot streaming is also configured. Saving a routing and rescheduling are two separate acts, which is the same discipline as save versus re-schedule on the Gantt.

The Diagram Should Answer the Question

The reason to draw a routing rather than only tabulate it is that branch structure is spatial. A mirror, a split, and a substitution look different from three feet away, and a missing connector is obvious in a way a broken cell value never is.

User Solutions has been modeling real shop routings since 1991, in ten-person job shops and in organizations including the US Navy, GE, BAE Systems, and Cummins, where the Nimitz program alone ran more than 26,000 tasks. Complex routings are readable when the picture is honest about which machines are alternatives and which are running together, which is the heart of finite capacity scheduling.

See the full EDGEBIC platform, start from the visual scheduling pillar guide, or bring your most branched routing to a demo and let US draw it on the canvas with you.