NX Electrical Routing Automation at a Special Machinery Builder: Three Tools, One Continuous Harness Process
This NX electrical routing automation case study shows how a special machinery builder turned an hours-long, error-prone harness hand-off into a matter of minutes — a change the client confirmed for anonymous publication. Three NX Open tools carry the process: a reuse library manager that connects the NX Reuse Library with Teamcenter and a third-party electrical engineering system, an exchange manager that runs PLMXML import and export as one controlled action, and a segmentation tool that separates stocks, overstocks and fixing elements such as cable ties so that every segment can be measured exactly. About 50 users work with the toolchain.
The story here is the unglamorous end of harness design. The routing itself was never the client’s problem — Siemens NX handles that well. The problem was everything around it: where components come from, how the design travels to the electrical side and back, and whether the lengths that purchasing orders against can be trusted. The client reports that the full hand-off pass per routed assembly went from several hours to a few minutes, and that PLMXML export errors went from a recurring problem to rare exceptions.
A special machinery builder now runs electrical routing in Siemens NX as one continuous process: components come from a single governed reuse library, PLMXML exchange with the electrical side runs as one managed action, and segments are separated automatically at the end of routing so their lengths can be measured exactly. The client reports the full hand-off pass went from several hours to a few minutes, and PLMXML export errors became rare exceptions. About 50 users work with the routing toolchain.
LAST UPDATED AUGUST 25, 2026 · FIGURES CONFIRMED BY THE CLIENT FOR ANONYMOUS PUBLICATION
- Client
- Special machinery builderNOT NAMED
- Sector
- Special machinery
- Users
- About 50 users of the routing toolchain
- Scope
- Three NX Open tools — reuse library management, PLMXML import/export, segment separation for length determination
- Environment
- Siemens NX with Routing (Electrical), Teamcenter, a third-party electrical engineering system
- Our role
- NX Open development — inteliScience SRL
The situation: lengths nobody could trust, exchanges nobody wanted to run
In special machinery, almost every machine order is a variant — and almost every variant carries its own wire harness. The harness is designed in the NX Routing (Electrical) environment, but a routed model on its own is not a manufacturing document. Before anyone can cut wire or order stock, three things have to be true: the components in the model match what the company actually builds with, the design data survives the round trip to the electrical engineering side, and the length of every segment is exact.
None of the three — component source, electrical round trip, segment lengths — was reliable. Routing components lived in several places at once: the NX Reuse Library, Teamcenter and a third-party electrical engineering system, with no automatic alignment between them, so nothing guaranteed that the three agreed. PLMXML import and export, the exchange that connects mechanical routing with the electrical side, was a sequence of manual steps, and the client confirms that export errors were a recurring problem. And because stocks, overstocks and fixing elements such as cable ties sat together in the routed model, segment lengths could not be measured cleanly — deriving the figures for manufacturing and purchasing was a manual pass that, as the client confirms, took several hours per routed assembly.
The industry context says this is not one company’s bad luck. In the CAN Newsletter (1/2020), Ulrike Hoff and Dan Scott (Siemens) put the usual lead time for a wiring harness, from design release to product delivery, at approximately 23 to 26 weeks — and name the elimination of manual steps, with information cascading automatically between steps, as the way to shorten it. Writing in Wiring Harness News, John Judkins (Siemens) puts numbers on the same idea: a model-based approach with a continuous digital thread can cut design errors by 50 percent and formboard design time by up to 85 percent. Both figures are industry context from named sources, not this project’s results — the project’s own figures are below and carry their own stamps.
What changed: before and after
- manualPick routing components from three libraries with no alignment between them
- manualRun PLMXML import and export as a sequence of individual steps
- fragileExport errors are found late, not where they are caused
- by handDerive segment lengths from a model where stocks, overstocks and cable ties sit together
- repeatDo all of it again for the next machine order
- governedComponents come from one reuse library, aligned across NX, Teamcenter and the electrical system
- one actionPLMXML import and export run as a single managed step
- reliableExchange errors are rare exceptions, confirmed by the client
- exactSegments are separated automatically at the end of routing, so every length is measurable
- repeatableThe same routed assembly produces the same lengths and the same exchange, every time
What the client reports
The figures below come from the client and were confirmed for anonymous publication; they describe what the routing team observes in production use, not a controlled study, and the stamps say so.
Confirmed by the client for anonymous publication.
The full routing-to-manufacturing hand-off per routed assembly
client-confirmed Separating segments, determining lengths and exchanging the data used to take several hours of manual work; with the three tools the client reports a few minutes. No exact before and after times were released, so this is stated as the client stated it. The bar indicates the order of magnitude described, not a measured ratio.
PLMXML export errors
client-confirmed Qualitative statement, no metric attached. No error rate was released, so no percentage is claimed. The client confirms that exchange failures are now rare exceptions rather than an expected part of the process.
Users working with the automated routing toolchain
client-confirmed The tools are in production use across the client’s routing user base.
How do three tools make one process?
Three tools make one process by covering the three points where harness data used to leak: where components enter the design, where the design crosses to the electrical side, and where lengths leave for manufacturing. Each is an NX Open application, each is useful on its own, and together they close the loop.
One source of truth for routing components
A library manager creates and maintains a local reuse library and keeps it aligned with Teamcenter and the third-party electrical engineering system, so the NX Reuse Library, the PLM system and the electrical tool no longer disagree. Designers pick from one governed catalogue instead of three unaligned ones.
OUTPUT — consistent components at the start of every routing jobThe round trip to the electrical side becomes one managed action
An exchange manager wraps PLMXML import and export in the routing environment: fewer manual steps, defined settings, a repeatable result. The exchange runs the same way every time, for every user.
OUTPUT — reliable data exchange between mechanical routing and electrical engineeringExact lengths fall out of the model, at the end of the process
As the final step of routing, the segmentation tool separates stocks, overstocks and fixing elements such as cable ties into individual segments, and every segment can then be measured correctly. Because the separation is rule-based and runs as a command, it can be repeated after a design change — the same assembly state always produces the same segments and the same lengths. The figures that manufacturing cuts to and purchasing orders against come from the model, not from a manual pass.
OUTPUT — trustworthy segment lengths for manufacturing and purchasingWorth saying plainly: there is no language model anywhere in this toolchain. Library alignment, exchange and segmentation are rules and geometry — they must be reproducible, and a reproducible rule should never be handed to a probabilistic system. Where we do use AI at inteliScience, it assists people; it does not decide what gets manufactured.
What the solution is built on
Supported Siemens extension points and the client’s own conventions — upgrades mean revalidation, not rewrites.
Related work on the same ground: NX Open custom tool development, NX workflow automation, and, for the programming side of this kind of work, NX Open programming for machine builders.
What we would do differently
We would settle the shared conventions before writing the first tool. All three tools touch the same objects — components, segments, attributes — and each needs the same answers: what counts as a segment, which attributes identify a component across NX, Teamcenter and the electrical system, which library is authoritative when two disagree. Writing those answers down first makes every tool that follows cheaper. We would also build an automated PLMXML round-trip test — export, re-import, compare — before the first production exchange: an exchange should prove itself in a test harness, not in production.
When does electrical routing automation pay off?
Electrical routing automation pays off when harness work repeats and its output feeds someone else. The case is strong when most machine orders carry their own harness variant, so the hand-off pass runs weekly rather than yearly; when manufacturing and purchasing consume the lengths directly, so a wrong figure becomes scrap or a wrong order; and when a separate electrical engineering system makes the PLMXML round trip a permanent part of the process. It is a weak case when harnesses are rare one-offs, when nobody downstream consumes exact lengths, or when there is no second system that speaks PLMXML to exchange with — then the manual pass, however tedious, is not where the money goes.
Frequently asked questions
Why do harness lengths from NX need segment separation at all?
What is PLMXML and why does the exchange break?
How long does a project like this take?
When would this not be worth it?
Why is the client not named?
Your purchasing orders are only as good as the lengths behind them.
Send us one routed assembly and the way you determine segment lengths today — we will show you how much of that hand-off is deterministic, on your own model.
Talk to an NX Routing Engineer