Anyone who wants to renew, extend or simply maintain an overhead line system properly needs reliable data about what is already there. On most tram and trolleybus networks, that record exists in the form that was standard for decades: 2D drawings. Carefully drawn, technically dense – but hard to verify, difficult to evaluate, and far from what clients increasingly expect.

Pressure is building from two sides. First, BIM is becoming a requirement: Swiss Federal Railways have been tendering construction projects with BIM since 2024, planning with it since 2025, and intend to build with it from 2028, with commissioning and operation following through 2031. For tram and trolleybus networks there is no comparable requirement. Second, the maintenance backlog has grown: the 2025 network condition report puts it at 9.5 billion Swiss francs for SBB alone, 12 per cent more than the year before. Anyone who has to prioritise what gets renewed needs one thing first: reliable as-built data.

The good news: much of that data already exists. It sits inside the 2D drawings – it only has to be extracted.

A 2D drawing is a database in graphic form

An overhead line drawing is not merely a picture with a few labels. Pole schedules, support point data and plan text already carry a large part of what a component model needs: pole lengths and types, foundation dimensions, bracket geometry, attachment heights, tensile loads. A well-maintained as-built drawing is a database in graphic form.

Which leads to something essential: the work is not modelling. Nobody has to rebuild thousands of poles by hand. The work is decoding and verifying – reading the information encoded in the drawing in a structured way and securing its meaning before anything is built on it.

The model is not the goal – the asset record is

Our route from the 2D drawing therefore runs through an intermediate stage: a structured asset dataset. All supports, attachment points, wires and special components are first extracted from the CAD drawing with their attributes – as data, not as a drawing. The 3D model is generated from that dataset afterwards. The model is one output of the record, not the purpose of the exercise.

That intermediate step is not a detour. It is where the value sits:

Cross-checks instead of assumptions

The most delicate part of any as-built digitisation is interpretation: what does this attribute actually mean? From which reference point is this height measured? Guess here, and you build in errors nobody will ever spot – the model looks complete, after all.

The answer lies in the drawing itself: its data is internally redundant. The same quantity often appears more than once – once as an attribute, once in the geometry. That is exactly what we use to cross-check every interpretation systematically rather than believe it. Misreadings and contradictions surface that way before they enter the digital asset record.

Which set of cross-checks works on a given drawing standard depends on how that standard is built – that part of our work we demonstrate on a reference section rather than describe on the internet. The rule behind it we are happy to state: whatever fails a cross-check does not enter the asset record unexamined. Checks like these surface false assumptions before they travel further. That is what separates a verifiable record from a pretty picture.

How much detail? Levels of use, not quality grades

Not every task needs the same information depth – and more is not automatically better. Following the principle of Level of Information Need as set out in EN 17412-1, depth follows purpose: as much as necessary, as little as possible. We work with three levels of use across the asset life cycle, not quality grades – each is complete for its purpose, and every object carries its level with it:

Paying for survey accuracy where a register would do buys precision nobody uses. And a thirty-year-old drawing does not become survey-accurate just because a model is derived from it. So we state the level per object: derived from the drawing is not the same as surveyed. Moving to the next level is an extension, not a fresh start.

Terrain and buildings come as open data

For spatial context, Switzerland offers high-quality open geodata. Since swisstopo releases its digital standard products as Open Government Data, the swissALTI3D elevation model, the swissBUILDINGS3D building models and further datasets are available at no cost, commercial use included.

What matters is holding the height logic of the overhead line consistently – in the drawing everything refers to the carriageway. With the elevation model properly underneath, the model carries those references through, and special cases such as embankments show up in the checks. Details like these decide whether a model later withstands clash detection or merely looks good.

Example: exceptional transport routes against your requirements

What a clean asset dataset makes possible is best shown by a check that almost falls out of it: wherever a route for exceptional transports passes under the overhead line, the contact wire height has to be right. We check that across the network and deliver the list of critical crossings with target, actual and difference – together with the route in the model.

Two things decide whether such a check is worth anything. First: the target value comes from your own requirements, not from a traffic dataset – a clearance height recorded there is the headroom at a structure, not a target for the overhead line. If your value is missing, we report that instead of grading it. Second: the assumptions are stated. The assumptions and data sources used are set out in the report – including where they work in favour of the result. Critical locations can then be examined deliberately instead of disappearing behind a green light.

What you end up with

And the effort scales the right way: the decoding work occurs once per drawing standard, not once per drawing. Once the conventions of a standard are calibrated, every further drawing of the same origin follows the same path – from a single section to an entire network. After a rebuild, updating the record is not a new project.

Two routes, one data model

Opening up existing records is one route. The other is new design work: support point concept, pole and foundation selection according to the operator's standards, assembly and material lists, drawing set – from preliminary design to detailed design. Both run in the same data model at BayTech.

That has an often underrated side effect: what is designed today is already part of the record on the day it goes into service – no second data capture. In new design work the same ladder runs top down: model and drawing set come first, the asset dataset falls out along the way.

Your data, your systems

The asset dataset is independent of the drawing software and survives any change of tool – including a change of service provider. Handover is in open formats: as IFC, in common GIS formats and as a table; integration with your existing GIS, BIM or asset management environment is agreed with you. You are not buying an island solution: you need no BayTech licence to keep working with the data we deliver.

Where we stand

Transparency is part of the deal: this service is under active development at BayTech. The tool chain from extraction through the asset dataset to the 3D model is in place – the component libraries we are extending continuously. Pole types, foundation tables and standard heights differ between operators not only in their values but in the selection logic itself, and every drawing standard needs its one-time calibration. Come in early and you feed the requirements of your own network straight into the ongoing development, and help decide which standards and component catalogues are covered first.

Start with one section, not a programme

Going digital with the overhead line requires neither a fresh start nor a network-wide resurvey. It does not even require a decision in principle: one existing section of one to two kilometres is enough to see what can reliably be won from your drawings – and what cannot. You supply the drawings and your standards, we process the section – an old or decommissioned one is very welcome, because the more varied the drawing conventions, the more the result tells you about your archive.

The most valuable data source for your overhead line has been in the archive all along. Evaluate it in a structured way, cross-check it consistently and combine it with the open geodata, and you get exactly the basis that renewal planning, asset management and BIM requirements will demand in the years ahead.