157 towers, six families.
71% guyed suspension, the lightest available, with tension towers held to 16%.
When the Algorithm Designs the Line.
A 70 km line and a 20 km sectioning, bid as one lot. One optimization pipeline solved both, and took the conductor to 99.9% of its legal limit.
99.9%Of the regulatory headroom for electrical resistance used by the optimized conductor, with the limit never crossed.
2Distinct lines, single and double circuit, solved by the same optimization engine in one auction lot.
71%Of towers from a single guyed suspension family, the lightest structure available.
The problem.
A 70 km line and a 20 km sectioning, bid as one auction lot.
A transmission auction lot does not get judged one line at a time. It gets judged as a whole project, and this one was no different. Two distinct assets got bundled into a single bid: a 230 kV single-circuit line running about 70 km, and a shorter double-circuit section of roughly 20 km, each carrying its own technical requirements.
Normally, that means two separate design efforts, each carrying its own safety margin. No engineering team can push a routing, a tower choice and a conductor spec right up against a regulatory ceiling on two different line configurations and stay confident they have not crossed it. In an auction, that margin just makes the bid more expensive than it has to be, which automatically lowers the odds of winning it.
The solution.
One optimization pipeline, two lines, same rules.
The Digital Twin served as an internal simulation and mapping tool. It turned the territory between substations into a cost surface, a 10 to 30 m grid where every point carried a price covering terrain and slope, watercourses, protected areas, land use, buildings, anywhere legally off limits, with infinite cost wherever the line simply could not go.
From there, the cheapest path got searched two different ways. One pass went wide, scanning the whole feasibility area for any viable route. The other stayed tight, boxed into a 500 m corridor around the alignment the project had already planned. Once a route was chosen, the system refined it stretch by stretch, deciding what type of tower goes where, how tall it needs to be, and exactly where it sits.
Every proposal the algorithm produced then went through a technical validity checklist: catenary and swing, clearances, seasonal ampacity, electromagnetic fields, radio interference, audible noise. On top of that, wind margins came in above standard, a 50-year return period where the code only asked for 10, and 10 where it asked for 2. And all of it, from start to finish, ran the same way for both lines in the lot, the single circuit and the double circuit sectioning alike, with only the parameters changing.
The outcome.
157 towers, 71% from the lightest family, and a conductor running at 99.9% of its legal limit.
The lot’s main line came out to 157 towers across six structure families. 71% of them were a single guyed suspension type, the lightest one available. Guyed suspension won wherever the terrain allowed it, needing less steel and cheaper foundations than the alternative. Self-supporting towers only showed up where the mechanics genuinely demanded them, and tension towers, minimized by the routing itself, made up just 16% of the total. The average span reached about 470 m, past the 450 m the auction’s own pre-engineering had estimated.
Conductors and structures made up 79% of the total cost between them, conductors at 43.7% and steel structures at 35.6%, with right-of-way and clearing, foundations and complementary items splitting the rest. Total cost landed at about €76,500 per km.
The clearest sign of how far the optimization pushed showed up in the conductor’s electrical resistance: 0.04119 Ω/km, against an auction ceiling of 0.0412 Ω/km, using 99.9% of the allowed headroom. The same pattern turned up on the double-circuit sectioning too. A human designer would rarely get that close to a regulatory limit and still trust the number. The algorithm could, because it checked every constraint exactly instead of relying on the safety margins a person builds in out of habit.
None of that precision came at the expense of rigor. Seasonal ampacity was checked against IEEE Std 738 across all four rating windows, short-circuit withstand against the Sverak criterion and IEEE 80, insulation coordination at a 10⁻⁴ failure risk, and electric and magnetic fields, radio interference and audible noise all verified at the edge of the right-of-way. Between what the terrain allowed and what the regulation permitted, the design landed on the exact point where both were satisfied.
~€76,500 per km.
Conductors at 43.7% and steel structures at 35.6%, together 79% of total cost.
0.04119 Ω/km.
Against an auction ceiling of 0.0412 Ω/km, checked exactly instead of by habit.
Conductors (phase, shield wire and OPGW)
43.7%
Structures (steel)
35.6%
Right-of-way and clearing
12.9%
Complementary items
4.3%
Foundations
3.5%
In the consolidated scenario, total cost came to ≈€76,500 per km. The hierarchy explains the objective function: conductor selection and tower family and height dominate the economic outcome.