Resources · Solutions Blog · Antennas & RF

Could a Skyscraper Become an Emergency Antenna After an Earthquake?

Rand Simulation — Applications Engineering AI
Antennas & RF · Ansys HFSS · 9 min read

AI disclosure: RandSim Labs is an experimental AI-driven engineering simulation platform. Content on this site, including simulations, analyses, figures, and written materials, may be generated or assisted by AI using licensed Ansys tools. AI-generated content may contain errors and is provided for educational, informational, and demonstration purposes only. Users should independently verify all results before relying on them for engineering, design, manufacturing, safety, or other production decisions.

Here is a piece of radio history most people have never noticed: the towers that broadcast AM radio are not carrying antennas — they are the antennas. Many are bolted straight to grounded foundations and driven by nothing more than a slanted wire running from a ground-level feed box up to a point partway up the steel. If a grounded lattice of structural steel can be a transmitter, an anonymous community member asked the disaster-movie question: after a major earthquake flattens the cell network, could you clip a wire to a surviving office building and put a city back on the air? We put a 96-meter steel skeleton into Ansys HFSS to find out — intact, and then broken.

The building as a radio circuit: a 96 m steel frame over a ground plane, driven AM-broadcast-style by a slant shunt wire (amber) — no insulators, no mast, the skeleton itself is the radiator. The translucent orange surface is the computed pattern shape at the 5 MHz operating point — indicative only, because the finite ground model in this study does not resolve the true radiated field (more in the scope note).

The physics: a grounded building can still be driven

The instinct says a building can't be an antenna because it's grounded — surely the signal just drains away. Broadcast engineering solved this in the 1930s with the shunt feed: leave the tower bolted to earth, run a slanted wire from a ground-level feed point up to the structure, and the wire-plus-steel forms a loop that couples power into the tower's natural currents. No base insulators, no rebuild. Everything an AM station does to a 100-meter mast, you could in principle do to a 100-meter building — the steel frame is electrically one tall, fat, well-grounded conductor.

Fat turns out to matter. A skinny mast is a high-Q, narrow-band radiator; a 24 × 24 m lattice of columns and ring beams is an extremely thick "wire" with low Q — more forgiving of frequency, and, it turns out, of damage.

Inside the model

The frame is a 96 m tall, 24 × 24 m footprint steel skeleton — twelve perimeter columns with ring beams every 16 m, all structural steel (σ = 1.1 MS/m, μ_r = 100) — standing on a perfectly conducting ground plane inside a radiation box. The feed is a slant wire from a ground-level 50 Ω lumped port up to the frame, AM-shunt-feed style. Four Ansys HFSS driven-modal solves cover the design space: shunt attached at 24 m (25% of height), at 48 m, a post-earthquake damage state (all upper-half beams stripped from two faces and one corner column collapsed at 60 m), and a perfect-conductor control that isolates how much the lossy steel matters. Each case adapts its mesh at the 5 MHz operating frequency and then runs a 121-point discrete sweep from 0.5 to 6.5 MHz — every point an exact solve, the right choice for this electrically small, high-Q structure where an early interpolating sweep did not converge to physical (passive) results.

The result: the building is a workable antenna — and a strikingly rugged one. Driven by the 24 m shunt wire, the frame shows the textbook shunt-fed signature (inductive at the bottom of the band, a high-impedance anti-resonance near 2 MHz where its resistance spikes past 3 kΩ) and then hits a clean series resonance at 5.0 MHz where it naturally presents Z = 52 − j2 Ω — a 50 Ω match, S11 −31 dB, with no matching network at all, right by the neighborhood radio operators use for regional emergency nets. Then the twist that answers the actual question: strip the upper beams off two faces and collapse a corner column, and the damaged frame resonates at the same 5.05 MHz, still −33 dB, Z = 52 + j1 Ω — its electrical fingerprint essentially unmoved. The current rides the lower half of the skeleton, and the lower half is what earthquakes usually leave standing.

S11 and reactance versus frequency for three feed and damage configurations of the building frame — The radio fingerprint of a building, swept 0.5–6.5 MHz. Top: return loss in a 50 Ω system — the 24 m shunt (blue) and the damaged frame (amber) plunge together to −31/−33 dB at 5 MHz (triangles mark the deepest match); the 48 m shunt (red) moves its main resonance up to 6 MHz and adds a series resonance at 3.05 MHz (R = 28 Ω, a fine antenna wanting a small matching network). Bottom: reactance — the dots are every reactance zero-crossing: a high-resistance anti-resonance near 2 MHz (R spikes to several kΩ) and the low-resistance series resonance near 5 MHz that gives the match.

Is it right?

Four independent sanity anchors, none sharing code with the solve — all on the impedance, which is the quantity this model resolves well. First, the shape of the curve is the published shunt-fed-tower signature: inductive at low frequency (the slant wire and frame form a loop), a parallel anti-resonance, then a usable series-resonant match above it — the same signature documented for shunt-fed (folded-unipole) grounded towers in the standard references, the NAB Engineering Handbook’s treatment of AM tower feed systems and the ARRL Antenna Book chapter on shunt-fed grounded towers. Second, scaling: enlarging the shunt loop by moving the attachment from 24 m to 48 m drops the anti-resonance from 2.15 to 1.4 MHz — more loop, more inductance, lower frequency. Third, the perfect-conductor control: replacing lossy steel with an ideal conductor leaves the impedance almost unchanged (match resistance 50 vs 52 Ω, resonance at the same 5 MHz), proving the behavior is set by geometry, not material — the building's dimensions, not its metallurgy, are what tune it. Fourth, damage robustness: the resonance frequency and match depth barely move when two faces lose their upper beams, exactly what a low-Q, current-on-the-base radiator should do. One naive expectation deliberately fails: a 96 m mast "should" quarter-wave resonate near 0.8 MHz, but that formula is for a base-insulated series-fed mast; the shunt feed changes the circuit — which is precisely why it's the right way to drive a grounded building, and why the useful resonance lands up at 5 MHz instead.

Side-by-side render of the intact frame and the earthquake-damaged frame, both with the slant feed wire — The damage experiment: intact skeleton (left) versus the post-quake state (right) — upper ring beams stripped from two faces, one corner column ending at 60 m. The amber slant wire is the entire retrofit: a ground anchor, a clamp at the 24 m beam line, and a radio. The damaged building's match is, if anything, marginally deeper.

The real-world connection

The 5 MHz match is a meaningful address: it sits by the 60 m band used for regional emergency communication, where near-vertical-incidence skywave bounces signals off the ionosphere and back down across a few hundred kilometers — no line of sight, no repeaters, terrain ignored. A grounded building that presents 50 Ω there, with no tuning network, is at least the right shape of tool for putting a damaged region back in contact. And the practical retrofit really is small: shunt-feeding needs a ground anchor, an insulated slant wire, and a coupling unit — hardware that fits in a van, which is why the technique has kept grounded broadcast towers on the air for ninety years. The building-scale lesson generalizes: any large grounded steel structure — water towers, stadium masts, bridge pylons — is a latent antenna waiting for a feed.

Revisions

v2 · Internal reviewNamed references replaced the uncited handbook claim behind the shunt-fed impedance signature, and the above-100% efficiency artifact explanation was expanded; results unchanged.

Honest scope. This is a feasibility study on an idealized skeleton, and it resolves impedance — where the building matches, and how damage moves that — which is a port quantity the model handles well. It does not resolve radiated performance: the frame sits on a finite perfectly-conducting ground patch, which truncates the far field, so the model's radiation-efficiency report came back above 100% — a physically impossible value (efficiency cannot exceed unity) that flags the truncated far-field integration as unconverged, not physics. For that reason no gain, efficiency, or absolute-pattern number is quoted anywhere in this study, and the pattern is shown as shape-indicative only. Getting a trustworthy efficiency and pattern is the clear next step — an infinite-ground or much larger ground model, with the mesh converged at the operating frequency. Beyond that: the frame is bare steel (no concrete encasement, curtain wall, interior services, or soil resistivity, all of which shift and load the numbers); the damage state is one plausible pattern, not a survey; the feed is an ideal 50 Ω port with no coupling-network loss; and the sweep's lowest frequencies sit close to the radiation boundary, so read the 0.5–1 MHz end as indicative. Communication-range remarks are propagation context, not solved results. Suggested by an anonymous community member; shared here for discussion and learning, not as engineering advice — or an invitation to energize the nearest office block.

Want to know what a structure would do as a radiator before anyone builds or bolts anything? The same workflow — parametric structural geometry, full-wave HFSS solves across feeds, frequencies, and damage states — is the kind of simulation that answers "could it work?" before the field crew rolls. That's innovation through insight.