The Golden Gate Bridge hums?

Turns out that modifying the design of a bridge after it's built is full of surprises. A bridge that hums to itself is one of those consequences...

The Golden Gate bridge has had a couple of post-building updates that didn’t quite turn out the way people thought they would. 

The Golden Gate bridge from Marin, looking back at San Francisco. (P/C Arnet Xavier, Pexels)

Built between 1933 and 1937, by 1951, swaying of the bridge in the breeze caused the engineers to add horizontal trusses on the bottoms of the two vertical trusses.  This turned the bottom of the bridge into a boxwork structure that completely solved the swaying problem by making the structure much stiffer. Good job.

That was a great solution until there was a talk of extending the Bay Area Rapid Transit (BART) line across the bridge and into Marin County. That was a great idea, except the additional trusswork added a significant amount of weight—so much that adding tracks and railcars to the bridge would have exceeded the safe carrying capacity, and the plans had to be dropped. 

Then, in 2014, the decision was made to add a protective net to the bridge to deter suicides from leaping off the deck and into the swirling waters over 700 feet (213 meters) below.  The net was to be put 20 feet (6 meters) below the sidewalk and extend out 20 feet (6 meters) as well. Woven from stainless steel rope, the netting would be extensions of the vertical struts and be all but invisible from street level. 

But as you’d expect, the weight of the net and the additional struts required that the bridge lose some weight elsewhere to compensate and stay within the margin of safety.  Part of the project was then to reduce the weight of the railing along the walkway.  The original railing was made of H-shaped balusters (the vertical part) connected to a substantial handrail.  The new, weight-saving design swapped the hefty H-balusters for simple vertical plates that would be far lighter but would also let much more of the famed San Francisco ocean breezes to blow thru the railing.

This would have been fine, except that the average wind speed across the bridge went up from 10 mph to around 15 mph, which is a very difficult crosswind when you’re riding a bike through the gale.  Since the winds at the Golden Gate can gust to 75 miles (121 kilometers) per hour, cross breezes are an important factor in the bridge’s design. 

What’s even more concerning, during certain winds, a very loud hum at 440 Hz (the pitch an orchestra tunes to) would emanate from the bridge.  (The hum was at 100 dB, the same as the sound of a jackhammer 1 meter away!) It was unpredictable and could be heard many times of day or night from as far as 1 mile (2 km) away during episodes of high winds.  [Kukura]

These unanticipated consequences caused multiple delays in the safety net installation as the engineers searched for a solution.  The hum problem workaround was found to be adding thin, U-shaped clips to tie rubber inserts over the windward face of the slate, which reduces the noise, but not the wind flowthrough.  The fixes needed to be done to reduce the annoying hum to satisfy those that lived nearby, and also those who biked across the bridge. The fix solves the hum, but not the crosswind. Bicyclists still need to be cautious.  [Petroski]

The thing is, bridge engineers are very, very aware of designing bridges with enough strength to handle all expected loads and all anticipated winds—every bridge designer can tell you the names, places, and dates of bridges that have failed spectacularly when weights and winds were accommodated. All bridge engineers have seen the video of the Tacoma Narrows Bridge twisting and vibrating itself to pieces in 1940.

But after the initial bridge design was completed in 1929, few people anticipated the need to also carry subway traffic over the bridge. In 1951, when winds through the channel caused the deck to sway and twist, worried engineers thought that the surface would become unstable and break itself in the same way Tacoma Narrows Bridge was torn apart by vibrations in damaging winds. But adding those few tons of trusses in 1951 cut into the engineering margin of safety, just enough to prevent adding any other additional loads, such as the BART train extension across the bridge.

Likewise, when the railings were changed, the wind load would have been calculated and tested extensively.  Apparently though, the baluster design wasn’t tested at full size, so the resonant properties of the slats weren’t apparent as a consequence of the design.  [Kukura]

And the bridge just kept humming along, much to the annoyance of those who lived within earshot.

(Check out the YouTube video of it humming…)

— REFERENCES —

[Kukura]  Kukura, Joe. “Engineers furiously trying to fix Golden Gate Bridge’s constant humming sound.” SFIST (May 17, 2021) https://sfist.com/2021/05/17/engineers-furiously-trying-to-fix-golden-gate-bridges-constant-humming-sound/

[Petroski] Petroski, Henry. "Altering an Icon: San Francisco's Golden Gate Bridge has undergone several retrofits for increased stability and safety--sometimes with unintended consequences." American Scientist 111.1 (2023): 24-28. https://www.americanscientist.org/article/altering-an-icon