Why Your Spring Broke In November (Every Year)

In November 2025 I replaced 41 broken torsion springs. In May of the same year I replaced 9.

That's not anecdote. That's my shop calendar. Same number of houses in Richmond. Same number of garage doors. November had four and a half times the spring failures of May. And November 2024 looked exactly the same. And November 2023. And every November I've worked since I started apprenticing in 2018.

There's a reason, and it's not bad luck. It's three reasons stacked on top of each other.

Reason one: cold steel is brittle steel

Most residential torsion springs are made from oil-tempered carbon steel. ASTM A229 is the standard spec. The steel has a published yield strength and a published fatigue life, both measured at standard room temperature — call it 20°C.

Drop the steel to 2°C and a few things happen at the metallurgical level. The grain structure stays the same, but the steel becomes slightly less ductile. The ductile-to-brittle transition isn't dramatic for spring steel — this isn't WWII Liberty ship territory — but it's measurable. Springs that have been quietly accumulating micro-cracks all summer are more likely to propagate those cracks into a complete fracture when the steel is cold.

Add to that: the spring contracts when it cools. A residential torsion spring shrinks about 1.2 millimetres in length between a 20°C summer afternoon and a 2°C November morning. That doesn't sound like much. But the spring is pre-loaded — it's wound up to a specific number of turns to balance the door. When the steel contracts, the effective tension changes, and the stress distribution along the coil changes with it.

The weakest coil pays for that change first.

Reason two: November is the cycle-count cliff

Garage door springs are rated in cycles. One cycle = one open + one close. The most common residential spring you'll buy off a builder's spec sheet is rated for 10,000 cycles.

A family with two cars uses the garage roughly 4 to 6 times a day. Call it 5. Five cycles × 365 days = 1,825 cycles a year. Ten-thousand-cycle spring ÷ 1,825 cycles per year = 5.4 years of life.

But most doors I open are older than 5.4 years. The math should mean they've all already failed. Why haven't they?

Two reasons. First, families don't open their garage 5 times a day, 365 days a year. They open it more in summer (camping, beach, the boat at Steveston Harbour) and less in winter. Second, springs are slightly over-spec'd at the factory. A "10,000-cycle" spring usually has another 1,500 to 3,000 cycles in it before complete failure.

So real-world residential spring life clusters around 7 to 12 years for stock springs. That means we're constantly replacing springs from doors installed in 2014, 2015, 2016, 2017, 2018.

Now layer on this: most Richmond homes get their first big-volume garage use of the year in late spring through fall — yard work, vacation prep, the kids in and out. By October, an aging spring has just completed its highest-use months. The fatigue cracks that started in July are now mature.

Then the temperature drops. The steel contracts. The brittle threshold gets closer.

The spring breaks on the first cold Tuesday in November.

Reason three: Richmond is corrosive

This is where Richmond fails harder than the rest of the Lower Mainland.

Richmond is built on the Fraser River delta. We are surrounded on three sides by salt water. Steveston is 400 metres from the sea. Sea Island is the airport. Terra Nova backs onto the dyke. The west and south sides of the island take direct onshore wind year-round, and that wind carries microscopic salt aerosol inland for kilometres.

Salt aerosol settles on steel. In our humidity (annual average around 80% RH at YVR), the salt holds moisture against the metal. The metal corrodes from the surface inward. Each tiny pit is a stress concentrator — the spring fails earlier at that point than the bulk metal would.

I have replaced 5-year-old springs in Steveston that looked 12-year-old. I have replaced 12-year-old springs in McLennan (further from the water) that looked 6-year-old.

If you live within sight of water, your springs work harder than a manufacturer's cycle rating implies.

So why November specifically?

Stack the three:

Steel is slightly more brittle in cold weather.
Springs have just come off their highest-use season and are at peak fatigue.
Richmond's humid, salty air has been pitting the surface for years.

The first cold morning of the season is the day all three line up. That morning is, historically, somewhere between October 28 and November 18 in Richmond. Look at any November weather chart and you'll see the first dawn under 3°C falls in there almost every year.

That's the spring's last day.

Cost reality check (2026, Lower Mainland CAD) - Single torsion spring replacement (one spring, residential): $245–$385 - Pair replacement (recommended — they age together): $385–$625 - Upgrade to IPPC-90 corrosion-coated, 25,000-cycle spring: add $120–$220 - After-hours / emergency same-day: add $185–$285 - In Steveston, Burkeville, or Sea Island, where corrosion is worst, I strongly recommend the corrosion-coated 25,000-cycle upgrade. The math: $200 upfront for roughly double the lifespan in a corrosive environment is the cheapest insurance you'll buy this decade.

What you do about it

Schedule the inspection in October. Not November. October. Spring failures in October are 80% predictable from a five-minute visual inspection. The two-second test in post 1 catches most of them. A real inspection catches the rest.

Upgrade to IPPC-90 springs if you're near the water. They cost more. They last longer in Richmond air. Math works out.

Replace springs in pairs. When one breaks, the other is on the same fatigue curve. You'll be paying me to come back in 6 to 14 months for the second one — plus another full service-call charge — if you don't.

Don't try to "get one more winter out of it." I have heard this sentence approximately 400 times. It means: pay double in February when the spring breaks during the freezing rain warning and you can't get the car out for work.

A specific Terra Nova call from last year

A nurse on Fentiman Place. She'd been hearing a tick on every door cycle since August. Husband said it was probably fine. November 4, 6:40 a.m., she's leaving for an early shift at Richmond Hospital. Spring lets go. The bang is loud enough that her neighbour calls her from the next driveway.

She calls me at 6:52. I get there at 7:30. New pair of IPPC-90 springs, balanced, lubed, out by 9:15. She makes the back half of her shift.

If she'd had me look at it in August when the tick started, it would have been a $325 scheduled job at her convenience. In November, on emergency response, it was $675.

Two hundred and fifty dollars and a missed shift is the price of "probably fine."

That's the post.