Network Rail and Bam start work on 1 September on four bridges over roads in Forest Gate, east London. Talbot Road and Stracey Road first, then Strode Road and Latimer Road on 28 September. The four structures are 130 years old, they carry the London Overground Suffragette line between Leytonstone High Road and Woodgrange Park, and the replacements — as New Civil Engineer reports — are expected to last until around the year 2150.

That figure is the story. Not the closure, not the contractor. A stated design life of roughly 125 years, written down, in a press release, before the first deck comes out.

What a number like 2150 actually commits you to

I know the arithmetic in my own steel. A span that carries traffic for 130 years outlives its signalling three times over, outlives the fastener catalogue, outlives every engineer who stamped the drawing. Network Rail’s own statement is careful about why these four are going: the old bridges are monitored closely to check they are safe, but they need replacing to avoid unplanned disruption and keep everyone moving reliably in the future. That is the honest reason. Not collapse. Unscheduled loss of the path.

Because that is what a life-expired bridge costs a network — not a failure, a timetable. My worst days were never structural. They were the minute I could not offer.

←TODAY: Four 130-year-old bridges come out in spring 2027, replaced by decks specified to reach 2150.
→3012: The asset that survives is the one whose replacement window was designed in, not discovered.
Fulcrum: A 125-year design life is a promise to a maintainer who is not yet born — and it is only real if you also hand them the access to keep it.

The sequence is the engineering

Read the programme carefully, because it is a masterclass in working over a live railway. Phase one is not demolition. It is extending the support structures at the ends of the existing bridges so they can carry the new decks — abutment work done while trains still run above and traffic still passes below. Only then, in spring 2027, do the old decks come out and the new spans go in. Completion is late summer 2027; weekend closures on the Suffragette line follow through 2027, plus shorter road closures across the area, with the road and pavement under each bridge shut for the duration.

So the load path is transferred before anything is removed. New bearing shelves, then the swap. Two years of preparation for a weekend of lifting. Joe Kennedy, infrastructure director for Network Rail Anglia, framed it plainly: thousands of people cross these bridges every day by train, through residential east London, and the four new ones will serve “several generations.”

Where this lands on your desk

You are probably not detailing a rail bridge. But you are specifying a façade, a roof build-up, a services riser — and almost none of it carries a written design life or a stated replacement route. PAZ has covered the cost of that silence: our piece on the Embarcadero Freeway traced how a double-deck viaduct optimised for one variable — throughput, under the 1948 Trafficways Plan — failed on every variable left outside the model, and came down in 1991 after just 32 years. Forest Gate is the inverse discipline: name the horizon, then design the sequence that reaches it without shutting the city.

Atelier: Most offices carry design life implicitly, buried in a product datasheet nobody opens after tender. That is where it goes missing — and where the client’s facility manager inherits a component with no documented replacement access. Monday move: add a design life and a replacement access column to your material and component schedule for the current project, and fill both for the ten highest-cost line items. Ten rows, one hour. The gaps you cannot fill are your real risk register.

Hack: Cost a replacement cycle before you argue about the cheaper option. A component with a 30-year life inside a 125-year asset is bought four extra times, and only discounted future cash makes that comparison honest. Run the whole-life sum in three lines of Python.

asset_life, comp_life, unit_cost, rate = 125, 30, 48000, 0.02
years = range(0, asset_life, comp_life)
print(sum(unit_cost / (1 + rate) ** y for y in years))
# 48000-CHF component, 30-yr life, 2% real discount -> ~131k over 125 years

Change comp_life to 60 and the number drops by roughly a third. That delta is the argument you take to the client — not “better quality,” but a figure with a discount rate attached.

The thing I would ask for

My warning is not about bridges falling down. Falling bridges get inspected. It is about the quiet loss of the second path: the diversionary route closed to save a budget, the maintenance window traded away, the access hatch detailed shut because it spoiled an elevation. Network Rail can do this work at all because the Suffragette line has weekend closures it can spend and neighbouring capacity to absorb them. Build that slack out of a system and the next 130-year replacement becomes impossible to schedule at any price.

Open your current component schedule and write the design life next to the ten most expensive items. Where the cell stays empty, you have found the thing your building cannot survive.

Source: New Civil Engineer (UK civil/infra)