Repair or Replace: How to Decide on Aging Building Railings

Somewhere between the first loose post and the full-building tender there is a decision a board or owner has to make, usually with incomplete information and always with someone in the room arguing that a repair will do. The decision is not fundamentally about money. It is about whether the existing assembly still has a coherent remaining service life, and whether the money you would spend on repairs buys any of it. Three questions answer that, in order.

Question one: is the structure and its anchorage still sound

Everything else is negotiable. This is not. If the posts, base plates and anchors can still develop the required capacity, you have an asset worth maintaining. If the anchorage is compromised — corroded embeds, spalled concrete at the slab edge, anchors that have lost embedment, base plates rotating under hand load — you are no longer discussing a railing repair. You are discussing structural remediation with a railing on top of it, and at that point replacing the guard while the slab edge is open is almost always the cheaper path.

The tell to watch for is corrosion jacking. When steel embedded in concrete corrodes it expands, and the expansion cracks the concrete from the inside. Rust staining below a base plate and hairline cracks radiating from anchor locations are early evidence of a process that only accelerates. Tightening the bolts on that condition does nothing except reset the clock on the next complaint.

Question two: can you still get the parts

This kills more repair strategies than corrosion does. Aluminum railing systems are built from proprietary extrusions, and dies get retired. A system installed in 1998 by a fabricator that has since been acquired, renamed or closed may have no available matching post, no matching base shoe, no matching top cap and no gasket that fits the channel. Glass is more forgiving in that a replacement panel can usually be made to size, but the hardware that holds it and the shoe it sits in may not be reproducible.

Before committing to a repair program, price a representative parts order. If a fabricator has to run a custom extrusion die to make forty replacement posts, the tooling cost alone can approach a meaningful fraction of the replacement cost, and you will have spent it on extending a system that is already at the end of its design life. Conversely, if the system is current, parts are catalogue items and the repair path opens up dramatically.

Question three: what happens to compliance

Existing guards are generally permitted to remain as built. But there is a threshold — different in wording across provinces, similar in effect — beyond which work stops being maintenance and starts being replacement, and replacement brings current requirements with it. If your existing guard is 1,000 mm tall with 125 mm openings, a like-for-like repair keeps that condition in service, while a replacement will need to meet the current minimum height, the 100 mm sphere requirement, and a non-climbable zone running roughly from 140 mm to 900 mm above the walking surface.

That is not an argument for repairing indefinitely. A guard with openings a small child can pass through is a real risk regardless of what the code allowed in 1984, and owners who knowingly maintain a non-compliant condition carry exposure that no maintenance budget offsets. It is an argument for making the decision deliberately and documenting the reasoning, rather than discovering the implications after the contract is signed.

Repair is the right answer when it buys you years. It is the wrong answer when all it buys is the interval until the next complaint.

When repair clearly wins

• Anchorage is sound and the deterioration is confined to seals, gaskets and setting blocks. Reglazing and resealing a guard line typically costs a fraction of replacement and can add many years to the assembly's service life, though how many depends heavily on exposure and on the condition of everything behind the seal.
• The finish has chalked or faded but the substrate is intact. Field refinishing costs far less than replacement, though colour and gloss uniformity is harder to achieve on site than in a factory line.
• Failures are isolated and random rather than clustered. Six broken panels across 400 on a fifteen-year-old building is attrition. Six broken panels on one elevation is a pattern.
• The system is current and parts are available off the shelf.
• The existing geometry already meets current code, so a repair carries no compliance penalty.
• The building has other envelope work scheduled in five to eight years and the guards can credibly be carried to that window and replaced as part of a coordinated project.

When replacement clearly wins

• Anchorage or slab edge deterioration is present on more than an isolated basis.
• Parts are unobtainable without custom tooling.
• The guard does not meet current height, opening or non-climbable requirements and the building's risk profile makes that untenable.
• Repair costs are accumulating year over year and the trend line is rising rather than flat.
• The building is already opening the slab edge for waterproofing, traffic coating or concrete repair. Removing and reinstalling existing guards to allow that work often costs 50 to 70 per cent of what new guards would cost installed, which makes reinstallation of a worn-out assembly poor value.
• The remaining components are within a few years of their own renewal point, so a repair simply resets a countdown that was already close to zero.

The arithmetic that settles the argument

Convert both options to annualized cost. Take the repair cost, divide by the years of additional service it credibly delivers, and compare that against the replacement cost divided by the service life of a new system. As rough 2026 planning figures: a one-off glass panel replacement including access typically falls somewhere around $900 to $2,500 depending on size, height and hardware availability; a building-wide reseal and gasket campaign commonly runs in the range of $30 to $80 per linear foot; field refinishing of aluminum components often lands in the range of $40 to $110 per linear foot; and full retrofit replacement of glass guards on an occupied building generally runs $400 to $850 per linear foot installed. All of those are approximate and vary substantially with access, quantity and region.

Run the numbers and a familiar shape appears. A reseal that costs, say, 8 per cent of replacement and buys fifteen years is an obviously good decision. A structural repair program that costs 45 per cent of replacement and buys seven years, on a system that will still be non-compliant afterward, is obviously not — even though it feels cheaper in the year you spend it. When the annualized numbers land close together, replace, because a new system generally carries a fuller and longer warranty than a repair does.

The hybrid option most buildings never consider

Between full repair and full replacement sits partial replacement, and it is underused. If the posts and anchorage are sound but the infill and hardware are finished, replacing infill only — swapping tired picket sections or degraded glass and its hardware while retaining the structural posts — can deliver much of the benefit at a fraction of the disruption. It works best where the existing posts were generously sized, where anchorage capacity has been verified by an engineer for the new configuration, and where the post spacing suits the new infill. It fails where the existing structure was designed to exactly the loads of the original, shorter guard, because a taller replacement raises the moment at the base.

The other underused hybrid is phasing by elevation according to exposure. Windward and salt-exposed faces deteriorate faster, and there is no rule requiring a building to replace all four elevations in the same year. Replace the failing face, maintain the others, and let the assessment data set the sequence. The one caution is that access and mobilization are large fixed costs, so phase in a way that reuses a single setup rather than paying for rigging four times.