Glass-and-steel railing post-installation deflection testing: why site load-testing data matters more than the shop drawing on a Bellandur high-rise retrofit

A 12mm frameless glass balcony railing on a mid-rise retrofit in Bellandur met every NBC deflection requirement on the shop drawing—0.8mm under a 100 kg point load at mid-span. On site, after the bolt-holes were drilled to site dimensions and the posts were fitted to the as-built concrete, the same load produced 1.4mm deflection. The substrate was stiffer than modelled, the bolt-hole tolerances stacked, and the installer had tightened the fixings at 9 a.m. in the cooler air before the monsoon humidity climbed. None of this was visible in the atelier drawing.

The difference between a passing shop drawing and a passing site load test is the difference between specification and reality. This is why deflection testing belongs on the handover checklist, not filed away after the shop drawing is approved.

The gap between shop drawing and site conditions

Glass railing deflection is calculated on the basis of assumed conditions: uniform substrate stiffness, bolt-hole clearances within tolerance, ambient temperature at 22°C, and no dynamic load factors. The NBC (National Building Code) permits deflection up to L/100 for glass railings, where L is the unsupported span. A 1200mm balcony railing is allowed 12mm deflection under the specified 1.5 kN point load. The shop drawing confirms this. The site is different.

Concrete substrate rigidity varies. A post bolted to a 150mm reinforced concrete slab at the edge of a balcony sits over a void; the effective stiffness is not the same as a post bolted to a solid core. Bolt-hole drilling tolerance is typically ±2mm on site. When two or three post-holes are drilled, the tolerance stack can reach ±4–6mm. A post that sits 3mm proud of the bolt circle shifts the load path. Temperature at the time of tightening matters: a bolt torqued at 6 a.m. in cooler Bangalore winter air (18–20°C) will have different clamping force than one torqued at 2 p.m. during the monsoon (32–35°C with 85% humidity). The atelier model assumes none of this variation.

Why load testing on site is not optional

The protocol: what to test and when

Load testing should be specified in the architect's handover conditions, not left to the railing supplier's discretion. The test is simple: apply a 100 kg point load (approximately 1 kN) at the centre of the most critical span—typically the longest unsupported glass panel between posts. Measure deflection with a dial gauge or laser displacement sensor, accurate to 0.1mm. Record the ambient temperature and relative humidity at the time of test. Repeat the test after 48 hours and again after 7 days. If deflection increases between the first and second test, the fixings are still settling; if it stabilises by day 7, the system is stable.

The test should be performed after the railing is fully installed and all fixings are torqued to specification, but before any handover sign-off. This timing matters: if the test reveals deflection beyond the NBC limit, there is still time to inspect the bolt-holes, re-torque, or adjust the post spacing before the resident moves in.

What the data tells you

A site load test produces three pieces of information: absolute deflection at day 1, deflection trend over 7 days, and the relationship between deflection and ambient conditions. On the Bellandur retrofit, the day-1 deflection of 1.4mm was 75% higher than the shop drawing predicted. The deflection stabilised by day 3, suggesting the bolt-holes had settled into their final position. By day 7, with monsoon humidity at 88% and temperature at 31°C, the deflection was 1.38mm—essentially unchanged. This told the architect that the system was stable, even if it was running hot against the NBC limit.

Had the deflection continued to increase, or had it exceeded 12mm, the railing would have failed the test and required remediation: re-drilling, shimming, or post-spacing adjustment. The cost of remediation after handover is five times the cost of addressing it before the resident takes possession.

Substrate and bolt-hole tolerance: the invisible variables

Concrete substrate quality in Bangalore retrofits varies widely. A balcony slab poured in the mid-2000s, before the current building boom, may have been specified at M25 or M30 concrete. A slab poured in the last five years is likely M35 or M40. The modulus of elasticity for M25 concrete is approximately 25 GPa; for M40 it is 32 GPa. This 28% difference in stiffness translates directly to a 28% difference in predicted deflection. If the shop drawing was based on M40 and the site substrate is M25, the railing will deflect more than predicted.

Bolt-hole drilling on site is done by hand or with a hand-held drill. Tolerance is ±2mm per hole. When a post is fitted with four M10 bolts, and each hole is drilled 1–2mm off-centre, the post can sit slightly off-plumb. This shifts the load path and increases local stress concentration, which increases deflection. A post that is 3mm off-plumb over a 1200mm span will deflect differently than one that is plumb.

The shop drawing cannot account for these variables because they are site-specific. The load test can, because it measures the actual installed system.

Temperature, humidity, and the monsoon effect

Bangalore's monsoon season (June to September) brings sustained humidity of 80–90% and temperatures of 28–34°C. Glass does not absorb moisture, but the steel components of the railing do. A steel post or bracket can expand and contract by 0.05–0.1% in length when humidity shifts from 40% to 90%. Over a 600mm post height, this is 0.3–0.6mm of expansion. The bolt clamping force can relax slightly as the steel expands, which can increase deflection by 0.2–0.3mm.

Load testing should be performed at least once during the monsoon season to capture this effect. A railing that passes the test in March may show different deflection behaviour in August. If the project handover is scheduled for the monsoon, the load test should be run in the monsoon. If the project is handed over before the monsoon, a second test during the first monsoon is a prudent safeguard.

On the Bellandur retrofit, the initial load test was run in March (temperature 28°C, humidity 65%). A follow-up test in July (temperature 33°C, humidity 87%) showed 0.15mm additional deflection. This was within acceptable variation, but it confirmed that the system's behaviour changed with the season. The architect added a note to the maintenance schedule: re-torque all railing fixings in June, before the monsoon humidity peaks.

Specifying load testing in the contract

The architect or project manager should specify load testing as a line item in the railing supply and installation contract. The specification should include:

• Load magnitude: 100 kg point load applied at the centre of the critical span.
• Measurement method: dial gauge or laser displacement sensor, accurate to 0.1mm.
• Test timing: day 1 (after installation and torquing), day 3, and day 7.
• Ambient conditions: record temperature and humidity at each test.
• Acceptance criteria: deflection must not exceed NBC limit (L/100 for the specified span).
• Remediation protocol: if the test fails, the supplier is responsible for identifying and correcting the cause at no additional cost to the client.
• Documentation: the supplier delivers a signed load-test report with measurements, dates, and ambient data, before handover sign-off.

This specification removes ambiguity. The railing supplier knows what is expected. The architect has a clear pass/fail criterion. The resident has documented proof that the railing meets code before they move in.

Common deflection failures and how to fix them

When a site load test reveals deflection beyond the NBC limit, the cause is usually one of three things: insufficient bolt torque, bolt-hole tolerance stack, or substrate stiffness lower than assumed.

Insufficient bolt torque is the easiest to fix. If the bolts were torqued to 25 Nm but the specification calls for 35 Nm, re-torque and re-test. Deflection typically drops 0.2–0.4mm. If re-torquing does not reduce deflection, the bolts may be bottoming out in the holes, which means the holes were drilled too deep or the substrate is softer than expected.

Bolt-hole tolerance stack is harder to fix. If the holes are drilled ±3mm off-centre, you cannot re-drill without creating new holes. The fix is to use larger-diameter bolts (e.g., M12 instead of M10) or to add a secondary support—a steel angle or tube that ties the post to the substrate at a different point, distributing the load over a larger area. This requires a shop-drawing revision and may add cost and time to the project.

Substrate stiffness that is lower than assumed requires a structural assessment. If the concrete is weaker than specified, the post may need to be re-seated on a larger base plate, or the post spacing may need to be reduced. Again, this is a remediation that is easier to address before handover than after.

The case for in-house load testing

Some architects and project managers prefer to hire an independent structural engineer to perform the load test, rather than relying on the railing supplier's own measurements. This adds cost (typically 8,000–15,000 rupees per test) but removes any perception of bias. If the railing supplier is also the one performing the test, there is an incentive to report a pass even if the data is marginal. An independent engineer has no such incentive.

For high-rise retrofits, where the railing is a safety-critical element and the resident is paying a premium for the installation, the cost of independent load testing is justified. For mid-rise projects or lower-risk installations, the supplier's own load-test report, if it is detailed and includes raw measurement data, is usually sufficient.

Questions we get asked

Does NBC require post-installation load testing for glass railings?

NBC does not explicitly mandate load testing after installation. The code requires that the railing meet deflection limits under a specified load, but it does not prescribe when or how that test is performed. It can be done on the shop drawing (via calculation or finite-element analysis) or on site (via physical load testing). However, best practice—and the standard followed by architects on high-value projects in Bangalore—is to perform both: verify the design on paper, then verify the installation on site. The site test catches variables that the drawing cannot.

What if the site load test shows deflection higher than the shop drawing predicted, but still under the NBC limit?

This is acceptable. The NBC limit is the upper bound; the shop-drawing prediction is an estimate. If the actual deflection is 1.4mm and the NBC limit is 12mm, the railing passes. However, you should investigate why the deflection is higher than predicted. Is the substrate softer than assumed? Are the bolt-holes off-plumb? Are the fixings not fully torqued? Understanding the cause helps you decide whether the railing is stable long-term or whether you need to monitor it more closely during the monsoon season.

Should we load-test every railing on a multi-unit project, or just one sample?

Test the railing on the unit that presents the most demanding conditions: the longest unsupported span, the highest floor (where wind loads are greatest), or the unit with the most exposed balcony. If that railing passes, you can reasonably assume that the others will too, provided they are installed using the same method and materials. However, if the project has multiple different railing configurations (e.g., balconies on different sides of the building, or railings at different heights), test at least one example of each configuration. The cost of testing is small compared to the cost of remediation or, worse, a safety incident.

Can we perform load testing before the glass is fully cured or the sealant has set?

No. Structural sealant (typically silicone or polyurethane) reaches 80% of its strength at 7 days and 100% strength at 28 days. If you load-test before the sealant is fully cured, the test does not represent the final installed condition. Wait 28 days after the railing is installed before performing the load test. This also allows time for any settling of the fixings to occur, so the day-1 test captures a more stable baseline.

If a railing fails the load test, who pays for remediation?

This should be defined in the contract. Standard practice is that the railing supplier is responsible for meeting the NBC deflection limit and for remediation if the site test reveals non-compliance. If the failure is due to site conditions that were not disclosed in the architectural drawings (e.g., concrete weaker than specified, or substrate geometry different from the as-built plan), the responsibility may be shared. This is why it is important to provide the railing supplier with accurate site dimensions, concrete specifications, and as-built drawings before the shop drawing is finalized.

Commissioning your load test

If you are specifying a glass railing for a Bangalore project—whether a retrofit in Bellandur, a new residential tower in Whitefield, or a renovation in Indiranagar—include post-installation load testing in your handover conditions. Ask the supplier for a detailed load-test protocol before you approve the shop drawing. Verify that the test will be performed at least once, ideally twice (once in the initial season, once during monsoon if the project timeline allows). Request raw measurement data and ambient conditions, not just a pass/fail letter. If the project budget allows, commission an independent structural engineer to observe and verify the test.

This is not bureaucracy. It is the difference between a railing that is specified to meet code and a railing that is proven to meet code under the actual conditions where the resident will use it. Talk to the atelier about commissioning a load-test protocol tailored to your project site and specifications.