Steel Silo Radiographic Weld Inspection: What We Actually Look For and Why It Matters

Radiographic weld inspection uses X-ray or gamma-ray imaging to detect internal defects in silo welds — cracks, porosity, lack of fusion, slag inclusions — without cutting anything open. For welded steel silos storing cement, clinker, coal, or fly ash, RT inspection on 10–25% of vertical and horizontal welds is standard practice, and it's the only method that gives you a permanent record of what's happening inside the weld metal. Here's what that looks like on a real project.

Why Radiography, Not Just Visual or Ultrasonic?

The defect types that'll kill a silo

Visual inspection catches surface cracks and undercut. That's maybe 15% of the problem. The defects that cause silo failures — the ones I've seen in fracture analyses — are internal: lack of sidewall fusion in vertical submerged arc welds, toe cracks buried under the reinforcement, slag lines running parallel to the fusion face. You can't see these. Ultrasonic testing finds them, but the record is an A-scan waveform that means nothing to anyone except the technician who shot it. Radiographic testing gives you a film or digital image that a third-party inspector, the client's engineer, and the contractor can all look at and argue over. That argument is the point.

I've been on projects where the client accepted UT results but rejected the same welds on RT. Not because UT was wrong — because the UT report didn't show enough detail to satisfy the insurance underwriter. RT doesn't have that problem.

When RT is mandatory vs. recommended

EN 14491 (dust explosion protection) doesn't explicitly require RT, but most specifiers tie weld quality to EN 1090 execution class, and EXC3 for silos means volumetric NDT on a defined percentage of welds. ASME Section VIII Division 1, UCS-57 requires full radiography for Category A and B joints in pressure vessels — and if your silo has a pneumatic conveying inlet operating above 0.5 bar gauge, some jurisdictions classify sections of it as a pressure boundary. In China, GB 50128-2014 (Code for Construction of Vertical Cylindrical Welded Steel Storage Tanks) specifies radiographic testing for tank wall butt welds with minimum 10% coverage for tanks over 5,000 m³. The exact percentage varies by code and client, but the trend is upward. I've seen specs demanding 100% RT on shell-to-bottom corner joints. That's expensive — about $12–18 per exposure on a 25mm wall — but it's cheaper than a leak.

The Physics in Practice: Sources, Sensitivity, and Setup

Iridium-192 vs. X-ray: which and when

For silo shell welds in the 8–30mm thickness range, Iridium-192 gamma sources are the workhorse. A 70 Ci Ir-192 source handles up to about 65mm of steel in a single exposure, gives good contrast on film, and the equipment is portable — one tech can carry a projector, film cassettes, and positioning magnets to the top of a 35m silo in a single trip. The downside: half-life is 74 days, so source activity drops fast. A source that was perfect for 20mm plate in week 2 is marginal for the same job in week 6. We've had to swap sources mid-project in Nigeria because the logistics chain couldn't get a replacement through customs in time.

X-ray machines (200–300 kV portable units) give better geometric unsharpness and thinner effective beam width. For critical joints — shell-to-bottom fillet welks on large-diameter silos, or nozzle reinforcement pad welds — I'll push for X-ray even though it's slower. The image quality difference at 18–25mm thickness is noticeable. IQI (Image Quality Indicator) sensitivity of 2-2T is achievable with both, but X-ray holds it more consistently on thinner sections.

Sensitivity, density, and what "acceptable" actually means

Film density matters more than most people realize. EN ISO 17636-1 requires a minimum density of 2.0 in the weld area (measured with a densitometer, not guessed). Below 2.0, you lose sensitivity to fine cracks. Above 4.0, you're overexposing and losing contrast in the toe region where cracks initiate. I've rejected films shot at density 1.6 — the tech said "it looks fine on the lightbox," but the wire IQI couldn't be resolved. That's a retake, which means another exposure, another radiation safety exclusion zone, another hour of lost time.

For digital radiography (CR or DR panels), the signal-to-noise ratio replaces density as the metric. A SNR of 70:1 is the minimum I'll accept for silo weld interpretation. DR panels are faster — image in 3–5 seconds vs. 5–15 minutes for film processing — but the spatial resolution is typically lower (100–200 μm pixel pitch vs. film grain at ~10 μm). For critical applications, film still wins. For production RT on hundreds of circumferential welds, digital is the only economically viable option.

Weld Joints in a Silo: Where to Shoot and What to Accept

Vertical welds vs. horizontal welds: different risks

Vertical welds in a silo shell are typically single-V or double-V butt joints, 8–25mm thick, done with submerged arc welding (SAW) on the ground during plate joining, or shielded metal arc welding (SMAW) during erection. These are the highest-quality welds in the structure — done in controlled positions, inspected before the plate goes up. RT coverage of 10–20% on vertical welds is standard. The defects we find here are usually slag inclusions from SAW flux entrapment or lack of sidewall fusion on thicker sections.

Horizontal welds (circumferential joints) are the problem children. These are field-welded, often in wind, sometimes in rain, with the welder standing on a platform that sways. The quality is inherently lower. I specify 25–50% RT on circumferential welds for silos over 20m diameter, and 100% on the first three passes of any new welder's work. The defect rate on horizontal field welds is typically 3–5x higher than on shop vertical welds. That's not a criticism — it's physics and ergonomics.

Acceptance criteria: EN, ASME, and the gray zone

EN ISO 5817 Level B (stringent) is the default for most silo projects in Europe and the Middle East. Under Level B, a single porosity indication up to 3mm diameter is acceptable in a 100mm weld length, but any crack — regardless of length — is a reject. Lack of fusion is always a reject. Slag inclusions up to 2mm thick and 10mm long are acceptable in isolation. ASME Section VIII uses a different framework: acceptance is based on indication type and length relative to thickness, with cracks and lack of fusion always rejected.

The gray zone is cluster porosity. Five pores in a 50mm length might each be individually acceptable, but clustered porosity can indicate a systematic gas shielding problem. I've written rejection reports for cluster porosity that technically met ISO 5817 Level B — because the pattern told me the welder was about to produce a much worse weld two joints later. That's engineering judgment, not code compliance. Some clients love it. Some hate it.

Field Execution: Safety, Logistics, and the Stuff That Goes Wrong

Radiation exclusion zones and production scheduling

A 70 Ci Ir-192 source produces a dose rate of roughly 5.2 mSv/hr at 1 meter. The controlled boundary for unbadged personnel is typically set at 2.5 μSv/hr (or whatever the local regulator specifies — in China it's GB 18871, in the US it's 10 CFR 20). For a typical silo site, that means a 30–50m exclusion zone during exposures. On a congested plant site, that's half the laydown area shut down for 15–30 minutes per exposure.

We schedule RT for early morning or night shifts when other crews aren't active. On one project in Saudi Arabia, we did all radiographic work between 0200 and 0500 — 40°C daytime temperatures made film handling miserable anyway, and the exclusion zone was easier to enforce when the site was empty. Shot 120 exposures in three nights. The client's safety officer slept through all of them.

Film handling, processing, and the humidity problem

Unprocessed film is sensitive to heat, humidity, and pressure. In tropical environments — Southeast Asia, West Africa, coastal India — we've had entire batches of film fogged by humidity before they were even exposed. Storage in sealed bags with desiccant is mandatory. Processing chemistry temperature must be held at 20±2°C; in a field darkroom with no air conditioning, that means processing at 0400 when ambient temperature finally drops.

Digital panels eliminate the chemistry problem but introduce their own: CR plates (imaging plates) are sensitive to scratches and must be erased between uses. A scratched plate produces artifacts that look exactly like cracks. I've seen a $3,000 rejection-and-rework cycle triggered by a scratched CR plate, not a real defect. We now carry three spare plates per project and replace any plate that shows surface damage.

Case Study: 8,000-Tonne Cement Silo, East Africa

We commissioned a 32m-diameter, 38m-high cement silo with 18mm shell plate. The spec called for 20% RT on vertical welds and 25% on circumferential welds, per EN ISO 5817 Level B. Total weld length was approximately 1,400 linear meters of vertical seam and 280 linear meters of circumferential seam. We planned 320 exposures using Ir-192 gamma sources with D4 film.

First week: 47 exposures shot, 4 rejected. Three were cluster porosity on circumferential welds — all from the same welder, same shift. Root cause was contaminated shielding gas (moisture in the CO₂ line from a faulty dryer). Fixed the dryer, re-shot the rejected joints, all passed. The fourth rejection was a toe crack on a vertical weld — traced to hydrogen-induced cracking from damp electrodes. The electrode oven had been unplugged overnight. We implemented a daily electrode bake log and the problem disappeared.

Total RT cost: approximately $6,200 (film, processing, technician time, source rental). Rework on the four rejected joints: $1,800. The alternative — no RT, and a crack propagating through a circumferential weld under cyclic loading from daily filling/emptying — would've been a leak, a cleanup, and a potential structural assessment costing $50,000–150,000. The math isn't close.

Frequently Asked Questions

How much RT coverage should I specify for a new silo?

For a standard cement or clinker silo under GB 50128, 10% minimum on vertical welds and 10% on circumferential welds is the code floor. I recommend 20% vertical and 25–50% circumferential for silos over 20m diameter or in seismic zones. For silos storing abrasive materials (coal, slag, alumina) where wall thinning is a long-term concern, I push for 100% RT on the bottom three shell courses — that's where erosion is fastest, and a weld defect accelerates localized wear.

Can radiographic testing detect corrosion under the silo floor?

No. RT requires access to both sides of the weld — source on one side, detector on the other. Under a silo floor, you've got concrete on one side and compacted fill on the other. For floor plate corrosion, use ultrasonic thickness scanning (grid pattern, 300mm spacing minimum) or acoustic emission testing if the silo is in service. RT is a weld inspection method, not a corrosion survey method. We've had clients ask us to RT the floor to find thin spots. That's not how it works.

What's the turnaround time for RT results on a typical silo project?

Film-based RT: 2–4 hours from exposure to interpreted result on the lightbox, assuming an on-site darkroom. Digital radiography: 15–30 minutes. For a silo with 200–400 exposures, a two-person RT crew can typically complete 25–40 exposures per shift. Interpretation and reporting adds another day. Plan for RT to add 5–8 working days to a typical silo erection schedule. Trying to compress it leads to rushed shooting, missed defects, and re-shoots that cost more time than the original inspection.