Welding Inspection and Repair Standards for Steel Silos: What Every Plant Owner Needs to Know

Steel silo welding isn't like structural steelwork — the loads are different, the consequences of failure are worse, and the inspection standards are stricter than most contractors realize. This article covers the actual welding standards that apply to industrial steel silos (AWS D1.1, EN 1090, GB 50661), the inspection methods that catch defects before they become failures, and the repair procedures that keep your silo in service without tearing the whole shell apart.

Why Welding Quality in Steel Silos Is Non-Negotiable

The Load Profile Nobody Talks About

Most structural steelwork deals with static, predictable loads. A silo doesn't cooperate. You've got the dead weight of the stored material — 3,000 tonnes of cement powder pushing outward with a lateral pressure that peaks at roughly 0.38 times the vertical pressure, depending on the Janssen model parameters. Then there's the thermal cycling. I've measured a 40°C temperature swing on the south face of a silo in Algeria between 6am and 2pm. That shell is expanding and contracting every single day, and every weld joint is taking that fatigue cycle.

On top of that, you've got the filling and discharge dynamics. When you pneumatically convey material into a silo at 150 tonnes per hour, you're not just adding weight — you're creating pressure surges, particle impingement on the shell wall, and differential settlement that twists the foundation ring. The welds at the shell-to-cone junction see the worst of it. I've seen circumferential butt welds crack at that junction after 18 months because the original weld had 3mm of lack of fusion that nobody caught during construction.

What Makes Silo Welding Different

Here's the thing most general contractors don't understand: a silo shell is a pressure-retaining structure, even if it's not classified as a pressure vessel under ASME. The stored material exerts outward pressure on the wall, and the welds have to resist that hoop tension continuously. A 25mm-thick shell plate storing clinker at 1,200 kg/m³ density generates roughly 180 MPa of hoop stress at the bottom ring. Your weld has to match that — not just in static strength, but in fatigue resistance over 20+ years of thermal cycling.

That's why we don't treat silo welding like a warehouse frame. The acceptance criteria are tighter, the inspection coverage is higher, and the repair protocols are more conservative. I've rejected entire circumferential weld runs on new construction because the contractor was using a WPS qualified for structural steel, not for the specific plate thickness and joint configuration we had on site. Cost us three days. Saved us a catastrophic failure five years later.

Applicable Welding Standards for Steel Silos

AWS D1.1 and EN 1090: The Two Frameworks You'll Encounter

If you're working in North America, the Middle East, or most of Southeast Asia, AWS D1.1 (Structural Welding Code — Steel) is the default reference. It covers the qualification of welders, the qualification of welding procedures (WPS/PQR), and the acceptance criteria for visual and non-destructive testing. For silo work, the relevant sections are typically 4 (qualification), 5 (fabrication), and 6 (inspection). The key number to remember: under AWS D1.1, the maximum allowable undercut for dynamically loaded members is 0.5mm. For silo shells, I apply that same limit even though the code doesn't explicitly classify silos as "dynamically loaded" — because they are.

In Europe and increasingly in Africa and the Middle East, EN 1090 (Execution of Steel Structures) is the governing standard. It uses execution classes from EXC1 to EXC4, with most industrial silos falling into EXC3. That class requires 100% visual inspection of all welds, plus a minimum of 10% non-destructive testing on butt welds and 5% on fillet welds. The NDT has to be performed by a certified inspector — Level 2 minimum under EN ISO 9712. I've seen projects in Nigeria where the contractor tried to use a Level 1 technician for ultrasonic testing on shell butt welds. That's not acceptable under EXC3, and we shut down the weld crew until they brought in qualified personnel.

GB 50661 and Chinese National Standards

For projects in China or using Chinese design codes, GB 50661 (Code for Welding of Steel Structures) applies. It's broadly aligned with AWS D1.1 in terms of acceptance criteria, but it has specific requirements for preheat temperatures that are more conservative in cold-weather conditions. For Q345B steel (the most common grade for silo shells in China), GB 50661 requires a minimum preheat of 50°C when the ambient temperature drops below 5°C and the plate thickness exceeds 25mm. I've been on sites in Inner Mongolia where the ambient was -15°C and we had to maintain 80°C preheat on 30mm shell plates. The welders hated it. The welds were perfect.

There's also GB 50205 (Code for Acceptance of Construction Quality of Steel Structures), which governs the inspection and acceptance process. It specifies the sampling rates for NDT, the documentation requirements, and the defect classification system. Class I welds (full penetration butt welds in tension) require 100% ultrasonic testing. Class II welds require a minimum of 20% UT. For silo shell circumferential and longitudinal butt welds, we treat them as Class I — no exceptions.

Inspection Methods: What We Actually Look For

Visual Inspection (VT) — The First and Most Important Step

Every single weld gets a visual inspection before any NDT. That sounds obvious, but I've been on sites where the NDT crew showed up before the VT was completed, and they were testing welds that had visible undercut, overlap, or arc strikes. You're wasting money if you're doing UT on a weld that already fails visual.

What we're looking for: undercut depth (measured with a welding gauge, not guessed), weld profile (convexity and concavity relative to the plate surface), surface porosity (any pore larger than 1.6mm is a reject), arc strikes (anywhere on the parent metal, not just the weld), and crater cracks at stop-start positions. On a typical silo shell, I'll find 3-5 stop-start positions per circumferential weld run. Each one is a potential crack initiation site. The welder needs to grind those out and restart with a proper run-off tab.

The inspection has to happen at the right time. For single-pass fillet welds, VT is done immediately after welding. For multi-pass butt welds, we inspect after each pass — especially the root pass. If the root has lack of fusion or excessive penetration, the subsequent passes won't fix it. I caught a root pass defect on a 20mm butt weld in Vietnam that would have been buried under six more passes. The welder had the amperage set 20A too low for the root opening. We stopped, re-qualified his parameters, and re-ran the joint. Took 45 minutes. The alternative was a crack propagating through the shell wall under cyclic loading.

Non-Destructive Testing: UT, RT, MT, and PT

Ultrasonic testing (UT) is the workhorse for silo shell inspection. It's fast, it doesn't require radiation safety zones, and it's excellent at detecting planar defects like lack of fusion and cracks. For butt welds in shell plates 12mm and thicker, we use a 45° and 60° probe combination, calibrated on a reference block with a 3mm side-drilled hole. The acceptance criteria under AWS D1.1 for tension welds: no indication longer than 6mm for material under 25mm thick, no indication longer than 8mm for material 25-50mm thick. But for silos, I tighten that to 4mm maximum regardless of thickness. The fatigue loading doesn't forgive.

Radiographic testing (RT) is used selectively — typically on the shell-to-bottom cone junction, on nozzle reinforcement pads, and on any weld where UT gives an ambiguous indication. RT gives you a permanent record (the film or digital image), which is valuable for documentation. But it's slow, it requires exclusion zones, and it's less effective at detecting planar defects oriented perpendicular to the beam. I'd say we use RT on maybe 15% of the total weld length on a typical silo project. The rest is UT.

Magnetic particle testing (MT) and liquid penetrant testing (PT) are surface methods. We use MT on fillet welds — especially at the hopper-to-shell junction, where the weld is in tension and surface-breaking cracks are the primary concern. PT is used on non-ferritic materials or where MT isn't practical (like inside a narrow manway). Both methods are sensitive to surface preparation — the weld has to be clean, dry, and free of paint or heavy scale. I've seen MT give false indications because the contractor hadn't removed the spatter from the weld toes. Clean the surface first. Always.

Common Weld Defects in Silo Construction

Porosity and Gas Inclusions

Porosity is the most common defect we find, and it's almost always a process issue — not a material issue. The causes: moisture in the shielding gas (for GMAW/FCAW), damp electrodes (for SMAW), or wind blowing the gas shield away on outdoor sites. I was on a project in Malaysia during monsoon season where the relative humidity was 95% and the contractors were using E7018 electrodes straight from the oven — but the oven was 200 meters from the weld position. By the time the electrodes reached the welder, they'd absorbed enough moisture to produce visible porosity in the first pass. We moved the holding oven to within 10 meters of the silo and the problem disappeared.

The acceptance criteria: under AWS D1.1, individual pores up to 3mm are acceptable in most applications, but the total pore area in any 100mm length of weld can't exceed 2% of the weld cross-section. For silos, I apply a tighter limit — no pore larger than 2mm, and no cluster of pores within any 50mm segment. Clustered porosity creates a stress concentration that's worse than a single large pore.

Lack of Fusion and Incomplete Penetration

These are the defects that keep me up at night. Lack of fusion (LOF) is a planar defect — it's essentially a crack that was there from day one. It occurs when the weld metal doesn't fuse to the parent metal or between passes. The root cause is almost always incorrect welding parameters: amperage too low, travel speed too fast, or electrode angle wrong. On a silo shell, LOF at the root of a butt weld is the most dangerous defect because it's in the highest stress zone and it's oriented perpendicular to the hoop stress.

We found a 40mm-long LOF indication on a circumferential butt weld during UT inspection of a 2,500-tonne fly ash silo in India. The weld had passed visual inspection — the surface looked fine. But the UT showed a clear planar indication at the root, 3mm below the surface. We cut out the entire 1.2-meter section and re-welded it. The root cause was a welder who'd been running the root pass at 140A when the WPS specified 160-180A. He was trying to avoid burn-through on the 16mm plate. Instead, he created a defect that would have grown into a through-wall crack within two years of thermal cycling.

Repair Procedures: When a Weld Fails Inspection

Grinding Out vs. Full Removal

Not every defect requires cutting out the entire weld. Surface-breaking defects — undercut, porosity at the surface, minor overlap — can be repaired by grinding. The rule is: grind out the defect, blend the repair area smoothly into the surrounding weld profile (maximum slope of 1:3), and re-inspect with MT or PT. If the defect is deeper than 2mm below the surface, you need to weld back up using a qualified repair WPS with reduced amperage (typically 10-15% below the original parameters) to minimize heat input and distortion.

For subsurface defects detected by UT — LOF, internal porosity, cracks — you have to remove the defect completely. The standard method is to gouge out the defect from one side (or both sides if the plate thickness allows), verify removal with MT or PT, and then re-weld. The repair weld length has to extend at least 50mm beyond each end of the original defect. I've seen contractors try to just "cap over" a UT-indicated defect with an additional weld pass. That doesn't work — the defect is still there, and now it's buried under more weld metal that's masking the UT signal.

Re-Inspection and Documentation

Every repair gets re-inspected using the same method that found the original defect, plus an additional method if the repair is in a critical location. A repaired butt weld in the bottom shell ring gets UT plus MT. A repaired fillet weld at the hopper junction gets VT plus MT. The repair has to meet the same acceptance criteria as the original weld — there's no "relaxed" standard for repairs.

Documentation is non-negotiable. Every repair gets a unique number, a location sketch, the original defect description, the repair method, the re-inspection result, and the inspector's signature. I've been called back to projects three years later because a repaired weld failed in service, and the only way to figure out what went wrong was to pull the repair records. If the records don't exist, you're guessing. And guessing with a 3,000-tonne silo above your head is a bad career move.

Field Welding Challenges Nobody Warns You About

Weather, Access, and the Human Factor

Shop welding is controlled. Field welding is a fight against everything the environment throws at you. Wind is the biggest enemy of GMAW and FCAW — even a 5 km/h breeze can disrupt the shielding gas envelope and cause porosity. On exposed sites, we use wind shields around the weld position, but that adds time and the welders don't always want to set them up. I've had arguments with crew supervisors who said "it's just a little wind." It's not. It's a defect waiting to happen.

Access is another issue. The inside of a silo under construction is a confined space — limited ventilation, limited lighting, and the welder is often working in an awkward position on a scaffold. Fatigue sets in fast. I've seen weld quality drop noticeably after the third hour of continuous overhead welding inside a silo shell. The solution is rotation — switch welders every 2 hours, even if it means more setup time. The alternative is a higher defect rate and more repairs, which costs more in the long run.

Then there's the human factor. I was on a project in Algeria where we had a sandstorm roll through at 3pm. The wind hit 60 km/h, sand was getting into everything, and the welders kept working because the schedule was tight. By the time I got to the site, they'd completed 8 meters of circumferential weld. We cut out every meter. The porosity was so dense you could see it with the naked eye. The schedule delay cost us four days. Working through the sandstorm would have cost us the same four days plus the cost of the original weld metal, the grinding, the re-welding, and the re-inspection. The math isn't complicated.

Material Condition and Fit-Up

The best WPS in the world won't save a weld if the joint fit-up is wrong. For butt welds, the root opening has to be within ±2mm of the WPS specification. The root face has to be within ±1mm. The alignment offset (mismatch) between the two plates can't exceed 2mm for plates up to 20mm thick, or 3mm for thicker plates. I've seen fit-up offsets of 6-8mm on field joints where the shell plates weren't properly aligned before tack welding. The welder tries to compensate by adding more weld metal, which increases residual stress and distortion. The result is a weld that looks massive but has internal defects from the excessive heat input.

Surface condition matters too. Mill scale, rust, paint, and moisture all contribute to hydrogen absorption and porosity. The joint area has to be cleaned to bright metal for at least 25mm on each side of the weld. For high-strength steels (yield above 355 MPa), we also require preheat to drive off moisture — typically 50°C minimum, verified with a contact thermometer, not an infrared gun. Infrared guns give you surface temperature, but the heat soak through the plate thickness is what matters. I've measured a 30°C difference between the surface reading and the mid-thickness temperature on a 40mm plate. That's enough to make the difference between a good weld and a hydrogen crack.

Case Study: Weld Repair on a 5,000-Tonne Clinker Silo

We were commissioned to inspect a 5,000-tonne clinker silo in Central Asia that had been in service for seven years. The plant reported intermittent dust leakage at the shell-to-cone junction — not a structural failure yet, but a warning sign. Our inspection found three circumferential cracks in the fillet weld connecting the shell to the bottom cone, ranging from 45mm to 120mm in length. UT confirmed the cracks were through-wall in two locations.

The root cause was a combination of factors: the original weld had been made with a 6mm fillet when the design called for 8mm, the weld toe had severe undercut (measured at 1.2mm), and the cone-to-shell junction had been subjected to differential settlement that added a bending stress the original design hadn't accounted for. The repair procedure: we removed the entire fillet weld around the full circumference (approximately 28 meters), ground the shell and cone surfaces back to bright metal, re-welded with a qualified WPS using E7018 electrodes at 160-170A, and performed 100% VT plus 100% MT on the completed repair. The repair took 12 working days with a crew of four welders and two inspectors. The silo has been in service for three years since with no further issues.

FAQ

Q: What percentage of welds on a steel silo require non-destructive testing?

A: For a typical industrial silo built to EXC3 under EN 1090, you're looking at 100% visual inspection on all welds, 10% minimum UT on butt welds, and 5% minimum on fillet welds. In practice, we run closer to 20-25% UT on shell butt welds and 10% on fillet welds because the cost of finding a defect early is a fraction of the cost of a failure in service. For critical junctions — shell-to-cone, shell-to-roof, nozzle penetrations — we do 100% UT plus surface NDT.

Q: Can a cracked silo weld be repaired without emptying the silo?

A: It depends on the location and the extent of the crack. Surface cracks on the exterior shell can often be repaired with the silo in service, provided the internal pressure is low and the repair is done with controlled heat input (below 2 kJ/mm for most carbon steels). Through-wall cracks or cracks at structural junctions require the silo to be emptied and cleaned before repair. I've done external repairs on silos holding up to 500 tonnes of material, but anything beyond that and the shell deflection under load makes weld quality unreliable. Empty the silo. It's the right call.

Q: How often should an operating silo be inspected for weld integrity?

A: A full visual inspection of all accessible welds annually, plus a targeted UT inspection of high-stress junctions every three years. For silos storing abrasive materials (clinker, slag, coal), the shell-to-cone junction and the lower 3 meters of the shell wall should be inspected annually with UT because abrasion thinning and weld fatigue concentrate in those zones. After any seismic event or foundation settlement incident, a full weld inspection is mandatory before returning the silo to service.