What Weld Undercut Actually Does to a Steel Silo (and Why 0.5mm Matters More Than You Think)

Weld undercut is a groove melted into the base metal alongside the weld toe that wasn't filled with weld metal. In a steel silo holding 5,000 tonnes of clinker, a 0.5mm undercut along a vertical seam can become a fatigue crack within 18 months of cyclic loading. This article covers what causes it, how much is acceptable, and what we've learned from fixing it on sites across Southeast Asia and the Middle East.

What Undercut Is and Where It Forms in Silo Welds

The geometry of undercut in shell-to-shell and shell-to-hopper joints

Undercut looks like a sharp notch running parallel to the weld line. You'll find it on the parent plate surface, right where the weld metal meets the base material. In silo construction, it shows up most often in three locations: vertical shell seams (the long welds running up the cylinder wall), horizontal circumferential seams (the ring welds connecting shell courses), and the shell-to-hopper transition joint where the cylinder meets the cone bottom.

The hopper transition is the one that keeps me up at night. That joint sees concentrated bending stress from the full weight of the stored material above, plus thermal cycling if you're storing hot clinker at 150–200°C. A 0.3mm undercut at that location acts like a pre-notch in a Charpy specimen — it gives the crack a starting point with zero incubation time.

I remember a project in Vietnam where we were commissioning a 4,200-tonne cement silo. The shell-to-hopper weld had been done by a subcontractor who'd never built a silo before. They'd run the vertical seams at 320 amps on a 4.0mm E7018 electrode — that's at least 40 amps too high for a 2G position weld. The undercut was visible to the naked eye, running 3–4mm deep in places. We cut out 14 metres of weld and re-did it. The plant manager wasn't happy about the two-week delay, but he'd have been a lot less happy with a crack propagating through his hopper six months after startup.

Why vertical welds on large-diameter silos are the worst offenders

Vertical welds on a silo shell are almost always done with the electrode travelling upward — vertical-up position. This is the hardest position to control because gravity is pulling the molten pool down while you're trying to push it up. The welder has to maintain a precise travel speed: too fast and the pool doesn't have time to wet out the sidewalls, leaving a groove. Too slow and you get excessive reinforcement but potential burn-through on thin shell plates.

On a 30-metre-diameter silo with 12mm shell plate, the vertical seams can run 8–10 metres in a single pass. That's a long time for a welder to maintain consistent technique, especially when it's 38°C on the platform and the electrode oven is 200 metres away on the ground. We've measured travel speed variations of ±30% on a single weld pass when the welder gets tired after hour three. That variation is the primary driver of intermittent undercut.

How Undercut Compromises Structural Integrity

Stress concentration factors and fatigue life reduction

A sharp undercut groove creates a stress concentration factor (Kt) that depends on its depth-to-radius ratio. For a typical undercut with a root radius of 0.1mm and depth of 0.5mm, Kt falls between 4 and 6. That means the local stress at the notch root is four to six times the nominal stress in the shell plate.

Now consider what a silo actually experiences. Every filling and emptying cycle applies a full load-unload cycle to the shell. A silo operating at 80% capacity utilisation cycles roughly 290 times per year. Over a 25-year design life, that's 7,250 cycles. At a stress concentration factor of 5, the effective fatigue stress range is five times higher than the design calculation assumes. The S-N curve doesn't lie — that can reduce fatigue life by a factor of 25 or more.

EN 1993-4-1 (the Eurocode for silo structures) requires fatigue assessment for silos subjected to more than 20,000 equivalent load cycles over their lifetime. Most industrial silos exceed that threshold within the first two years. If your weld quality doesn't match the fatigue category assumed in the design — and undercut effectively downgrades your joint by two or three fatigue categories — you're running an unaccounted safety deficit from day one.

Corrosion initiation at undercut grooves

The undercut groove traps moisture. In a cement plant environment with 70–90% relative humidity and ambient SO₂ from the kiln exhaust, that trapped moisture becomes an electrolyte. The groove geometry creates an oxygen concentration cell — oxygen is depleted at the bottom of the notch, making it anodic relative to the surrounding plate. Galvanic corrosion starts at the root and works outward.

We inspected a coal silo in Indonesia after seven years of service. The shell had lost 1.2mm of thickness from general corrosion — expected for that environment. But at undercut locations along the vertical seams, we measured localized pitting up to 3.5mm deep. The undercut had acted as a corrosion concentrator, tripling the local metal loss rate. That silo needed 22 individual patch repairs before it could return to service.

Acceptable Limits and Inspection Standards

EN ISO 5817 quality levels for silo welds

EN ISO 5817 defines three quality levels for fusion-welded joints: B (stringent), C (intermediate), and D (moderate). For structural silo welds subject to fatigue loading, the applicable level is typically C. Under the C level, the maximum permitted undercut is 0.5mm for material thicknesses between 3mm and 6mm, and 1.0mm for thicknesses above 6mm.

But here's where it gets tricky. Those limits apply to individual measurements. EN ISO 5817 also specifies that undercut shall not be continuous — it must be intermittent with a minimum spacing of 50mm between occurrences. A 0.4mm undercut that runs continuously for 2 metres is a rejection, even though each individual measurement is below the 0.5mm limit. The standard treats continuous undercut as a more severe defect because it creates a continuous stress raiser rather than isolated notches.

For silos designed to EN 1993-4-1, the executing class under EN 1090 determines the inspection rigor. Execution class EXC3 (required for structures with high consequence of failure) mandates 100% visual inspection plus 10% volumetric NDT (ultrasonic or radiographic) on all main shell welds. Undercut is assessed visually and measured with a welding gauge — a simple mechanical gauge with a sliding pointer that reads depth to 0.1mm resolution.

When 0.5mm becomes a rejection criterion

The 0.5mm limit isn't arbitrary. It corresponds to roughly 4–8% of the shell plate thickness on a typical silo (6–12mm plate). At that depth, the stress concentration factor becomes significant enough to affect fatigue performance. Below 0.5mm, the notch is shallow enough that local yielding at the root blunts the stress concentration during the first few load cycles — the material effectively "heals" itself through plastic deformation.

Above 0.5mm, the notch is too deep for blunting to fully compensate. The stress concentration persists through the full fatigue life. That's why the standard draws the line there. For critical welds — shell-to-hopper joints, ring beam connections, and any weld in a zone of high membrane stress — we apply a tighter internal limit of 0.3mm, regardless of what the code permits.

Root Causes in Field Welding Conditions

Travel speed and amperage mismatches on vertical-up welds

The single most common cause of undercut on silo vertical seams is excessive travel speed. The welder moves the electrode too fast along the joint, and the arc doesn't have enough time to melt the sidewalls of the groove. The weld metal bridges across the top but leaves a channel along each edge.

On a 10mm plate with a single-V preparation, the correct travel speed for a 4.0mm E7018 electrode at 160–180 amps is roughly 120–150mm per minute. We've timed welders on site who were running at 200mm/min or more — they were trying to finish the seam before their shift ended. At that speed, undercut is guaranteed.

The fix is procedural, not technical. We specify maximum travel speeds in the WPS (Welding Procedure Specification) and enforce them through weld pass length tracking. If a welder completes a 10-metre vertical seam in under 70 minutes, that's a red flag. We also require a welding supervisor to spot-check every third vertical seam with a gauge before the welder moves to the next joint.

Wind, humidity, and electrode moisture on outdoor sites

Most silo construction happens outdoors. Wind speeds above 8m/s disrupt the gas shield on FCAW (flux-cored arc welding) processes, causing arc instability that produces intermittent undercut. We've had to stop welding on a site in Oman when the shamal winds hit 15m/s — the welds were coming out with undercut so severe you could catch a fingernail in it.

Electrode moisture is the other silent killer. E7018 electrodes are low-hydrogen and must be kept in a 250°C holding oven until use. On a site in Myanmar, we found electrodes sitting in an open bucket on the platform for six hours. The moisture content had risen to 0.8% — the maximum permitted is 0.2%. The resulting welds had undercut combined with hydrogen-induced cracking. We scrapped 340 metres of completed weld and sent the crew back to the classroom for a two-day re-qualification.

Prevention and Repair Protocols

Welding procedure qualification for silo-specific conditions

A WPS qualified for a workshop environment doesn't automatically transfer to a field silo site. The Welding Procedure Qualification Record (WPQR) must account for the actual conditions: outdoor wind, temperature extremes, plate fit-up tolerances, and the specific positions encountered on a silo (vertical-up for shell seams, flat or horizontal for circumferential joints).

We qualify our silo WPSs with a minimum of two test plates welded in the vertical-up position, with intentional wind exposure up to 5m/s. The test welds are visually inspected, then sectioned for macro-etch examination. We measure undercut depth on three cross-sections per weld. If any section shows undercut exceeding 0.3mm, the WPS fails qualification and we adjust parameters — usually reducing travel speed by 10–15% or increasing electrode weave width by 2–3mm.

The cost of a proper WPQR qualification is roughly $2,500–$4,000 per procedure, including materials, welding time, and metallographic examination. That's cheap insurance against a rework bill that can run $15,000–$30,000 for a single silo if undercut is found during final inspection.

Grinding, re-weld acceptance criteria

When undercut is found, the repair protocol depends on depth. For undercut up to 0.5mm that's intermittent and within EN ISO 5817 limits, we smooth-grind the groove to blend it into the surrounding plate. The grind must achieve a minimum radius of 2mm at the root — we check this with a weld profile gauge. No re-welding is needed; the grinding removes the stress raiser.

For undercut exceeding 0.5mm or continuous undercut of any depth, we grind out the defect completely, then re-weld using a qualified repair WPS. The repair weld is inspected to the same criteria as the original. We don't accept "cosmetic" repairs — a thin layer of weld metal over an undercut groove doesn't fix the stress concentration. The groove has to be fully excavated first.

On a 6,000-tonne clinker silo in Saudi Arabia, we found 47 linear metres of undercut along the lower shell course during final inspection. The root cause was a batch of FCAW wire with inconsistent flux fill — the wire manufacturer had a quality control lapse. We ground out every metre, re-welded with a different wire batch, and re-inspected. The repair cost was $22,000 and added 11 days to the schedule. The alternative — leaving it — would have meant a potential fatigue crack within three years, with repair costs exceeding $150,000 including lost production.

Case Study: Undercut Discovery on a 3,500-Tonne Fly Ash Silo

A fly ash storage silo in Gujarat, India — 18m diameter, 22m shell height, 3,500-tonne capacity — was commissioned in March 2022. The shell was 10mm and 12mm ASTM A572 Grade 50 plate, welded with FCAW using E71T-1C wire. The fabricator had qualified their WPS in a workshop with no wind exposure.

During pre-commissioning inspection in August 2022, we found undercut on 60% of the vertical seams in the lower three shell courses. Depths ranged from 0.3mm to 1.1mm, with continuous runs up to 4 metres. The undercut was concentrated on the advancing side of the weld — the side where the gun angle was too steep relative to the plate surface.

Root cause analysis identified three contributing factors: (1) the WPS travel speed of 180mm/min was too fast for the 12mm plate in vertical-up position — it should have been 130mm/min; (2) the site had no wind protection, and afternoon winds of 6–10m/s were disrupting the CO₂ shield; (3) the welders had not been trained on the specific gun angle required for vertical-up FCAW on thick plate.

We repaired 89 metres of weld by grinding and re-welding. The repair protocol required grinding to sound metal (minimum 1mm below the undercut root), then re-welding with a corrected WPS at 130mm/min travel speed and 20° gun angle instead of the original 35°. All repair welds passed visual inspection and 10% ultrasonic testing. The silo entered service in November 2022 with a corrected WPS and mandatory wind screens for all future welding on site.

Frequently Asked Questions

Can undercut be repaired without re-welding?

Yes, if the undercut is shallow (≤0.5mm) and intermittent. Grinding the groove to a smooth profile with a minimum 2mm root radius removes the stress concentration. This is acceptable under EN ISO 5817 quality level C. For deeper or continuous undercut, grinding alone isn't sufficient — you need to excavate the defect and re-weld.

Does undercut affect the silo's pressure rating?

Indirectly, yes. A silo shell under internal pressure from pneumatic conveying (typically 15–25 kPa for cement injection) experiences hoop tension. Undercut reduces the effective cross-sectional area at the weld and creates a stress concentration. For thin-shell silos (6–8mm plate), a 0.5mm undercut represents 6–8% of the section — that's a meaningful reduction in pressure-bearing capacity. For thick-shell silos (16mm+), the effect is less significant but still relevant for fatigue assessment.

How do you distinguish undercut from a lack-of-fusion defect?

Undercut is a surface-breaking groove at the weld toe, visible from the outside. Lack-of-fusion is an internal defect between weld passes or between weld metal and base plate — it can't be seen visually. Undercut is detected by visual inspection and measured with a gauge. Lack-of-fusion requires ultrasonic or radiographic testing. The two defects often co-exist because the same welding parameters (excessive travel speed, incorrect angle) that cause undercut also promote lack-of-fusion.