Steel Silo Weld Porosity: How to Prevent It and Why It Matters
Weld porosity is gas pockets trapped in the weld metal during solidification — tiny voids that weaken the joint and create leak paths. In a steel silo holding 5,000 tonnes of cement, even 2% porosity in a shell weld can cut fatigue life by 40% and let moisture creep into the stored material. The fix isn't complicated, but it demands discipline at every step from plate cutting to final inspection.
What Weld Porosity Actually Is (and Isn't)
The physics of gas entrapment
When you're welding, the molten pool is a hungry thing. It pulls dissolved gases from the atmosphere, from moisture on the plate surface, from oil, from rust, from the coating on your filler wire. Hydrogen, nitrogen, carbon dioxide — they all dissolve into the liquid metal at welding temperature. Then the pool cools and solidifies, and the solubility of those gases drops by roughly 90%. The gas has to go somewhere. If it can't escape to the surface before the metal freezes, it gets trapped as a pore.
The size range matters. Micro-porosity — pores under 0.5 mm — is almost unavoidable in manual welding and usually harmless. Macro-porosity, anything above 1 mm, is a different animal. That's where you get interconnected voids, wormholes, and the kind of defect that shows up on an X-ray and makes the QA engineer's morning miserable.
Porosity vs. other weld defects
Don't confuse porosity with lack of fusion or slag inclusions. They look similar on a radiograph but have completely different causes and remedies. Porosity is round or slightly elongated, uniformly distributed or clustered. Slag is irregular, often linear. Lack of fusion is a sharp, planar discontinuity. I've seen a contractor in Indonesia spend three days grinding out what they thought was porosity, only to find it was actually tungsten contamination from a bad TIG start. Wrong diagnosis, wasted time, same problem still there.
The key distinction: porosity is a gas problem. Everything else is a process or technique problem. Fix the gas, fix the porosity.
Root Causes in Silo Fabrication
Moisture and hydrogen — the number one enemy
Here's the thing. Hydrogen is everywhere. It's in the moisture on your plate at 70% relative humidity. It's in the cellulose coating on E6010 electrodes. It's in the lubricant on your wire feed liner. And it dissolves into molten steel like sugar into hot tea — about 2.5 mL per 100 g of liquid metal at 1,600°C. When that metal freezes, the solubility crashes to roughly 0.2 mL per 100 g. The excess hydrogen has to escape. If the weld cools too fast — say, on a 12 mm plate in 5°C ambient temperature — it doesn't get the chance.
I remember a project in Vietnam where the night shift was welding silo shell plates in the open. Morning dew had settled on the joint surfaces. The radiographic rejection rate hit 35% on the first lift. We stopped, set up portable heaters to bring the plate temperature to 25°C minimum, and ran a 150°C preheat on all joints for 50 mm either side of the weld. Rejection rate dropped to 4% overnight. The fix cost us two days and about $3,800 in gas and labor. The alternative — re-welding 35% of the seams — would have cost $45,000 and three weeks.
Shielding gas failures
Your shielding gas is the only thing standing between the molten pool and the atmosphere. When it fails, porosity follows within seconds. The most common failure modes in silo work:
Flow rate too low. Below 15 L/min for GMAW, the gas envelope collapses and air gets aspirated into the arc. I've seen welders set their regulators to 8 L/min to "save gas" and then wonder why their welds look like Swiss cheese.
Flow rate too high. Above 25 L/min, you get turbulence. The laminar flow breaks down and actually pulls air into the shielding zone. It's counterintuitive, but 30 L/min can give you worse coverage than 12 L/min.
Wind. Outdoor silo construction means wind. At 3 m/s — a light breeze you barely feel — the shielding gas envelope is already compromised. At 5 m/s, it's gone. We use windbreaks on every outdoor job. Canvas screens, plywood barriers, whatever's available. A 2 m high windbreak on the windward side reduces gas velocity at the weld zone by 60-70%.
Leaks in the gas delivery system. Cracked hoses, loose fittings, a regulator diaphragm that's been bumped one too many times. A leak that loses 2 L/min out of a 20 L/min flow is a 10% reduction in shielding effectiveness. That's enough to push you from acceptable to rejectable porosity.
Surface contamination
The plate surface is where most porosity problems start, and it's the easiest thing to control. Mill scale, rust, paint, grease, cutting fluid residue — all of these decompose in the arc and release hydrogen and oxides into the weld pool.
For silo shell plates, the joint preparation area needs to be cleaned to bare metal for at least 25 mm back from the weld groove. Not wire-brushed. Ground. A flap disc or grinding wheel to bright metal. I don't care if the plate looks clean to the naked eye — if it's been stored outside for more than 48 hours in a humid climate, there's moisture in the oxide layer that you can't see.
On a job in Shandong, we had a batch of Q345B plates that had been stored at the fabricator's yard for three weeks through a rainy period. The mill scale looked intact, but when we started welding, porosity was everywhere. We ended up shot-blasting the entire batch before cutting. Added $12 per tonne to the fabrication cost, but the alternative was rework that would have cost five times that.
Prevention Strategies That Work
Pre-weld preparation — the boring stuff that saves projects
Nobody gets excited about pre-weld preparation. It's not glamorous. But it's where 80% of porosity prevention happens. Here's the checklist we run on every Manxing silo project:
1. Plate storage. Covered, off the ground, with ventilation. If the ambient humidity is above 80%, we run dehumidifiers in the storage area. Target: keep plate surface temperature at least 3°C above the dew point.
2. Joint cleaning. Grind to bright metal 25 mm either side of the joint. Remove all mill scale, rust, paint, and cutting dross. Wipe with acetone if there's any chance of hydrocarbon contamination.
3. Electrode handling. Low-hydrogen electrodes (E7018, E7016) need to be kept in holding ovens at 120-150°C. Once a hermetically sealed can is opened, the electrodes go into the oven within 4 hours. Any electrode that's been exposed to ambient air for more than 2 hours gets re-baked at 300°C for 1 hour before use. This is non-negotiable.
4. Preheat. For plate thickness above 25 mm, or ambient temperature below 10°C, we preheat to 80-120°C measured 75 mm from the joint. This slows the cooling rate and gives dissolved gases time to escape before the metal freezes.
Welding procedure optimization
The welding procedure specification (WPS) is your contract with the weld. If it's wrong, the weld will be wrong. For silo shell welds, the critical parameters are:
Wire feed speed and voltage. Too high a voltage lengthens the arc, widens the shielding gas coverage gap, and increases nitrogen pickup. For GMAW with 1.2 mm wire, we run 22-26 volts and 200-280 A. Push the voltage above 28 V and you'll see porosity climb.
Travel speed. Too fast, and the solidifying metal traps gas before it can escape. Too slow, and you get excessive heat input that can cause other problems. For vertical-up welds on silo shells — which is most of them — we target 150-250 mm/min. The welder needs to watch the pool, not the clock, but that's the range.
Contact tip to work distance. For GMAW, 12-15 mm. Any further and the shielding gas coverage drops off. Any closer and you get spatter buildup on the nozzle that disrupts gas flow. We check this every 30 minutes of arc time.
Back-groove and back-weld. Full-penetration single-V butt joints on silo shells need back-gouging or back-grinding before the second side is welded. If you don't, you get lack of fusion at the root, and the trapped slag and gas from the first pass create porosity in the second pass. We use air arc gouging with a 10 mm diameter carbon electrode, followed by grinding to bright metal.
Environmental controls
You can't control the weather, but you can control the welding environment. For outdoor silo construction, we set up three lines of defense:
Windbreaks. As mentioned earlier, canvas or plywood barriers on the windward side. Minimum 2 m high, positioned 1.5 m from the weld joint. This cuts wind speed at the arc by more than half.
Preheat and interpass temperature control. We monitor interpass temperature with contact thermometers, not infrared guns — contact is more accurate on rough steel surfaces. Maximum interpass temperature for Q345B is 250°C. Go above that and you risk metallurgical issues in the heat-affected zone.
Rain and condensation. No welding when it's raining, period. No welding on wet surfaces, period. If the plate has been rained on, it gets dried with torch heat and re-cleaned before welding resumes. We've had situations where a sudden tropical downpour caught us mid-weld on a 10,000-tonne clinker silo in Malaysia. We stopped, covered the work, waited two hours, then ground out the last 100 mm of every open joint before restarting. Lost half a day. Saved a month of rework.
Detection and Repair
NDT methods for porosity
You can't see porosity with your eyes — not the kind that matters, anyway. Surface-breaking pores are visible, but they're the tip of the iceberg. The dangerous porosity is internal, and you need non-destructive testing to find it.
Radiographic testing (RT). This is the gold standard for porosity detection. X-ray or gamma ray source on one side, film or digital detector on the other. Porosity shows up as dark spots or clusters on the radiograph. Sensitivity is typically 2% of wall thickness — meaning a pore needs to be at least 2% of the plate thickness to be reliably detected. For a 10 mm shell plate, that's 0.2 mm. Most porosity that matters is larger than that.
Ultrasonic testing (UT).strong> UT can detect porosity, but it's less reliable than RT for this specific defect. Porosity tends to scatter the ultrasound beam, which can mask other defects. We use UT for thickness verification and crack detection, RT for porosity.
Acceptance criteria. Per EN ISO 5817, for quality level B (the normal requirement for silo structures), the maximum pore diameter is 2 mm for wall thickness up to 10 mm, and 3 mm for wall thickness 10-20 mm. Clustered porosity — multiple pores within a 25 mm diameter circle — is limited to a total area of 16 mm². Exceed those limits and the weld gets rejected.
When to repair vs. accept
Not every pore is a rejection. A single 1 mm pore in a 50 m long shell weld is not going to cause a structural failure. The silo will hold its load, the cement will stay dry, and nobody will ever know it's there. The question is always: does the defect exceed the acceptance criteria in the applicable standard?
For silo shell welds, we follow EN 1090-2 execution class EXC2 for most applications. That means 10% radiographic sampling of all butt welds, with acceptance per EN ISO 5817 level C. If the defect is below the acceptance threshold, we document it and move on. If it's above, we repair.
The repair is straightforward: grind out the defective area, re-weld using the same WPS, and re-inspect. But here's the catch — you need to grind out the entire defect. If you just cap over a porous weld, the porosity is still there, hidden under a new layer of sound metal. The radiograph will look fine, but the structural integrity hasn't improved. We've seen contractors try this. It never ends well.
Standards and Quality Control
Applicable codes
Silo welding isn't a free-for-all. The standards that govern our work include:
EN 1090-2. European standard for execution of steel structures. Defines weld quality levels, inspection requirements, and acceptance criteria. For silos, EXC2 is the minimum; EXC3 for seismic zones or high-consequence facilities.
EN ISO 5817. Quality levels for fusion welding. Defines pore size limits, cluster limits, and other defect acceptance criteria. Level B for static structures, Level C for fatigue-loaded structures like silo shells.
AWS D1.1. American Welding Society structural welding code. Used on projects in the Americas and the Middle East. Similar acceptance criteria to EN ISO 5817 but with different nomenclature.
GB 50205. Chinese national standard for acceptance of steel structure engineering quality. Used on domestic projects in China. Weld quality grades I, II, and III, with Grade II being the normal requirement for silo shells.
Documentation requirements
Every weld on a Manxing silo gets documented. The welder's ID, the WPS number, the date, the ambient temperature, the preheat temperature, the interpass temperature, the electrode batch number, the shielding gas flow rate. If it's not documented, it didn't happen.
For radiographic testing, we keep the radiographs for 10 years minimum. The report includes the weld location, the defect type, the defect size, the acceptance criteria applied, and the pass/fail determination. If a defect is found, the repair report references the original radiograph and includes the post-repair radiograph.
This isn't bureaucracy. It's traceability. If a silo shell fails in service — and I've seen it happen — the first thing the investigator asks for is the weld documentation. If you can't produce it, you're liable. If you can, you can prove the weld was acceptable when it left your hands.
Case Study: 8,000-Tonne Cement Silo — Porosity Crisis and Recovery
A cement plant in central China commissioned an 8,000-tonne steel cement silo, 22 m diameter, 28 m shell height, Q345B steel, 8-16 mm shell thickness. The fabricator was a local shop with decent equipment but limited experience on large silos. They started shell welding in November, ambient temperature 2-8°C, humidity 75-85%.
After the first three shell courses were welded, we ran radiographic testing per EN 1090-2 EXC2 — 10% sampling of all vertical and horizontal butt welds. The rejection rate was 28%. Porosity was the primary defect, mostly clustered, with individual pores up to 3.5 mm diameter. The acceptance limit per EN ISO 5817 Level C was 2 mm for the 10 mm plate and 3 mm for the 16 mm plate. Most of the defects were in the 10 mm vertical seams.
Root cause analysis pointed to three issues: no preheat, electrodes stored at ambient temperature without oven, and wind exposure during welding with no windbreaks. The fabricator had been welding structural steel for buildings, where the requirements are less strict and the consequences of porosity are lower. A silo shell is a different beast — it's a pressure vessel, a weather barrier, and a structural element all at once.
We implemented the full prevention protocol: preheat to 100°C minimum, electrode holding ovens at 150°C, canvas windbreaks on all four sides of the silo, and joint grinding to bright metal before every weld. We also added a hold point — no welding starts until the QC inspector verifies preheat temperature and joint cleanliness.
The remaining five shell courses were welded under the new protocol. Radiographic rejection rate dropped to 3%. The rework on the first three courses cost $28,000 and added 12 days to the schedule. The prevention measures for the remaining courses cost $6,500. Total cost of the porosity problem: $34,500. Total cost of prevention: $6,500. The ratio speaks for itself.
FAQ
Q: Can weld porosity be completely eliminated?
No. Micro-porosity — pores under 0.5 mm — is a natural consequence of the welding process and is generally harmless. The goal is to eliminate macro-porosity above 1 mm and clustered porosity that exceeds acceptance criteria. With proper procedures, you can get rejection rates below 2%. Zero is unrealistic.
Q: Does porosity affect silo structural integrity?
It depends on the size, distribution, and location. A few isolated 1 mm pores in a shell weld won't cause a failure. But clustered porosity covering more than 5% of the weld cross-section can reduce fatigue strength by 30-50%. In a silo that cycles daily — filling and emptying — fatigue is the governing design case. That's why we take porosity seriously even when the static load capacity isn't affected.
Q: What's the cost of radiographic testing on a silo project?
For a typical 5,000-tonne cement silo with 200 linear meters of butt welds, 10% radiographic sampling means about 20 radiographs. At $80-120 per exposure including film, processing, and interpretation, the total cost is $1,600-2,400. That's less than 0.5% of the total fabrication cost. It's the cheapest insurance you'll ever buy.
Get It Right the First Time
Manxing has delivered over 2,000 steel silos across 30+ countries. Our welding procedures are qualified per EN 1090 and AWS D1.1, our QC inspectors hold CSWIP and AWS certifications, and our radiographic rejection rate averages 2.3% — well below the industry norm of 8-12%. If you're planning a cement, clinker, or fly ash storage silo, talk to us before you finalize the fabrication spec. We'll help you avoid the mistakes that cost projects months and tens of thousands of dollars.