Magnetic Particle Testing for Steel Silos: What Every Plant Owner Should Know

Magnetic particle testing (MT) is a non-destructive method that finds surface and near-surface cracks in ferromagnetic steel — the exact failure mode that brings silos down. For cement, clinker, coal, and slag storage, MT on shell welds, hopper seams, and stiffener connections isn't optional. It's the cheapest insurance you'll ever buy against a catastrophic release. Here's what actually matters when you're specifying it.

How Magnetic Particle Testing Works on Silo Welds

The Physics in Plain English

Run a magnetic field through steel. Any crack perpendicular to that field interrupts the flux lines and creates a leak — a "flux leakage field" — right at the defect. Spray iron particles on the surface, and they cluster at that leak like filings around a magnet. You see the crack. Simple concept. Tricky execution.

The field has to be oriented correctly. A crack parallel to the magnetization direction won't show up. That's why we always use two perpendicular magnetization passes — longitudinal and circumferential — on every weld we test. One pass misses what the other catches. I've seen inspectors skip the second pass to save time. That's how a 2-millimeter toe crack on a 4,000-tonne clinker silo went undetected for three years until it propagated through 60% of the shell circumference.

Wet Method vs. Dry Method

Wet MT suspends fluorescent or visible dye particles in a carrier liquid (usually water with corrosion inhibitors). It flows into tight, shallow discontinuities and gives you high sensitivity — down to 1-millimeter surface-breaking cracks under good conditions. This is what we use on silo shell welds, hopper seams, and manway penetrations.

Dry MT blows powder directly onto the surface. It's faster for large areas and works better on rough, as-welded surfaces where wet particles pool unevenly. But it's less sensitive to tight, shallow cracks. For silo work, I default to wet fluorescent MT on finished welds and reserve dry powder for pre-weld joint prep checks on bevels and root passes.

The fluorescent method under UV-A light (365 nanometer wavelength, minimum 1,000 µW/cm² at the inspection surface per ASTM E1444) gives you contrast that visible dye can't match. A crack that's invisible under white light jumps out in blacklight. That's not a nice-to-have. On a 30-meter-diameter silo with 800 linear meters of weld, your eyes are tired by hour three. Fluorescent contrast keeps you honest.

AC vs. DC Magnetization

Alternating current magnetizes only the surface layer — skin effect limits penetration to about 1-2 millimeters. That's perfect for surface cracks. Direct current (rectified or battery-powered) penetrates deeper, up to 5-6 millimeters, and finds subsurface defects like buried lack-of-fusion near the root.

For silo shell welds, we use AC prods for surface-breaking crack detection and switch to a DC yoke or headshot for subsurface verification on full-penetration joints. The prods are spaced 75-200 millimeters apart, and we apply current at 3-5 amps per millimeter of prod spacing. Too little current and you miss tight cracks. Too much and you burn the surface at the prod contact points — which creates its own defects and a whole new set of arguments with the welding contractor.

Where to Test: Critical Zones on a Steel Silo

Shell-to-Hopper Connection

This is the highest-stress weld on the entire silo. The full weight of the stored material — plus seismic loads, wind loads, and thermal cycling — funnels through this joint. On a 5,000-tonne cement silo, that connection sees compressive loads exceeding 800 kilonewtons per meter of circumference. Any crack here is a structural emergency.

We MT 100% of the shell-to-hopper weld on every project. No sampling. No "statistical representation." Every millimeter. The weld is usually a full-penetration T-joint with a backing ring or a one-sided closure weld, and the toe area on the shell side is where fatigue cracks initiate. That's where we concentrate the prod passes.

Vertical and Circumferential Shell Seams

Vertical seams carry hoop tension from material pressure. Circumferential seams carry the vertical dead load plus wind-induced bending. Both get MT, but the priority depends on the silo's aspect ratio.

On tall, slender silos (height-to-diameter ratio above 2.5), vertical seams are critical. On low, wide silos (ratio below 1.5), circumferential seams govern. Most cement and clinker silos fall in the 1.8-2.2 range, so we test both equally — typically 20-30% of vertical seams and 100% of circumferential seams, with 100% on the lower three shell courses where pressure is highest.

Here's a number that matters: the lateral pressure of stored cement at the bottom of a 40-meter-tall silo is roughly 250-300 kilopascals. That translates to hoop tension of 3,750-4,500 kilonewtons per meter in a 30-meter-diameter shell. A 10-millimeter lack-of-fusion defect in a vertical seam under that load doesn't stay 10 millimeters for long.

Stiffener and Ring Beam Welds

External stiffeners — usually channel or angle sections welded to the shell — resist wind and seismic loads. Their attachment welds are fillet welds, typically 6-10 millimeters, and they're loaded in combined shear and bending. Toe cracks at the stiffener-to-shell junction are common, especially on the windward side where cyclic loading from vortex shedding adds fatigue.

We MT the first and last 100 millimeters of every stiffener-to-shell weld, plus any mid-span welds on stiffeners longer than 2 meters. Ring beams at the top of the silo — supporting roof loads and sometimes a rail-mounted roof — get 100% MT on all attachment welds.

Standards and Acceptance Criteria

Which Standard Applies

For structural steel silos, the primary reference is AWS D1.1 (Structural Welding Code — Steel) for acceptance criteria and ASTM E1444/E1444M for the MT method itself. In European projects, EN ISO 17638 covers the technique and EN ISO 23278 defines acceptance levels. Chinese projects follow GB/T 15822 and GB 50205. The method is similar across standards; the acceptance thresholds differ slightly.

AWS D1.1 for statically loaded structures accepts surface cracks up to 3 millimeters in length in tension zones. For cyclically loaded structures — which includes most silos due to filling and emptying cycles — the acceptance criterion drops to zero for cracks of any length. That's right: any detectable crack is a reject. You repair it and re-test.

What "Acceptable" Actually Looks Like

Under AWS D1.1, relevant indications are evaluated by length, not width. A linear indication (crack, lack of fusion, slag line with sharp edges) longer than 3 millimeters in a static structure is rejectable. In a cyclic structure, any linear indication is rejectable. Rounded indications (porosity, spherical slag) are evaluated by size and spacing — generally, isolated pores under 3 millimeters are acceptable if they don't cluster.

The inspector's call on whether an indication is "linear" or "rounded" is where experience matters. A 2-millimeter slag inclusion with a sharp tail looks linear under UV light. A 4-millimeter pore with a smooth profile looks rounded. The difference determines whether you grind and re-weld or sign off. I've watched Level II technicians disagree on this exact call. That's why we require Level III review on any indication within 50 millimeters of a structural joint.

Field Conditions That Make or Break an MPI Inspection

Surface Preparation

MT is a surface method. If the surface is covered in mill scale, rust, heavy paint, or weld spatter, you're not testing the steel — you're testing the coating. The surface must be clean and dry, with no more than a thin, uniform layer of primer (under 50 micrometers). We specify grinding or wire-brushing of the weld toe area plus 50 millimeters on each side before MT begins.

On a project in Vietnam, the painting contractor had applied 200 micrometers of epoxy primer over the shell welds before we got to site. The MT showed nothing — not because there were no defects, but because the coating masked them. We stripped 120 square meters of primer with needle guns, re-tested, and found four toe cracks in the lower shell course. The delay cost three days. The alternative — not finding those cracks — would have cost a lot more.

Temperature, Humidity, and Access

MT works from -10°C to 60°C ambient, but the sweet spot is 10-35°C. Below 5°C, the carrier liquid doesn't flow properly and particles clump. Above 45°C, the liquid evaporates before indications form. In desert environments — I'm thinking of a clinker silo in Saudi Arabia where ambient hit 52°C — we tested between 4am and 9am only. The crew started at 0330 to prep surfaces before sunrise.

Humidity above 85% causes condensation on the surface, which dilutes the particle suspension and washes out faint indications. We carry a portable dehumidifier for enclosed hopper inspections and a simple wipe-test — if a white cloth picks up moisture from the steel, we wait.

Access is the practical killer. MT requires the inspector to see the surface at close range — ideally 300-600 millimeters — under UV light. On a 40-meter-tall silo, that means scaffolding or a suspended platform. Scaffolding costs $15-25 per square meter per week. A swing stage runs $800-1,500 per day including operator. Budget for access before you budget for the inspection itself. I've seen projects where the MT line item was 40% of the total NDT budget and access was another 35%.

Common Defects Found in Silo Welds

Toe Cracks and Undercut

Toe cracks initiate at the weld toe — the junction between weld metal and base metal — where stress concentration is highest. They're usually 1-5 millimeters long, perpendicular to the weld direction, and caused by hydrogen embrittlement or solidification stress. On silo shells, they appear most often on the shell side of circumferential seams where the weld profile creates a sharp transition.

Undercut is a groove melted into the base metal adjacent to the weld toe and left unfilled. It acts as a stress riser and crack initiation site. AWS D1.1 allows up to 1 millimeter of undercut for static structures and zero for cyclic. In practice, any undercut deeper than 0.5 millimeters on a silo shell gets ground out and re-welded, because the filling-emptying cycle makes every silo a cyclic structure whether the drawings say so or not.

Lack of Fusion and Incomplete Penetration

Lack of fusion (LOF) is a planar defect — a crack-like discontinuity with sharp edges — that MT detects reliably when it breaks the surface. It occurs when weld metal doesn't fuse to the base metal or between weld passes. On full-penetration silo welds, LOF at the root is the most dangerous defect because it's on the tension face and loaded directly by hoop stress.

Incomplete penetration is similar but specific to the joint root — the weld didn't reach the back side of the joint. Both defects show as linear indications under MT. Both are rejectable in cyclic service. The repair is straightforward: grind out the defect from both sides, re-weld, and re-test. But finding them in the first place requires the right magnetization angle and enough current density. I've seen MT reports that listed "no relevant indications" on welds that radiography later showed had 40-millimeter LOF segments. The MT was done with prods spaced 250 millimeters apart — too wide for the defect orientation. The standard says 75-200 millimeters. The inspector ignored it.

Cost, Timing, and Integration with Silo Projects

What MT Actually Costs

Field MT on a steel silo runs $8-15 per linear meter of weld, depending on location, access difficulty, and whether you're using fluorescent or visible dye. A typical 3,000-tonne cement silo — 20 meters diameter, 35 meters tall — has roughly 600 linear meters of shell welds and 63 meters of hopper circumference weld. At $12/meter average, that's $8,000-9,000 for a complete MT survey. Add access costs and you're at $15,000-25,000 total.

Compare that to the cost of a silo failure. A 2019 incident at a slag storage facility in Eastern Europe — shell crack propagated through a circumferential seam, 1,200 tonnes of material released, cleanup and rebuild cost $2.3 million. The MT survey that would have caught the initiating crack? Maybe $12,000. The math isn't close.

When to Schedule MT in the Build Sequence

MT happens after welding is complete and the weld has cooled to ambient temperature — minimum 24 hours for welds over 25 millimeters thick to allow hydrogen to diffuse. It happens before painting, before insulation, and before the silo is put into service. On a typical EPC schedule, MT is a 3-5 day activity sandwiched between mechanical completion and pre-commissioning.

We also specify MT on repair welds — any weld that was ground out and re-welded gets re-tested at 100%. And we recommend a follow-up MT survey at 12-18 months of service, focused on the shell-to-hopper connection and the lower three shell courses. That's when fatigue cracks from the first filling cycles start to show up.

Case Study: 4,000-Tonne Clinker Silo — Indonesia

A cement plant in East Java commissioned a new 4,000-tonne clinker silo — 24 meters diameter, 38 meters shell height, bolted construction with 8-millimeter shell plates. The EPC contractor (not us — this was a project we were brought in to consult on after the fact) had completed welding and applied a shop primer before MT was performed. The MT was done with visible dye, white-light method, by a local NDT contractor using prods spaced at 200 millimeters.

The report showed "no relevant indications" on 95% of the welds tested. When we arrived for a second-opinion survey, we stripped the primer on the lower four shell courses and the hopper connection, switched to wet fluorescent MT with prod spacing at 100 millimeters, and found 14 linear indications in the circumferential seams — 8 toe cracks (2-7 millimeters long) and 6 lack-of-fusion segments (10-35 millimeters long). Three of the LOF defects were on the shell-to-hopper weld.

The root cause was a combination of factors: the primer masked surface defects, the wide prod spacing missed defects oriented parallel to the prod axis, and the visible dye method lacked the sensitivity to catch tight toe cracks under field conditions. All 14 defects were ground out, re-welded, and re-tested. The delay was 11 days. The cost of the rework was $18,000. The cost of not finding those defects? A clinker silo operating at 80% capacity with a 35-millimeter LOF defect in the hopper connection is a structural failure waiting for the right wind load or seismic event.

Frequently Asked Questions

Can MT detect defects inside the weld, or only on the surface?

MT detects surface and near-surface defects — typically down to 2-3 millimeters below the surface with AC magnetization and up to 5-6 millimeters with DC. It won't find a defect buried 10 millimeters deep in the weld center. For volumetric inspection of the full weld cross-section, you need ultrasonic testing (UT) or radiographic testing (RT). On critical silo welds, we often specify both MT and UT — MT for surface and near-surface, UT for the interior. The combination gives you full coverage.

How long does MT take on a typical silo?

A two-person MT crew — one technician operating the magnetizing equipment, one inspector interpreting indications under UV light — can cover 40-60 linear meters of weld per 8-hour shift under good conditions. On a silo with 600 meters of shell welds and 60 meters of hopper welds, that's 11-16 shifts for 100% coverage. With access setup and surface preparation, plan for 5-8 working days on a typical 3,000-5,000-tonne silo.

Is MT required by code, or is it just good practice?

It depends on the jurisdiction and the standard. EN 1993-4-1 (Eurocode 3: Design of steel structures — Silos) requires non-destructive testing on a percentage of welds, with the percentage increasing for higher consequence classes. For Consequence Class 3 silos — which includes most industrial silos storing hazardous or high-value materials — the code requires NDT on 100% of structural welds. AWS D1.1 requires visual inspection on all welds and NDT on a percentage specified by the engineer. In practice, for silos storing cement, clinker, coal, or slag, we specify 100% MT on all shell and hopper welds. It's not just good practice — it's the standard of care that any competent engineer would specify.