4.3 How Discontinuities Distort Magnetic Fields

The third official MT general topic prior to 15 December 2026 is Effects of Discontinuities on Materials. The exam is asking how a real crack, a buried inclusion, a pore, a hardness change, or a keyway each distorts the flux — and how that distortion looks in particles.

Leakage is strongest when the discontinuity cuts the flux

Flux lines behave like a crowd in a corridor. A door that stands across the corridor stops people; a door that lies along the wall barely does. A discontinuity that is perpendicular to the flux (the crack length cuts the lines at about 90°) produces the strongest leakage field. As the angle closes, leakage falls. A tight crack parallel to the flux may produce no reportable indication even though the crack is real and rejectable by another method.

That is the same 90° rule as Flux Fields, now stated from the discontinuity's point of view. Interpretation items use it: a longitudinal seam that showed on the head shot and vanished on the coil shot is behaving exactly as physics predicts, not 'healing' between stations.

A 45° crack still leaks, but more weakly. Do not treat 45° as a substitute for two perpendicular magnetizations. The procedure still requires the second direction so that some shot is close to 90° for every possible crack orientation.

Surface-breaking cracks, subsurface inclusions, and rounded porosity

Not every void leaks the same way. Shape, sharpness, and depth decide how much flux is forced out of the surface where particles live.

A surface-breaking crack is the condition MT was built for. Tight fatigue and grinding cracks are planar, high length-to-width slots. They interrupt a large fraction of the near-surface flux and print a crisp line.

A rounded pore of similar opening width is a much weaker magnet. Flux slips around a sphere in three dimensions instead of being forced out through a long slot. That is why MT is a poor porosity survey compared with radiographic testing (RT), and why a single faint rounded particle cluster is not automatically 'a crack we almost missed.'

A subsurface inclusion sits under a ligament of sound metal. Some flux never leaves the steel. The leakage that does reach the surface is spread out, so the particle pile is wider and less intense than a surface crack of the same length. If the inclusion is deep enough, there is no surface indication at all.

Permeability change versus a void

A crack is a void: permeability drops from that of steel to that of air. A permeability change without a void — cold work, a hard heat-affected zone (HAZ), a brazed interface, magnetic writing from a previous residual field, or a local alloy difference — also bends flux lines. Particles can collect there even though there is no opening.

Interpretation rule:

• A void (crack, seam, lap, open inclusion) is a candidate relevant indication. Evaluate it against the acceptance standard.
• A permeability variation is often a nonrelevant indication. Confirm by residual-field mapping, demagnetizing and retesting, etching, hardness check, or another method.
• Mechanical hold-up of particles on roughness, lint, or drain lines is a false indication — there was no leakage field.

Do not call every particle pile a defect. Defect is an accept/reject decision after evaluation against the specified standard. The general exam punishes that word when the stem has only given you a buildup.

Depth sensitivity: AC versus HWDC and DC

How deep a discontinuity can still leak to the surface depends on how the field is produced.

• Alternating current (AC) is confined to a thin skin by eddy currents (skin effect). It is outstanding for surface-breaking fatigue and grinding cracks and is the usual yoke current for in-service weld toes. It is the wrong current if the procedure is hunting a subsurface shrinkage crack in a casting.
• Half-wave direct current (HWDC) (single-phase rectified) penetrates deeper than AC. The pulses also agitate dry particles so they remain mobile instead of sitting wherever they first land. HWDC plus dry powder is a classic casting and weld technique when some near-surface depth is required.
• Direct current (DC) and full-wave direct current (FWDC) penetrate deepest. Particle mobility is poorer unless the bath is flowing or the part is gently moved, because there is no AC vibration. Residual techniques after a DC shot are common on high-retentivity parts.

There is no single published 'AC sees 0.040 in, DC sees 0.250 in' number that ASNT posts as an exam constant. Depth is procedure-, material-, and magnetization-dependent. Teach the order: AC = surface; HWDC = deeper with particle mobility; DC/FWDC = deepest field, watch particle mobility. If a stem gives a procedure depth claim, use the procedure.

A surface fatigue crack on a girder toe is an AC-yoke problem. A subsurface hot tear a few millimetres under the surface of a ferromagnetic steel casting is an HWDC or DC problem. Using the wrong current type is not a 'light indication of the same crack'; it can be a complete miss.

Geometry, cold work, scratches, and other look-alikes

Real parts generate leakage that is not a crack.

Geometry. Threads, keyways, spline roots, sudden section changes, hole edges, and weld toes are flux concentrators. Particles collect along the geometric contour. That is usually nonrelevant after you recognize the feature. A crack in a thread root sits off the geometric helix or is sharper than the thread; do not dismiss every root line as 'just the thread.'

Cold work. Localized plastic deformation changes permeability. A press-fit, a stamped identification number, a peened repair, or a badly dressed grind can print a fuzzy indication that follows the worked zone rather than a single crack path.

Permeability changes from heat. A hard HAZ or a local hard spot in a plate can attract particles in a broad band. Demagnetize and retest, or confirm with hardness or another NDT method, before you write a crack.

Scratches and tool marks. A shallow scratch may simply trap dry powder (false). A deeper gouge can leak like a weak crack (relevant, or at least requiring evaluation). Wipe, re-apply with lighter particle density, and look at sharpness and whether the mark is visible under white light as a geometric groove.

Magnetic writing. Residual fields from a previous magnetizing shot, from contact with another magnetized part, or from a magnetic chuck can write streaks. Demagnetize and remagnetize in the required directions. Writing that disappears after a controlled demagnetization was never a void.

Over-magnetization and rough surface. Heavy background hides both relevant and nonrelevant patterns. Reduce field or particle application; do not 'interpret through' a purple fog of fluorescent bath.

Banding and segregation in some bar stock can produce faint, diffuse longitudinal indications from permeability layering rather than from a single seam. Compare with a known seam (sharp, tight, often one-sided) and with the product-form chapter's wrought-steel story.

Relevant, nonrelevant, and false — used on purpose

Keep the three words that the fundamentals chapter introduced, now applied to MT only:

• Relevant indication: leakage from a discontinuity that must be evaluated (crack, seam, lap, open inclusion).
• Nonrelevant indication: real leakage or particle collection from a condition the procedure does not treat as rejectable — geometry, permitted section change, magnetic writing, many permeability variations.
• False indication: particles present for a non-magnetic reason — lint, dirt, drain lines, inadequate removal of excess bath, mechanical lodging in a scratch.

A Level II who writes 'defect' on a keyway corner fails the exam item even if the particles were spectacular.

Realistic exam scenarios

A carbon-steel pin shows a sharp longitudinal indication after a circular head shot and nothing after the coil shot. That is the expected orientation behavior of a longitudinal seam, not a random 'intermittent' crack. Evaluate it as relevant against the acceptance standard.

An AC yoke on a girder toe finds a tight toe-line indication. The same area examined with a DC prod setup looking for subsurface shrinkage in an adjacent casting pad is a different technique for a different depth. Do not treat 'we already MT'd it' as covering both.

Dry powder hangs in a coarse grind scratch after an AC yoke pass. White-light visual shows the groove. After a light reclean and a thinner powder cloud, the linear pile is gone. Classify the first pile as false hold-up, not as a healed crack.

A press-fit collar leaves a fuzzy circumferential band after a coil shot. Demagnetization and a second coil shot still show a broad band exactly at the interference fit. Another method (or disassembly) is needed before anyone calls it a crack. Treat it as a permeability/geometry nonrelevant until proven otherwise.

Rounded fluorescent specks appear in a weld cap after an AC yoke shot. They do not line up, they wipe partly with the bath, and radiography of the same weld later shows scattered porosity. Those specks were weak surface pores or hold-up — not a substitute for an RT porosity map, and not automatically rejectable cracks.

What this topic is testing

Ask four questions on every indication stem: Did the discontinuity cut the flux? Was it open to the surface or buried? Was the current type capable of that depth? Are you looking at a void, a permeability change, geometry, or dirt? The correct classification — relevant, nonrelevant, or false — falls out of those answers.