Both methods find surface-breaking defects. Both are cheap, fast and portable. Both have been in use for the better part of a century. And most of the time the choice between them is made by one property of the material, before any of the other considerations get a look in.
Here is how to decide, and what each method will and will not tell you.
The constraint that decides it
Magnetic particle inspection works by magnetising the component. A surface or near-surface discontinuity distorts the magnetic field and creates a leakage field at the surface, which attracts applied ferromagnetic particles into a visible indication.
That mechanism requires the component to be ferromagnetic. Carbon steels, low-alloy steels, cast irons and martensitic stainless steels are. Austenitic stainless steels, aluminium, copper, titanium, nickel alloys and most non-ferrous materials are not, and no amount of technique compensates.
So the first question is not which method is better. It is: will the material hold a magnetic field? If no, penetrant, and the decision is made. If yes, you have a genuine choice — and MPI is usually the better one.
Duplex stainless steels sit awkwardly here. Their mixed austenitic–ferritic microstructure means they are weakly magnetic, and MPI on duplex can produce confusing results. Penetrant is the safer choice unless a technique has been specifically qualified.
Where MPI wins
It finds sub-surface defects. This is the one people forget. Penetrant only finds defects open to the surface, because the mechanism is capillary action — if the penetrant cannot get in, there is nothing to draw out. MPI detects discontinuities slightly below the surface too, because the leakage field extends beyond the metal. Not deep — a few millimetres at best, depending on technique and defect size — but the difference between "surface-breaking only" and "surface and near-surface" is real.
It is faster. A penetrant inspection has a fixed time cost you cannot compress: penetrant dwell (typically 10 to 30 minutes depending on the product and temperature), then developer dwell before evaluation. MPI is close to immediate — magnetise, apply particles, look. On a run of components, the throughput difference is substantial.
It tolerates a rougher surface. Penetrant is unforgiving of surface condition, because a rough or contaminated surface holds penetrant everywhere and produces background that masks real indications. MPI copes better with as-welded and lightly dressed surfaces.
It is more forgiving of tight defects. A very tight crack may not admit penetrant at all, particularly if it is contaminated or has been smeared closed by machining or grit blasting. The magnetic leakage field does not need to physically enter the defect.
Where penetrant wins
It works on anything non-porous. Austenitic stainless, aluminium, titanium, nickel alloys, copper, and non-metals such as certain ceramics and plastics. On a mixed-material fabrication it is often the only method that covers everything.
Orientation does not matter. This is a significant practical advantage and it is underappreciated. MPI detects defects that lie across the magnetic field; a defect parallel to the field produces little or no leakage and can be missed entirely. Proper MPI therefore requires magnetisation in two directions roughly perpendicular to each other, and skipping the second direction is one of the most common ways MPI is done badly. Penetrant has no directional sensitivity at all.
No electrical supply, no magnetisation equipment. Aerosol penetrant kits go anywhere, including places where you would rather not introduce electrical current. On confined-space or remote work that matters.
No demagnetisation afterwards. Components that will later be machined, welded or used near instrumentation often need demagnetising after MPI, which is an extra step and occasionally a difficult one on large or complex geometries.
Sensitivity: fluorescent versus visible
Both methods come in colour-contrast and fluorescent forms, and this choice affects sensitivity more than the choice between methods does.
Colour-contrast — red penetrant against white developer, or black particles against white contrast paint — is evaluated in normal lighting. Convenient, portable, sufficient for a great deal of general fabrication work.
Fluorescent systems are evaluated under UV-A in a darkened environment and are materially more sensitive. The eye detects a bright indication against a dark field far better than a dark indication against a light one. Where sensitivity matters — aerospace, safety-critical, thin-section, or looking for fine fatigue cracking — fluorescent is the answer.
The trade is environmental. Fluorescent inspection requires controlled lighting, a period of dark adaptation for the inspector's eyes, and a UV source that has been checked for intensity. On an open site at midday that is a problem. In a booth it is not.
What both share, and where both fail
Both are surface methods. Neither tells you anything about the volume of the weld. A component that passes MPI or penetrant may still contain porosity, slag or lack of fusion in its interior, and if that matters you need a volumetric method.
Both depend heavily on surface preparation. Paint, scale, rust, oil and heavy oxide will defeat either. Cleaning is not preliminary to the inspection — it is part of it, and it is where most of the time on a poorly scoped job goes.
Both are sensitive to process control in ways that are invisible in the result. Penetrant dwell times that were cut short, developer applied too thickly, MPI field strength never verified, UV intensity never measured — none of these announce themselves. They just quietly reduce sensitivity, and the report still says "no relevant indications".
Both also require that the inspector can actually see the surface. Access is a genuine constraint and no technique compensates for a face you cannot get an eye onto.
Acceptance criteria: what counts as an indication
Finding something is not the same as failing something, and surface methods generate a lot of things that are not defects.
The vocabulary matters. An indication is anything the method produces. A relevant indication is one caused by an actual discontinuity, as opposed to a non-relevant indication caused by geometry — a change of section, a press fit, a machining mark, or on MPI a permeability change at a weld boundary. Distinguishing them is a substantial part of what a Level 2 is certified to do, and it is why an inspector who reports every mark on the surface is not being thorough, they are being untrained.
Relevant indications are then classified by shape. A linear indication has length at least three times its width; anything else is rounded. Most acceptance criteria treat linear indications far more severely, because they signal cracking or lack of fusion rather than trapped gas.
Beyond that, the criteria depend entirely on the code — ISO 23278 for magnetic particle on welds, ISO 23277 for penetrant, and ASME setting out its own in Section V with acceptance in the construction code. They use different thresholds, and a weld can pass one and fail another.
The instruction is the same as for every method: agree the acceptance criteria in writing before anyone mobilises. Whether a 4 mm linear indication is acceptable is a much cheaper argument to have at a desk than standing over the component.
A practical decision path
Is the material ferromagnetic? If no — penetrant. Decision made.
Is it duplex? Penetrant, unless MPI has been specifically qualified for it.
Do you care about near-surface as well as surface-breaking? MPI.
High throughput, many similar parts? MPI — the dwell times dominate on penetrant.
Rough as-welded surface you cannot dress? MPI tolerates it better.
Complex geometry, awkward orientation, or you cannot reliably magnetise in two directions? Penetrant.
Is sensitivity the priority? Fluorescent, in either method, in controlled lighting.
Does the code name a method? Then that is your answer, and a change needs a documented justification.
The mistake worth avoiding
The most common failure is not choosing the wrong method. It is choosing the right method and then running it in a way that quietly loses sensitivity: single-direction magnetisation on MPI, shortened dwell on penetrant, inadequate cleaning on either, or fluorescent evaluation in a room that is not actually dark.
Every one of those produces a clean report. That is precisely the problem — a surface method done badly does not fail visibly, it just stops finding things. Which is why the procedure, the process controls and the certification of the person doing it matter more here than the method selection ever will.
We run both methods on site and in our Cumbria, Southampton and Durham facilities, UK coverage within a couple of hours, colour-contrast and fluorescent, to ISO and ASME acceptance criteria. If you are not sure which your scope needs — or whether what you are currently getting is being run properly — ask us.
