Erosion and Erosion-Corrosion in Piping: Elbows, Tees and Injection Points

Short answer: erosion and erosion-corrosion thin piping where flow changes speed or direction: elbows and bends, tees, reducers, the run downstream of control valves, orifices and pump discharges, and injection points. The loss is localized and directional (grooves, gullies, horseshoe-shaped pits), so a single spot UT reading per fitting often misses it. Inspectors use UT grid scanning, automated corrosion mapping and profile radiography to find the thinnest point, then track it as a condition monitoring location under API 570.

This guide is for piping inspectors, mechanical integrity engineers and reliability teams who need to decide where to place condition monitoring locations (CMLs) for velocity-driven damage and which NDE method will actually see it. It explains the mechanism, the locations the API piping documents single out, method selection, a worked example and the records the inspector of record expects. It does not reproduce any code text: for requirements, use the current edition of API 570 (fifth edition, February 2024) and its supporting recommended practices API RP 574 and API RP 571.

What erosion and erosion-corrosion are, and why the difference matters

Pure erosion is mechanical wear: solid particles, liquid droplets, slurries or two-phase flow strike the pipe wall and remove metal. Erosion-corrosion is the combined case, where a corrosive fluid would normally form a protective film or scale on the steel and the flow keeps stripping that film away, so corrosion proceeds much faster than it would in still fluid. In practice most refinery and chemical plant damage is erosion-corrosion: the fluid is corrosive enough to attack bare metal and fast or turbulent enough to keep the metal bare.

API RP 571 groups the two together and describes the damage as localized metal loss in the form of pits, grooves, gullies, waves, rounded holes and valleys, often with a clear directional pattern that follows the flow. The rate depends on the velocity and concentration of whatever is doing the impacting (particles, droplets, slurry or two-phase flow), the size and hardness of the particles, the hardness and corrosion resistance of the pipe material, and the angle of impact.

The distinction matters for inspection planning. Pure particle erosion tends to be concentrated where particles impact, such as the outside radius of an elbow carrying catalyst or sand. Erosion-corrosion driven by turbulence can show up on the inside radius, immediately downstream of a weld root, or several diameters downstream of a disturbance where the flow reattaches to the wall. If you place CMLs only where you expect particle impact, you can miss turbulence-driven loss entirely. A related mechanism, flow-accelerated corrosion in steam and feedwater systems, is managed with its own industry programmes in power generation and is outside the scope of this page.

Where erosion-corrosion concentrates in piping

The API piping inspection documents have long listed the places where velocity-driven loss is most likely. The second edition of API 570 (1998), a public copy of which remains available as an NRC exhibit, listed these locations: downstream of control valves, especially where flashing occurs; downstream of orifices; downstream of pump discharges; at any point of flow direction change, such as the inside and outside radii of elbows; and downstream of piping configurations such as welds, thermowells and flanges that produce turbulence, particularly in velocity-sensitive services such as ammonium bisulfide and sulfuric acid. That list is quoted here for history; the current edition and API RP 574 should be used for today's wording, but the physics has not changed.

• Elbows and bends. Particle erosion favours the outside radius (extrados). Turbulence-driven erosion-corrosion can attack the inside radius and the straight run just downstream. Short-radius elbows and closely spaced elbows in different planes are worse than long-radius bends.
• Tees. Where flow enters through the branch, it impinges on the opposite wall of the run; where flow turns from run to branch, the branch entry and the run wall just downstream take the turbulence. Tees used as elbows (one run end blanked) also create dead pockets on the blind end.
• Reducers and expanders. Velocity rises through a reducer, and an expander creates a recirculation zone just downstream.
• Downstream of control valves, letdown valves and orifices. Pressure drop can cause flashing or cavitation, and the resulting two-phase jet attacks the downstream pipe for several diameters.
• Pump discharges and compressor outlets. High velocity and turbulence in the first fittings after the machine.
• Protruding weld roots, thermowells and partially open valves. Local obstructions shed eddies that erode the wall just downstream.
• Injection points and mixing points, covered in detail in the next section.
• Catalyst and coke handling. In refineries, FCC reactor/regenerator systems, slurry piping and coker equipment are classic erosion services.

Injection points: their own circuit under API 570

Injection points get special treatment in API 570 because they combine concentrated chemicals, mixing turbulence and often a temperature change in a short length of pipe. API 570 defines injection points as locations where relatively small quantities of material are injected into a process stream to control chemistry or other process variables, such as water wash, corrosion inhibitor, neutraliser or caustic injection. Points where two process streams join (mixing tees) are not injection points in the code's definition, although they may need similar attention.

The code treats each injection point as a separate inspection circuit with its own boundaries and inspection frequency. The 1998 edition, for example, recommended an upstream limit of the greater of 12 inches (300 mm) or three pipe diameters, and a downstream limit of the lesser of the second change in flow direction past the injection point or 25 feet (7.6 m) past the first change in direction, with more extensive examination from 12 inches upstream to at least ten pipe diameters downstream during scheduled inspections, and named UT and/or RT as preferred methods. These are historical values: confirm current-edition circuit limits and intervals in a licensed copy of API 570 before you build a plan. The companion guide on API 570 injection point inspection covers circuit limits, CML placement and frequency in depth; this page focuses on the velocity and turbulence side.

The practical lesson is that the damage at an injection point is often a mix of mechanisms: erosion-corrosion where the injected stream impinges, acid or caustic attack where the chemical has not yet mixed, and sometimes thermal fatigue where a cold injection meets hot pipe. Scan coverage, not a handful of points, is what finds it.

Why spot UT misses erosion-corrosion

A traditional thickness programme takes one or a few point readings at each CML with a dual-element UT probe. That works well for uniform corrosion, where any reading represents the whole area. Erosion-corrosion is the opposite: the thinnest point might be a groove a few millimetres wide on one side of an elbow, and a reading a short distance away can show nearly full wall. Point readings at fixed locations can therefore report a healthy elbow right up to a leak.

Older editions of API 570 already pointed in this direction: where localized corrosion or erosion is suspected, the 1998 edition recommended wide-area techniques such as ultrasonic scanning, profile radiography or eddy current rather than relying on spot readings, and called for measurements in each of the four quadrants of pipe and fittings where appropriate, with special attention to the inside and outside radius of elbows and tees, recording the thinnest reading and its location. Modern practice extends that with grid scanning, encoded C-scan corrosion mapping and phased-array UT, which record a full thickness map instead of a few numbers. The table below compares the options.

On a typical refinery circuit, corrosion mapping or grid scanning is used at the fittings most exposed to velocity, profile RT is used where insulation removal is costly or for small-bore and complex geometry, and spot UT is kept for trending once the thinnest location is known. For the UT technique itself, see pipe wall thickness inspection UT procedures.

A worked example, described qualitatively

Consider a carbon steel overhead line on a crude unit with a wash-water injection point, three elbows before an exchanger bank, and a control valve upstream of a drum. Previous inspections took one UT reading on each elbow extrados and one at the injection point. All readings showed slow loss.

An improved plan, set by the owner's API 570 inspector, would first re-examine the damage mechanisms with the process engineers: the stream contains ammonium salts and water, so erosion-corrosion at turbulence points and corrosion where wash water impinges are both credible. It would then redefine the injection point as its own circuit using the current-edition limits, and replace single readings on each elbow with a full scan of the elbow and the first few diameters downstream. The control-valve outlet spool would get profile RT because it is insulated and short. Every scanned area would be referenced to fixed datum points so the next inspection compares like with like.

Suppose the scans find a directional groove on the inside radius of the second elbow, deeper than anything the old spot readings had shown, and localized thinning one to two diameters downstream of the injection quill. The NDE report gives minimum thickness, location and a map of each area. The inspector then calculates corrosion rates and remaining life under API 570, decides whether the fitting can stay in service until the next inspection, and sets the next date. The owner might replace the elbow with a long-radius or upgraded-material fitting, adjust the injection quill design, or add a permanent sensor at the hot spot. None of those decisions are made by the NDE contractor.

From readings to corrosion rates and intervals

Erosion-corrosion data feeds the same calculations as any other thinning: short-term and long-term corrosion rates, remaining life against the required thickness, and the next inspection date. Two cautions are specific to velocity-driven damage. First, rates can change suddenly when process conditions change, for example when throughput rises, a pump is swapped, or the injection rate changes, so the inspection plan should link to integrity operating windows and management of change. Second, comparing a scan minimum with an old spot reading at a different point can produce a misleading rate. Where the method has changed, flag it in the record so the inspector can judge whether the apparent rate is real. The guide on remaining life calculations and corrosion rates explains the arithmetic, and the inspector of record applies the code rules.

Offshore and production piping teams often meet the erosional velocity guideline from API RP 14E. Published reviews of that equation point out that it was derived for limited conditions and is frequently misapplied, for example to flows carrying sand. Treat any velocity limit as a design screening tool, not as evidence that erosion is absent; inspection data remains the evidence.

Records the inspector of record expects

• The circuit and CML identification, with drawings or isometrics showing every scanned area and its boundaries.
• The method used at each location (spot UT, grid, C-scan, profile RT), the procedure reference, technician certification and equipment calibration.
• Minimum thickness and its location within the scanned area, the scan map or radiograph, and the nominal and previous thickness for comparison.
• Notes on surface condition, temperature at the time of measurement (UT readings drift with temperature) and any areas that could not be examined.
• Photographs of external indicators such as leaks, stains or insulation damage near the fitting.

Good records let the next crew return to the same areas and let auditors under OSHA PSM or a provincial regulator see why each location was chosen.

Common mistakes on erosion-corrosion inspections

• Only checking the elbow extrados. Turbulence-driven loss can sit on the intrados or downstream of the fitting.
• Too few readings per fitting. One or two points per elbow on a velocity-sensitive circuit.
• Treating mixing tees like plain tees. They may not be injection points by definition, but mixing of dissimilar streams can be just as aggressive.
• Not re-evaluating after process changes. Higher throughput or a new injection chemical can change rates quickly.
• Relying on guided wave for fittings. It is a screening tool for straight runs, not for elbow clusters.
• Losing location control. Scans not tied to datum points make trend comparison meaningless.
• Using the 1998 numbers as current. Old public copies of API 570 are useful history; build plans from the current edition.

Regulatory overlay in the USA and Canada

In the United States, process piping at refineries and chemical plants is commonly covered by OSHA Process Safety Management (29 CFR 1910.119), whose mechanical integrity element requires inspection and testing to follow recognised and generally accepted good engineering practice. API 570 and API RP 574 are widely cited for in-service process piping. Pipelines regulated by PHMSA under 49 CFR 192 or 195 follow those rules instead, and the boundary between plant piping and regulated pipeline should be clear in the inspection plan. In Canada, provincial regulators such as ABSA in Alberta and TSSA in Ontario administer pressure piping requirements, often referencing CSA standards and owner integrity programmes. Confirm the rules that apply at your site with your jurisdiction.

How Atlantis supports erosion-corrosion inspection

Atlantis NDT performs the thickness and scanning work that the owner's API 570 inspection plan calls for: UT thickness surveys at CMLs, grid scanning of elbows, tees and reducers, encoded corrosion mapping and phased-array UT, and profile radiography through insulation by radiography crews licensed where the work is performed. Technicians are ASNT-certified and work under ASNT Level III oversight. We deliver minimum thickness, scan maps and location references to the owner's API-certified inspector, who remains inspector of record and sets corrosion rates, remaining life and intervals. Atlantis does not perform risk-based inspection or fitness-for-service assessments. See piping circuit and CML inspection and API 570 piping inspection support, or request a quote for erosion-corrosion scanning. Quotes are returned within 24 hours.

Frequently asked questions

What is erosion-corrosion in piping?

It is accelerated metal loss where flowing fluid strips away the protective film or scale that would normally slow corrosion, so corrosion and mechanical wear act together. It appears as directional grooves, gullies, pits and rounded holes at places where flow speeds up, turns or becomes turbulent.

Where does erosion-corrosion occur in elbows?

Particle erosion usually attacks the outside radius, while turbulence-driven erosion-corrosion can attack the inside radius and the straight pipe just downstream. Scanning the whole elbow and the downstream run is safer than reading one point.

What is the difference between erosion and corrosion?

Erosion is mechanical removal of metal by particles, droplets or high-velocity flow. Corrosion is chemical or electrochemical attack. Erosion-corrosion is the combination, and it is usually faster than either alone.

How do you detect erosion-corrosion?

With wide-area methods: UT grid scanning, automated UT corrosion mapping or phased-array C-scan, and profile radiography. Spot UT is useful for trending once the thinnest point is known but can miss localized grooves.

What counts as an injection point under API 570?

A location where relatively small quantities of material are injected into a process stream to control chemistry or other process variables. Points where two process streams join are mixing points, not injection points, under the code definition.

How long is an API 570 injection point circuit?

The code sets an upstream and a downstream limit around the injection point. Older editions gave specific distances; check the current edition of API 570 for the limits and intervals to use today.

Can profile radiography measure erosion?

Yes, at the tangent points of the pipe it shows the wall profile directly, and the image also shows internal loss patterns and deposits. It works through insulation, which makes it useful on short insulated spools and small bore.

Is guided wave testing suitable for elbows?

Not as a primary tool. Guided wave screening is designed for long straight runs; elbows, tees and closely spaced fittings reduce its range and sensitivity. Use scanning UT or radiography for fittings.

Does API RP 14E erosional velocity prove a line is safe?

No. It is a design guideline with known limitations, especially in sand-bearing flow. Inspection data showing actual wall loss is the evidence the inspector relies on.

Who sets the inspection interval for an eroding elbow?

The owner's API-certified piping inspector, applying API 570 and the owner's programme to the measured corrosion rate and remaining life. The NDE contractor provides the thickness data.

Next turnaround coming up? Ask us to scope elbow and tee corrosion mapping, see the erosion-corrosion glossary entry, or request an injection point scanning quote.