Materials Science · Durability in real environments

Combined mechanical and environmental damage

Distinguish sustained environment-assisted cracking from corrosion fatigue, trace a possible crack-tip coupling, and use matched crack-extension records to choose informative controls and protection tests.

Useful background: Energy, stability and environmental context; Fatigue, creep and time-dependent damage; Joining, coatings and surface engineering. Each worked case explains the quantities it uses.

Back to the chapter

After reading, you can

Explain why mechanical and chemical damage can interact.
Distinguish stress-corrosion cracking and corrosion fatigue without assuming that their mechanisms cannot overlap.
Calculate and interpret finite-interval crack extension under paired loading and environmental histories.
Select controls and protection evidence that address the local crack, residual stress and actual combined exposure.

A crack changes the local environment

Mechanical opening exposes fresh surfaces and can alter transport. Chemical reactions can weaken a tip or assist growth. Stress-corrosion cracking and corrosion fatigue describe interactions that cannot be inferred from isolated dry strength and unstressed corrosion tests alone. Stress-corrosion cracking requires a susceptible material state, a relevant chemical environment and sufficient tensile stress; residual stress can contribute even without an applied external load.

Stress-corrosion cracking, abbreviated SCC, is environment-assisted cracking associated with tensile stress and a susceptible material state. It can develop during sustained loading without a repeated load cycle. Corrosion fatigue describes a corrosion-related change in fatigue damage under cyclic loading. Dry fatigue can occur without SCC, and a cyclic environmental test can contain both cycle-dependent fatigue and time-dependent SCC processes. The loading label is not a unique microscopic diagnosis.

An applied force is only one source of stress. Fabrication can leave residual stresses that remain after external forces are removed. Tensile stress near one surface can be balanced by compression elsewhere, so a zero overall force does not establish zero local tensile stress. A nominal force divided by area also does not resolve the stress field near a crack.

Keep initiation separate from propagation. A corrosion pit can alter local shape and promote the formation of a crack. A test beginning with an existing crack instead studies its subsequent growth. Neither the total mass removed nor the average corrosion penetration automatically gives that crack’s length, opening stress or growth history.

Follow one possible coupling, then test its explanation

A passive film is a surface layer that can slow further reaction. Opening an existing crack changes access to its surfaces; new fracture or local film rupture can expose fresh metal. For a susceptible passive alloy, a possible sequence is local deformation, film disruption, reaction at the exposed surface, and film repair. Repair is called repassivation. The next mechanical disturbance then encounters a chemically changed tip, rather than the original undisturbed surface.

The chemical contribution need not be simply the volume of metal dissolved. In some susceptible steels, reactions at an exposed surface can supply hydrogen that enters the metal and assists local cracking. Which step matters depends on the alloy state, local solution, electrochemical potential and loading. Corrosion, oxidation and reaction pathways develops the reaction/transport context; Temperature and environmental interactions gives a complementary film-rupture and time-scale explanation.

For example, research on Custom 465-H950 martensitic stainless steel immersed in controlled chloride environments combined fracture-surface observations with frequency-dependent growth measurements. Together they supported a time-dependent SCC contribution during cyclic loading. Cyclic disruption of the crack-tip film, enhancing hydrogen generation and uptake, was proposed as an enabling route. That proposal was not direct proof of every local reaction or a rule for other alloys and solutions.

This example explains why a sustained-load result cannot automatically be transferred to a small-amplitude cyclic history. It also shows why a fracture appearance alone is insufficient: a grain-boundary path does not by itself prove that the advance was controlled by elapsed time rather than cycles. Additional changes in waveform, chemistry and measurement response must be checked.

Assess the coupled state

Retain stress level, cyclic or sustained history, temperature and chemistry. An environment that is mild for an unstressed coupon can matter greatly for a stressed crack. A combined test should target the relevant mechanism and geometry.

The original invented comparison below uses four separate, initially precracked coupons. They have the same material lot/state, geometry, surface preparation and initial measured crack length a₀ = 0.500 mm. The crack dimension a is the same defined depth in every reading. Each exposure lasts 10.0 h at a maintained 25°C. These chosen records are not measurements of a named grade, a fit to the research example or a service prediction.

The reference atmosphere is dry nitrogen at ambient pressure. The reactive condition is an aerated aqueous 0.10 mol/L NaCl solution maintained at bulk pH 7.0. These settings define the fictional comparison; no claim of SCC susceptibility follows from those numbers alone. Bulk solution conditions need not equal the chemistry inside a crack. Real tests should record actual temperature, solution history and relevant potential rather than assume their labels guarantee equivalence.

In each environment, one coupon has sustained nominal tensile stress 100 MPa. Another cycles sinusoidally between 10 and 100 MPa at 1.00 Hz. Nominal stress uses the same original gross-section area; it is not crack-tip stress. Applied force and stress over an area explains that average. The two histories share a peak but not a mean or complete driving-force history. Environment is compared within each matched loading history, rather than attributing every difference between sustained and cyclic tests solely to cycling.

Crack extension is Δa = a_final − a₀. Over the measured interval, the average time-based rate is Δa/Δt. For the cyclic records, 10.0 h = 36000 s gives 36000 cycles, so an average per-cycle rate is Δa/36000. A sustained test has no repeated cycles to use as that denominator. Fatigue crack growth and variable loading explains why finite averages are different from a local growth law.

In these chosen records, reactive/reference extension is 0.080/0.020 = 4 for sustained loading and 0.180/0.060 = 3 for cyclic loading. Each comparison retains the same loading history and duration. The first pattern motivates investigation of sustained environment-assisted growth; the second shows an environmental difference under cycling. Neither ratio, from one chosen record per condition, establishes a population effect or uniquely assigns the microscopic route.

An unchanged nominal force history does not guarantee unchanged crack driving force as a grows. Do not add independently measured “corrosion” and “fatigue” rates as if the surfaces and mechanical states stayed unchanged. A proposed combined-growth model must be checked against combined-exposure observations, not just match two separate tests.

Original invented endpoint records, all at 25°C for 10.0 h with the same a₀ = 0.500 mm. Each row is one stipulated record; no replication, scatter estimate, constant-growth law or survival limit is supplied.
Environment	Nominal tensile history	a_final (mm)	Δa (mm)	Average Δa/Δt (mm/h)
Reference	Sustained 100 MPa	0.520	0.020	0.00200
Reactive	Sustained 100 MPa	0.580	0.080	0.00800
Reference	Cyclic 10–100 MPa, 1 Hz	0.560	0.060	0.00600
Reactive	Cyclic 10–100 MPa, 1 Hz	0.680	0.180	0.0180

Original invented comparison, not measured alloy data or a service qualification. Panel A shows the first two seconds of the prescribed nominal tensile histories that continue for ten hours. The sustained history is a solid blue line at 100 megapascals; the cyclic history is a dashed violet sine wave from 10 to 100 megapascals at one cycle per second, with mean 55 and range 90 megapascals. Both share a peak but not a mean or full loading history. Horizontal time axis is linear from zero to 2 seconds; vertical stress axis is zero-based and linear from zero to 120 megapascals. Panel B shows four crack-extension endpoint records after the same 10 hours at 25 degrees Celsius. Every coupon starts with the same defined crack depth 0.500 millimetre. Reference sustained extension is 0.020 millimetre, reactive sustained 0.080, reference cyclic 0.060 and reactive cyclic 0.180. The horizontal zero-based linear extension scale runs from zero to 0.20 millimetre. Reference bars are filled blue; reactive bars are hollow orange with dashed outlines, with explicit row labels in both cases. One selected record is given per condition; there are no replicate uncertainty bars or lines predicting intermediate or future growth. Nominal stress is based on original gross area, not the local crack-tip stress.

Change a pathway and observe the relevant damage

Choose a control that tests a stated explanation. A reactive specimen without an applied load can reveal pits or other attack, but its residual stress state still needs consideration. Matching reference/reactive tests at the same loading history tests an environmental difference. Interrupted crack and surface observations can separate pit formation, crack initiation, film change and growth of an existing crack. A single final mass does not separate these stages.

For a film-rupture explanation, useful evidence can include surface-film or scratch-repair measurements under the relevant electrochemical conditions, together with matched crack-growth observations. For a hydrogen-assisted explanation, a possible route for hydrogen entry is not enough: evidence must connect the conditions, uptake or sensitivity, and cracking response. An electrical crack-length signal also needs validation against geometry and material-state effects; not every signal change is real crack extension.

Protection should address the demonstrated route. Candidate approaches might change the susceptible material state, reduce a relevant tensile/residual stress, or limit contact with the environment. Treat these as hypotheses to test under the actual combined history. A changed processing state may also change microstructure, and a coating can be damaged or contain a local through-defect. A reduction in general mass loss alone does not demonstrate a reduction in stressed crack growth.

Protection, inspection and condition monitoring develops defect-sensitive evidence. Here, a proposed coating comparison should retain loading, temperature and environment, include relevant coating defects, and observe the crack endpoint as well as surface condition. Inspection capability must match crack size, location and orientation. “No resolved growth” means none was resolved by the validated method over that interval, not an exact zero rate or unlimited resistance.

Laboratory acceleration and real-service evidence explains the further transfer problem. A protective effect found for one material state, solution and waveform is bounded evidence for those conditions. It does not supply a universal safe stress, an inspection interval or a conversion to component life.

Worked examples

Work through the case

A metal loses little mass in a corrosion test but cracks under sustained load in the same medium. Are the observations incompatible?

Mass loss and crack propagation are different endpoints.
Localised damage can be severe without large total loss.
Assess the stressed crack environment.

No. Low general corrosion does not rule out environment-assisted cracking.

Compare the same loading history across environments

Use the four invented records. Find each crack extension and average time-based rate, then the per-cycle rates for the cyclic pair. What does the paired comparison support, and what remains unresolved?

Subtract the common initial depth. The four extensions are 0.520 − 0.500 = 0.020 mm, 0.580 − 0.500 = 0.080 mm, 0.560 − 0.500 = 0.060 mm and 0.680 − 0.500 = 0.180 mm.
Divide by 10.0 h. Average rates are 0.00200, 0.00800, 0.00600 and 0.0180 mm/h in the table’s order. These are measured-interval averages of the stipulated records, not predictions of subsequent slopes.
For the two cyclic records, N = 1.00 × 10.0 × 3600 = 36000 cycles. Their per-cycle averages are 0.060/36000 = 1.67 × 10⁻⁶ mm/cycle and 0.180/36000 = 5.00 × 10⁻⁶ mm/cycle. Do not divide a sustained result by a made-up cycle count.
Within the sustained pair the environmental extension ratio is 4; within the cyclic pair it is 3. A reference atmosphere is a comparison, not a guarantee of no time-dependent process. The nonzero reference sustained record has not been assigned a cause by these four numbers.
The records motivate combined-exposure and mechanism investigation. They do not establish a unique hydrogen, dissolution or fatigue decomposition; there are no replicate scatter or uncertainty data. They also omit the initiation period because every coupon began precracked.

The reactive records show more extension than their matched reference records here: fourfold under sustained loading and threefold under cycling. The two cyclic averages are approximately 1.67 × 10⁻⁶ and 5.00 × 10⁻⁶ mm/cycle. Mechanism, population effect and future life remain unestablished.

Test the proposed protection against the actual pathway

A candidate coating greatly reduces mass loss on unstressed coupons. The intended comparison involves precracked specimens under the reactive cyclic history above. What new test and observations would directly address the coating’s proposed benefit?

State the proposal: the coating is intended to restrict solution access and reduce the environment-assisted crack response, not merely reduce total material dissolved.
Compare coated and relevant uncoated specimens with matched substrate state, initial crack, geometry, loading, actual temperature, environment and observation time. Use repeated specimens to assess scatter; do not use the four invented records as a population model.
Include a relevant local coating defect or damage condition instead of assuming total covered area describes every exposure path. Check whether loading changes the film or adhesion and whether the crack remains accessible.
Measure the defined crack extension with a validated method and inspect local surface/coating condition. Mass loss can remain an additional endpoint, but it cannot replace the crack measurement.
Interpret the outcome within that material/defect/exposure history. The test could support a bounded reduction in observed growth; it would not prove that an existing crack has healed or establish an unrestricted component life.

Use a matched stressed, reactive coated/uncoated comparison with relevant defects, replicated crack-extension observations and local coating evidence. Unstressed mass loss alone does not test the proposed crack-protection function.

Try the reasoning

Can repeated loading break a protective film?
Show the answer and reasoning
Yes.
Mechanical and chemical states can interact. In a susceptible passive alloy, local deformation can expose fresh metal before the film reforms. The importance of that route must be tested for the actual material, solution and waveform.
Does adding isolated damage rates always predict the combined result?
Show the answer and reasoning
No.
Coupling can create another rate or mechanism. Mechanical history changes surfaces and transport; reactions can change the tip or load-bearing state. Separate rates therefore need an independently checked combined model before they are added.
An additional reading for the invented reactive cyclic coupon gives a = 0.540 mm after 5.0 h, between a = 0.500 at the start and 0.680 at 10.0 h. Calculate the first and second interval rates in mm/h. Does the full-record 0.0180 mm/h mean that every later hour will add that amount?
Show the answer and reasoning
First interval: 0.00800 mm/h. Second interval: 0.0280 mm/h, 3.5 times the first. The full-record average is not a constant future growth law.
The increments are 0.040 and 0.140 mm over equal 5.0 h intervals. Their rates are 0.040/5.0 and 0.140/5.0; the equally time-weighted mean is (0.00800 + 0.0280)/2 = 0.0180 mm/h. The changed slope could reflect an evolving crack and local state; these readings alone do not identify why.
A fictional specimen has original gross area 15.0 mm². A rig applies tensile force between 150 and 1500 N at 1 Hz. Does this match the table’s nominal cyclic stress history? Does that calculation establish the local crack-tip stress or an SCC limit?
Show the answer and reasoning
Yes: the nominal history is 10.0–100 MPa. It does not establish local crack-tip stress or an SCC limit.
Using 1 N/mm² = 1 MPa, the endpoints are 150/15.0 = 10.0 and 1500/15.0 = 100 MPa. The nominal range is 90.0 MPa and the minimum/maximum ratio is 0.10. Geometry, crack state, residual stresses and material/environment evidence remain necessary to interpret local growth or resistance.
An exposed welded specimen has no applied force. A learner calls it a zero-stress chemical control and treats an absence of visible cracks as proof that the material cannot undergo SCC. Identify both unsupported steps.
Show the answer and reasoning
No applied force does not exclude local residual tensile stress. No visible crack over one finite exposure does not establish universal SCC immunity.
Residual stresses can balance within a body while the external resultant is zero. The observed endpoint also depends on exposure duration, material state and detection capability. Characterise the relevant stress state and use matched stressed comparisons before interpreting susceptibility.
A dry reference test uses rapid cycles, while a reactive test includes long tensile holds and shows greater crack extension. Can the entire difference be assigned to the solution? Propose the missing comparisons.
Show the answer and reasoning
No. Environment and loading history changed together. Compare reference and reactive conditions under each of the two defined histories.
That paired design separates a conditional environmental comparison from a change in waveform and elapsed exposure. It still requires matching material/crack state, actual temperature and measurement, plus replication. A hold may allow different reactions or relaxation, so its effect cannot be removed by simply counting cycles.
A fracture follows grain boundaries during cyclic exposure. Is that appearance alone proof that growth was controlled by SCC rather than by cycles? What evidence would make the distinction more informative?
Show the answer and reasoning
No. Combine morphology with loading/time dependence, matched environmental controls and validated crack-growth measurements.
A fracture path describes where the crack advanced, not uniquely the rate-controlling process. Controlled changes of frequency or holds with other conditions retained can test time-versus-cycle behaviour, while local chemical or film evidence can test a proposed route. Coupled mechanisms can remain active together.

What to carry forward

Use relevant combined-exposure evidence.
General mass loss does not cover every damage mode.
SCC can act under sustained tensile stress; corrosion fatigue involves cyclic loading and an environmental interaction, and the two processes can overlap.
Use the same crack endpoint, loading history and exposure duration within an environmental comparison.
Finite crack-extension averages and a plausible pathway do not determine a unique mechanism, future life or an unrestricted protection claim.

References

Teaching resources and research supporting this lesson’s explanations and examples.