Material Selection, Testing and Failure
Ashby indices, weighted selection, toughness and fracture, tests, failure analysis and a worked selection.
The selection process
- Define the function (what does the part do?), the constraints that must be met (a minimum stiffness, a maximum temperature, a corrosive environment, a size limit) and the objective to optimize (minimum mass, cost, or maximum life).
- Screen the materials that fail the hard constraints (temperature, corrosion, availability, process).
- Rank the survivors by a performance index or a weighted score.
- Look for supporting information on the best few (data sheets, standards, experience), check the process (can it be made and joined? see Steels and Heat Treatment to Composites and Fiber-Reinforced Plastics) and test a prototype.
Performance indices (Ashby)
When a part must be light, the right way to compare materials depends on how it is loaded. Mike Ashby’s method derives a performance index from the function, so that the best material is the one with the largest index (the chart on Aluminum, Copper, Titanium and Other Alloys and the Ashby-style plot below show it as lines):
| Component and objective | Index to maximize |
|---|---|
| Tie rod, light and stiff | E ÷ ρ |
| Tie rod, light and strong | σy ÷ ρ |
| Beam (fixed width), light and stiff | E1/2 ÷ ρ |
| Plate (fixed width and length), light and stiff | E1/3 ÷ ρ |
| Beam, light and strong | σy2/3 ÷ ρ |
| Column, light against buckling | E1/2 ÷ ρ |
| Spring, minimum volume for the energy | σy2 ÷ E |
| Pressure vessel, leak before break | KIC ÷ σy |
Worked example: indices relative to steel
| Material | E ÷ ρ | E1/2 ÷ ρ (beam) | E1/3 ÷ ρ (plate) | σy ÷ ρ (tie) | σy2/3 ÷ ρ (beam) |
|---|---|---|---|---|---|
| Mild steel S355 | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 |
| 4140 Q&T | 1.00 | 1.00 | 1.00 | 2.54 | 1.86 |
| Aluminum 6061-T6 | 0.96 | 1.67 | 2.01 | 2.26 | 2.46 |
| Aluminum 7075-T6 | 0.96 | 1.64 | 1.96 | 3.96 | 3.52 |
| Titanium Ti-6Al-4V | 0.96 | 1.31 | 1.45 | 4.39 | 3.25 |
| Magnesium AZ31 | 0.95 | 2.05 | 2.65 | 2.50 | 3.03 |
| CFRP (along) | 3.14 | 3.98 | 4.32 | 21.40 | 13.24 |
| GFRP (along) | 0.79 | 1.80 | 2.38 | 8.15 | 6.50 |
| Pine (along) | 0.82 | 3.59 | 5.87 | 1.77 | 3.66 |
Every column is divided by the value for S355 steel, so a figure above 1 beats steel. For stiffness-limited tie rods no metal beats steel (all about 1), but for stiffness-limited beams and plates the lighter metals win handsomely (aluminum 1.7 times for beams, 2.0 times for plates), because a lighter material can be made deeper. In the strength column the advantage of 4140, titanium, 7075 and carbon is clear. Cost per kilogram changes the answer: titanium and carbon cost 40 times steel.
Weighted decision matrix
When many things matter, score each candidate from 1 (poor) to 5 (excellent) against each criterion, weight the criteria by importance (weights add to 1) and compute the weighted sum. It is a way of making the judgment explicit, not a substitute for it: test the result by changing the weights (a sensitivity check) and see whether the winner changes.
Worked example: the press piston rod
Worked example: choosing the rod material
Requirements: 80 kN compressive load, a hard surface for the seal, indoor shop environment, one-off to small series. Weights: Strength and fatigue 30 %, Wear and hardness 25 %, Corrosion resistance 15 %, Machinability and plating 15 %, Cost and availability 15 %.
| Option | Strength / fatigue | Wear | Corrosion | Machining / plating | Cost | Weighted score |
|---|---|---|---|---|---|---|
| 4140 Q&T, hard chrome | 5 | 5 | 3 | 4 | 4 | 4.40 |
| 1045 induction hardened, chrome | 4 | 5 | 3 | 5 | 5 | 4.40 |
| 304 stainless, hard chrome | 3 | 3 | 5 | 3 | 3 | 3.30 |
| 17-4PH, hardened | 5 | 4 | 5 | 3 | 2 | 4.00 |
| 6061-T6, hard anodized | 2 | 3 | 4 | 5 | 3 | 3.15 |
- Best: 4140 Q&T, hard chrome, with a weighted score of 4.40 of 5. The result agrees with the usual practice (Steels and Heat Treatment).
- Sensitivity: if the press runs in a washdown or marine environment and the corrosion weight rises to 0.35 (others scaled down), stainless options gain and 17-4PH or a 304 rod with a hard coating becomes the better choice. The scores are judgments; reasons matter more than the number.
- Check against the index: for a column that must not buckle, the index is E1/2 ÷ ρ: steel is 1 and aluminum 1.7. But the rod diameter is limited by the bore, and the steel gives the hardness the seal needs, so the index is only a guide here.
- Fatigue check: the pressure cycle makes the rod stress swing between 0 and 51 MPa (amplitude 26 MPa). The endurance limit of 4140 at 36 HRC is about 0.5 × 1 150 = 575 MPa, reduced by surface finish (0.77), size (0.85) and reliability (0.90) to 339 MPa: a margin of 13 times: fatigue of the rod itself is not a concern, but the chrome plating and the thread root are (see Fatigue and Failure).
Toughness, fracture and brittle failure
Strength alone does not make a material safe. A crack of length a in a stressed part fails suddenly when the stress intensity K = Y σ √(π a) reaches the material’s fracture toughness KIC. The critical crack length for a given stress is ac = (1 ÷ π) (KIC ÷ (Y σ))², with Y about 1.12 for a small edge crack. A tough material tolerates large cracks, a brittle one fails from a tiny flaw. For a working stress of 300 MPa:
| Material | KIC (MPa√m) | Critical crack length at 300 MPa |
|---|---|---|
| 4140 tempered at 540 °C | 100.0 | 28.2 mm |
| 4140 tempered at 200 °C | 50.0 | 7.0 mm |
| Aluminum 7075-T6 | 29.0 | 2.4 mm |
| Titanium Ti-6Al-4V | 75.0 | 15.9 mm |
| Gray cast iron | 20.0 | 1.1 mm |
| Alumina ceramic | 4.0 | 0.05 mm |
| Glass | 0.7 | 0.00 mm |
| Epoxy | 0.6 | 0.00 mm |
The 4140 rod tempered at 540 °C tolerates a crack of 28 mm; the same steel tempered at 200 °C (hard but brittle) only 7 mm: tempering for toughness is what makes it safe. Steel also suffers a ductile-to-brittle transition: below a certain temperature the toughness collapses (Charpy impact tests find it: S355J2 guarantees 27 J at −20 °C, S355K2 at −30 °C; the Titanic’s hull plate and many welded Liberty ships fractured in cold water). For service at low temperature, choose a steel with a guaranteed impact value at the lowest temperature, and aluminum, copper or austenitic stainless steel, which have no transition.
Material tests
| Test | What it gives | Standard |
|---|---|---|
| Tensile test | yield, ultimate strength, elongation, modulus | ISO 6892-1, ASTM E8 |
| Hardness (Rockwell, Brinell, Vickers) | resistance to indentation; estimates strength (UTS ≈ 3.45 HB for steel) | ISO 6508, 6506, 6507; ASTM E18, E10, E92 |
| Charpy impact | toughness at temperature; the transition | ISO 148-1, ASTM E23 |
| Fatigue (S–N) | endurance limit and life | ISO 1099, ASTM E466 |
| Fracture toughness | KIC | ASTM E399 |
| Creep and stress rupture | deformation at high temperature | ISO 204, ASTM E139 |
| Non-destructive testing | cracks, voids, thickness | dye penetrant, magnetic particle, ultrasonic, radiographic, eddy current |
| Metallography, chemistry | structure, grain size, composition | ASTM E3, E112; spark OES or XRF analysis |
Failure analysis
When a part breaks, the fracture surface tells how: keep it clean (do not fit the pieces together or wash them) and photograph everything. The usual modes:
| Mode | The surface looks like | Common cause |
|---|---|---|
| Ductile overload | cup and cone, necking, fibrous | load above yield: wrong material, a miscalculation, misuse |
| Brittle fracture | flat, bright, chevron marks pointing to the origin; little deformation | cold, a notch, a hard or embrittled material, a high rate |
| Fatigue | beach marks (arrest lines) and a small rough final fracture; origin at a stress concentration | repeated loading below yield: notches, welds, poor surface |
| Corrosion, stress corrosion, hydrogen | pits, branching cracks, intergranular fracture, corrosion products | environment plus stress (Corrosion and Protection) |
| Wear, fretting | polished, scored or pitted surfaces | sliding contact, vibration, poor lubrication |
| Creep | distortion, voids on grain boundaries | high temperature and sustained stress |
Fatigue is by far the commonest cause of failure of machine parts (some say 80 to 90 percent). Establish the origin, the mode and the root cause (a design stress concentration, a manufacturing defect, a material fault, a misuse or an unforeseen load), then fix that, not just the broken part: compare with the specification, run hardness and chemistry checks on the failed and a sound part, and check what changed. A competent materials engineer or a laboratory should handle serious failures, especially where people were hurt.
Related
- Previous: Composites and Fiber-Reinforced Plastics. Back to the overview: Materials and Processes.
- Material Properties and the Material Properties Table.
- Fatigue and Failure.
FAQ
Is the strongest material always the best?
No. Strength is one of many properties: stiffness, toughness, fatigue, corrosion, cost, availability and manufacturability all matter, and the best choice depends on the function and the objective.
What is a performance index?
A combination of properties that measures how well a material does a job: for a light stiff tie rod, E ÷ ρ. Each loading case has its own index.
How much safety factor should I use for brittle materials?
More than for ductile ones, because they give no warning: typically 3 to 10 for ceramics and glass, 2 to 3 on yield for ductile metals, and higher for fatigue and uncertain loads. Standards set the factors for regulated products.
When should I call a materials lab?
For any failure with injuries or large cost, for safety-critical parts, for unexplained repeats, and when a supplier’s material may be out of specification.
References
- Ashby, M. F., Materials Selection in Mechanical Design, Butterworth-Heinemann; Ashby, M. F., Jones, D. R. H., Engineering Materials 1, Butterworth-Heinemann.
- Callister, W. D., Rethwisch, D. G., Materials Science and Engineering, Wiley; ASM Handbook, Volume 11: Failure Analysis and Prevention; Dieter, G. E., Schmidt, L. C., Engineering Design.
- ISO 6892, ISO 148, ASTM E399 (tests); ASTM E1820 (fracture toughness); Anderson, T. L., Fracture Mechanics, CRC Press.
Last reviewed: October 2026. Properties and scores are typical illustrations; use data sheets and the standard that applies.