Gains:
- Ability to explain material selection criteria (strength, density, cost, corrosion) and Ashby index approach
- Ability to produce candidate bill of materials, comparison matrix and justification draft with AI
- Ability to verify AI recommended material properties from reliable data sheets and standards
The choice of materials largely determines whether a machine part will work, how long it will last, how much it will cost, and how it will be produced. The same geometry becomes a completely different part when made from steel, aluminum or composite. Material selection; It is a matter of establishing a balance between strength (the stress that the material can bear without suffering permanent damage), density (unit volume mass), hardness, corrosion resistance, cost and manufacturability. In this balance, artificial intelligence is a powerful aid in generating candidates, creating comparison tables and drafting justification. However, the material property values returned by AI may be approximate, outdated or fabricated; Therefore, each numerical value should not be included in the design without being verified from the manufacturer's datasheet or material standard. In this unit, you will learn the logic of material selection, the Ashby approach, and how to safely incorporate AI into this process.
Material Selection Criteria
There is no single "best" when choosing a material; Priority varies depending on application. The basic criteria are: mechanical properties (yield strength, tensile strength, modulus of elasticity, ductility, fatigue strength), physical properties (density, thermal expansion, thermal conductivity), environmental resistance (corrosion, temperature, UV), manufacturability (machinability, weldability, castability) and economy (unit cost, availability, recycling). A good choice weights these criteria according to the need of the application.
Tip: Select the material in the order "function first, then material". First clarify what the part is supposed to do (light, hard, withstand heat); Decide on the material brand at the end. Ask the AI first "which material family for this function", do not ask for the brand directly.
Ashby Approach: Performance Index
The method developed by Cambridge's Michael Ashby systematizes material selection with a "performance index" based on the ratio of two properties. Ashby diagrams compare thousands of materials at a glance by placing two properties (e.g. strength-density, modulus-of-elasticity-density) on logarithmic axes and visualize the best set of candidates for a particular purpose.
Classic example: for a bar of minimum mass that does not break at a given load, the performance index is σ/ρ (ratio of strength to density). Materials with a higher ratio provide lighter solutions. For a beam with minimum mass and a certain stiffness, the index is E^(1/2)/ρ. In other words, the "material with the highest index according to the objective function" is selected, not the "highest strength material".
Purpose
Performance index
desired to be high
Light and durable rod
σ_flow / ρ
High strength, low density
Light and rigid beam
E^(1/2) / ρ
High modulus, low density
Light and rigid plate
E^(1/3) / ρ
High modulus, low density
Cheap and durable
σ_flow / (ρ cost)
High strength, low cost
Caution: The Ashby index narrows down the correct material family but does not make the final decision. Corrosion, manufacturability, supply and actual operating conditions are not included in the index; The engineer evaluates these separately.
Step by Step: Material Selection with AI
- Define the function and constraints. Load type, temperature, environment, mass/cost target.
- Determine the objective function. Is lightness, rigidity or cost your priority?
- Have AI produce a family of candidates. Not the brand, but the material group first.
- Ask for a comparison matrix. Make the criteria columns and the candidates rows.
- Verify each value. Cross check with datasheet/standard.
- Justify the decision. Why this material, with what risk?
Prompt that generates candidate material
Role: Mechanical engineer with experience in material selection.Application: A bracket carrying a vertical load of 30 kg, working in an outdoor environment.Priority: 1) corrosion resistance, 2) low mass, 3) low cost.Task: Recommend 4-6 candidate MATERIAL FAMILY (family, not brand). For each candidate: rough strength range, density, corrosion behavior, manufacturability, typical relative cost. Rule: If you give an exact number, state that it "must be verified" and write from which source (standard/datasheet) I need to verify it.
Prompt asking for comparison matrix
Table the following candidates as a comparison matrix: 6061-T6 aluminum, AISI 304 stainless, S355 structural steel, glass fiber reinforced polyamide. Columns: yield strength, density, corrosion, machinability, relative cost, typical use. State that the values are approximate. At the end: write with which standard/datasheet I should verify the strength value in each line.
Verification (fact-check) prompt
Question this claim with the meticulousness of a materials engineer, do not agree with me: "The yield strength of AISI 304 stainless is 290 MPa."- Under what conditions (annealed, cold worked) is this value valid?- In which standard is it defined (e.g. relevant EN/ASTM)?- How does the value change? Tell me where to look for verification.
Weak Prompt / Strong Prompt
Weak prompt:
What is the best material?
No applications, priorities and restrictions; AI gives a generic and unverifiable answer, and acts as if there is no single "best" ingredient.
Powerful prompt:
Suggest 4 candidate material families for a clamp carrying a cyclic load of 500 N, exposed to seawater, heated to 80°C. Priority: corrosion > fatigue > cost. For each candidate, write the plus/minus and which feature I need to verify and where. Mark the exact numbers as "must be verified".
The second prompt gives the environment, temperature, load type and priority; It expressly leaves the burden of verification to the engineer.
Three Mini Cases (By Numbers)
Case 1 - Fictitious datasheet. “Yield strength of 7075-T6 aluminum,” an engineer asks the AI; AI gives 505 MPa. The engineer looks at the manufacturer's datasheet: the typical value is indeed around ~503 MPa, which is fine. However, the same AI says "7075-T6's resistance to sea water is excellent"; whereas 7075 is prone to stress corrosion cracking. The engineer catches this misinterpretation and notes the need for anodizing/plating. Lesson: Even if the AI gives the number correctly, its interpretation may be wrong.
Case 2 - Save 3 kg with Ashby. For a drone arm, S235 steel (ρ = 7850 kg/m³, σ ≈ 235 MPa) is initially considered. Compared to the Ashby index σ/ρ, 6061-T6 aluminum (ρ = 2700 kg/m³, σ ≈ 276 MPa) gives a much higher index. Providing the same strength, the aluminum arm weighs ~0.16 kg per piece instead of ~0.45 kg compared to the steel arm; total savings of ~1.2 kg across four arms. The decision is justified by Ashby logic, the values are verified from the datasheet.
Case 3 - Cost index surprise. Stainless 316L is chosen because it "resists everything". But the application is indoor, low load. AI, with its σ/(ρ·cost) index, shows that galvanized S275 steel does the same job at approximately 4-5 times cheaper than 316L. The engineer finds the corrosion risk acceptable for the interior and switches to galvanized steel; unit cost drops from 100 TL to ~22 TL. The correct index prevents unnecessary costs.
Common mistakes
- Using AI without verifying its value: Neglecting the temper/heat treatment condition and obtaining the wrong strength.
- Thinking "highest strength = best": Overlooking the objective function (lightness, cost).
- Reducing corrosion to a single line: Skipping mechanisms such as stress corrosion and galvanic couple.
- Looking at a single property: Looking at strength and forgetting ductility, fatigue or workability.
- Jumping into brand selection too early: Locking in on a particular alloy before the function is clear.
In summary
- Material selection is a purposeful balance between strength, density, corrosion, manufacturability and cost.
- The Ashby approach narrows the set of candidates with a performance index (e.g. σ/ρ) according to the objective function.
- AI is powerful in candidate generation, comparison matrix, and justification outline; but their values must be verified.
- Each numerical feature should not be included in the design without being verified from the manufacturer's datasheet or the relevant standard.
- Even if the AI's number is correct, its interpretation (such as corrosion behavior) may be wrong; engineer checks.
Application task
Choose an actual part (e.g. bike frame tube, pump shaft, exterior bracket). Write down the environmental and load conditions of the application, determine the priority (lightness/cost/corrosion). Have the AI produce 4 candidate material families and a comparison matrix. Then verify at least two strength/density values in the matrix against a manufacturer's datasheet or standard; If there is a discrepancy, note why. Finally, choose a performance index that fits the objective function (such as σ/ρ), compare the two candidates, and justify your decision with a paragraph.
checklist
- [ ] The load, temperature, environmental conditions and priority of the application are written.
- [ ] The material family was determined first, the brand was left last.
- [ ] A performance index suitable for the objective function was selected.
- [ ] At least two values given by the AI have been verified with the datasheet/standard.
- [ ] Corrosion and manufacturability were evaluated separately (not reduced to a single line).
- [ ] The final material decision was written with justification and left to the engineer's approval.