Units
1. Introduction to Artificial Intelligence in Metallurgy: Boundaries, Validation, Responsibility and Ethics 2. AI in Material Selection and Decision Support Systems 3. Feature Prediction and Structure-Process-Property Relationships (Machine Learning) 4. Phase Diagrams, Thermodynamics and AI in CALPHAD Approach 5. Microstructure Image Analysis and Quantitative Metallography 6. AI in Optimization of Heat Treatment and Process Parameters 7. Manufacturing Processes: AI in Casting, Welding and Additive Manufacturing 8. Quality Control, Defect Detection and Statistical Process Control 9. AI in Damage Analysis, Root Cause and Breakage Investigation 10. Characterization Data Interpretation: XRD, SEM/EDS, DSC and Mechanical Testing 11. Standard Investigation, Compliance and Verification Discipline 12. Metallurgical Data Analysis and Frontiers of AI with Python
Unit 2 / 12

AI in Material Selection and Decision Support Systems

Gains:

  • Ability to structure the material selection problem with requirements, constraints and performance indices and accurately describe it to AI
  • Ability to use AI as a hypothesis generator when comparing candidate materials with the Ashby approach and weighted decision matrix
  • Ability to verify each material and feature value that AI recommends against data sheet, standard and cost/supply fact

Material selection is the most visible and responsible job of the metallurgical and materials engineer. Aluminum, carbon fiber or steel for a bicycle frame; 316 stainless or duplex stainless for a pump shaft; Which tool steel for a hot die? These decisions require simultaneously balancing dozens of conflicting constraints such as strength, weight, corrosion, temperature, machinability, cost and supply. Artificial intelligence (systems that process text and numbers and produce suggestions) accelerates the process of creating a candidate list, establishing a comparison matrix and writing the justification. But only the data sheet, standard and, where necessary, testing prove that the material you have chosen is truly suitable. In this unit, you will learn how to conduct material selection in a disciplined manner with AI.

Establishing the material selection problem correctly

A poor material choice often arises not from a bad model, but from an ill-defined problem. Asking AI “what is the best material” is pointless; because "best" depends on which function it will perform under which constraint. A sound material selection problem is defined by these four components:

  • Function: What will the piece do? (E.g. a rod carrying axial load, a vessel resisting pressure.)
  • Constraints: Limits that must not be exceeded. (E.g. yield strength ≥ 350 MPa, operating temperature 200 °C, resistance to salt water.)
  • Objective: What will we minimize/maximize? (E.g. minimize weight, minimize cost.)
  • Free variable: What can we choose? (Material and often a dimension, e.g. section.)

This framework is the basis of Ashby's material selection approach. A performance index is an expression that gives the combination of material properties that best meets a particular purpose; For example, an index such as specific strength is used for the lightest beam of a certain stiffness. Given the function and constraints, AI will help you derive the correct performance index and rank candidate material classes — but be sure to check the physical derivation of the index it derives.

Step by step: AI-powered material selection

  1. Write the requirements: Function, constraints, purpose, free variable. Let it be numerical and measurable.
  2. Identify candidate classes: Ask the AI ​​to list the material classes (steels, stainless, aluminum alloys, titanium, polymers, composites) and justify the suitability of each for this function.
  3. Eliminate with constraint: Remove grades that do not meet the constraints (e.g. the corrosion constraint eliminates carbon steel).
  4. Sort by performance index: Sort the remaining candidates by purpose (lightness, cost).
  5. Set up a decision matrix: Compare final candidates with a weighted scoring table.
  6. Verify: Obtain property values ​​of selected material from data sheet/standard, confirm supply and cost, test sample if necessary.
Tip: Never take the material properties given by AI as final values. Even if AI says "yield strength of 6061-T6 aluminum is 276 MPa", the actual value varies depending on the temper state, semi-finished product form (plate, bar, extrusion) and section thickness. Confirm the value from the EN/ASTM data sheet.

Weighted decision matrix

The weighted decision matrix is ​​a transparent and defensible method when choosing among the final candidates. Each criterion is given an importance weight, each candidate is scored on each criterion, and the weighted total is calculated. Below is an example of a seawater pump shaft (scores 1–5, sum of weights 1.0).

criterion

Weight

316L Stainless

Duplex 2205

Ni-Al Bronze

Corrosion resistance (sea water)

0.35

3

5

4

yield strength

0.25

3

5

3

Machinability

0.15

4

3

4

Cost (lower is better)

0.15

4

3

2

Supply/lead time

0.10

5

3

3

weighted total

1.00

3.40

4.30

3.45

This table highlights duplex 2205; But the table is not a decision, it is a ground for discussion. You determine the weights and scores; AI calculates them regularly and writes the justification. The basis for the scores (PREN value for corrosion, standard minimum for strength) must be verified.

three mini cases

Case 1 — Wrong temper, wrong strength. A designer asks AI "yield strength of 6061 aluminum" for a bracket; AI says "276 MPa". The designer uses this value, but the material supplied is in T4 temper and has a yield strength of only ~145 MPa. The part becomes permanently deformed at lower than expected load. Lesson: property value depends on the condition of the material (T4/T6); The value given by AI is meaningless without specifying the temper and must be confirmed from the data sheet.

Case 2 — Correct elimination. A team requests a list of candidates from AI for a valve body operating at 6 bar pressure and constant salt water contact. AI eliminates carbon steel and low alloy steels due to corrosion limitation; 316L recommends duplex and bronze and explains the PREN (Pitting Resistance Equivalent Number) concept for each. The team accepts this preliminary screening, but makes the final decision based on actual chlorine concentration and temperature corrosion test data. AI eliminated and experimented.

Case 3 — The reality of cost. For one product, AI recommends titanium Ti-6Al-4V; It is excellent in terms of lightness and corrosion. However, the context that the part will be produced in 50,000 units annually and that there is price pressure is not given. The per kilogram cost and processing difficulty of titanium make the product commercially unfeasible. The team adds context and asks again; This time a cost-effective stainless steel is chosen. Lesson: if you don't give cost and manufacturability constraints from the beginning, AI will give technically correct but commercially incorrect advice.

Copiable prompt templates

MATERIAL SELECTION FRAMEWORK TEMPLATE"Role: You are the assistant material selection specialist. Part: [description]. Fill in the frame below:- Function: [...]- Constraints (numerical): [strength, temperature, corrosion, weight...]- Purpose: [to be minimized: weight/cost]- Free variable: [material + section] Derive the appropriate performance index accordingly and show it STEP BY STEP. Give the property values with the source and Add 'must verify from datasheet' warning."

CANDIDATE SCREENING TEMPLATE"SIGN and justify material classes that do not meet the following constraints: [list constraints]. Sort remaining candidates by purpose. State by which standard (e.g. EN, ASTM) each candidate is defined. Do not give exact property numbers; give range and 'must be verified' rating."

DECISION MATRIX TEMPLATE "Set up a weighted decision matrix for the following candidates: [candidates]. My criteria and weights: [criteria=weight list]. Rate each candidate with 1-5 points and write the BASIS of each score (which feature, which source) in one sentence. Calculate the weighted total and give the result as a table. Mark the score bases as 'must be verified'."

CORROSION COMPATIBILITY TEMPLATE "Environment: [temperature, chlorine/acid concentration, pH]. Explain the PREN rationale for candidate stainless and tell which test (e.g. ASTM G48 pit test) it should be verified by. Do not finalize material selection without testing; just list candidates."

Weak prompt / Strong prompt

WEAK PROMPT: "What is the best material for a pump shaft to be used in seawater?"

STRONG PROMPT:"Role: You are the material selection expert. Part: centrifugal seawater pump shaft. Constraints: continuous seawater contact (~20,000 ppm chloride, 25-40 °C), yield strength ≥ 450 MPa, fatigue load available, to be machined by lathe, 300 pieces per year. Purpose: minimize lifetime cost. Handle candidate stainless and bronzes according to CONSTRAINTS and sort; write which standard and which verification test is required for each. Give the property values with the source, finalize."

Powerful prompt; It gives the environment numerically, sets a strength constraint, specifies the production quantity and cost objective, and requests the verification test. Thus, AI produces a list of candidates that are both commercially and technically suitable.

Common mistakes

  • Asking for the "best material" without quantifying the function and constraints.
  • Getting the property value without specifying the state of the material (temper, shape, section).
  • Not putting the cost, supply and manufacturability constraint into context.
  • Not controlling the physical derivation of the performance index that the AI ​​derives.
  • Taking the result as definitive without verifying the basis for the scores in the decision matrix.
  • Confirming environment-dependent decisions such as corrosion without experimentation.

In summary

In material selection, AI is a powerful decision support tool once you clarify the function and constraints: it produces a list of candidates, derives a performance index, constructs a decision matrix and writes justification. However, every property value must be verified from the data sheet and standard, and every corrosion/resistance critical decision must be verified by experiment. Framing the problem within the framework of function-constraint-objective-free variable is half of a good choice.

Application task

Choose a real or fictional part (e.g. an exterior fastener). Write the function, constraints, objective, and free variable numerically. Request a candidate list and performance index from AI with the "MATERIAL SELECTION FRAMEWORK" template. Then fill out the "DECISION MATRIX" template with at least three candidates. Finally, write in a paragraph which standard and test you will use to verify the yield strength and corrosion behavior of the material you have chosen.

checklist

  • [ ] I defined the function, constraints, purpose and free variable numerically.
  • [ ] I eliminated classes that did not meet the constraints with justification.
  • [ ] I got the property values ​​by specifying the material condition (temper/form).
  • [ ] I noted the basis for the decision matrix scores.
  • [ ] I put cost, supply, and manufacturability into context.
  • [ ] I made a plan to verify the selection with the data sheet, standard and necessary experimentation.
Previous Next