TASC Engineering Design cheat sheet

Iterative cycle: define problem → gather info → generate concepts → select → model/analyze → prototype → test → refine → communicate; loops back at any stage when test data contradict assumptions.

Design review gates occur at concept freeze, preliminary design, critical design, and test readiness; each gate requires documented exit criteria before the next phase begins.

Convergent thinking narrows options using constraints and criteria; divergent thinking expands the option set first - alternating too early collapses the solution space prematurely.

Verification answers 'did we build the product right' (meets spec); validation answers 'did we build the right product' (meets user need) - both are required.

Requirements traceability matrix links each requirement to a design element and a test; orphan requirements or untested elements are audit findings.

A design freeze is a configuration baseline; changes after freeze require an engineering change request with impact analysis on cost, schedule, and risk.

Root-cause tools (5 Whys, fishbone) are applied during problem definition, not after prototyping, to avoid solving the wrong problem.

Stage-gate metrics: concept screening uses weighted scores; detailed design uses failure modes, tolerance stack-up, and cost-to-target.

A requirement is verifiable if a specific test, measurement, or inspection can prove compliance; 'user-friendly' is not verifiable, 'task completed in ≤3 steps' is.

Functional requirements state what the system must do (verb + object + measurable condition); non-functional requirements state qualities such as reliability, mass, noise, or cost.

Constraints are non-negotiable boundaries (regulatory limits, envelope, budget ceiling); criteria are negotiable preferences ranked by weight.

Well-formed requirement attributes: unique ID, rationale, owner, verification method, and acceptance threshold with units.

MoSCoW or pairwise comparison assigns weights to criteria; weights must sum to 1.0 so weighted scores are comparable across concepts.

A requirement should be singular (one 'shall' per statement); compound requirements hide untested clauses and cause ambiguous acceptance.

Assumptions must be logged with an owner and a validation date; an unvalidated assumption is a latent design risk.

Interface requirements define exchange across boundaries - mechanical mounting, power, data protocol, and human handoff - and are frequent sources of integration failure.

Functional decomposition breaks the top-level function into subfunctions; each subfunction generates a morphological chart of working principles, and combinations form candidate concepts.

Brainstorming rules: defer judgment, encourage wild ideas, build on others, one conversation at a time - quantity precedes quality screening.

TRIZ resolves contradictions (improving one parameter degrades another) using inventive principles such as segmentation, nesting, and inversion.

Pugh matrix: a datum concept scores 0; alternatives score +, −, or 0 per criterion; the highest net plus count advances, and weak criteria drive redesign.

Weighted decision matrix score = Σ (criterion weight × rating); ratings use a consistent scale (e.g., 1–5) and weights must sum to 1.0.

Selection must separate must-have screening (pass/fail gate) from want-based scoring; a concept failing any constraint is eliminated regardless of score.

Reverse engineering and benchmarking generate concepts by analyzing existing solutions' working principles without copying protected designs.

Concept risk ranking uses probability × consequence; high-risk concepts need early bench tests before committing to detailed design.

Free-body diagrams isolate a body, show all external forces and moments, and are the prerequisite for any equilibrium or dynamic analysis.

Static equilibrium: ΣF = 0 and ΣM = 0; safety factor n = failure strength / applied stress, and n must exceed the value set by the applicable code.

Stress σ = F/A (axial), σ = Mc/I (bending), τ = Tc/J (torsion); deflection and strain follow from material stiffness E or G.

Tolerance stack-up: worst-case = sum of tolerances; statistical (RSS) = √(Σ tᵢ²) - RSS assumes independent, centered distributions.

Finite element analysis requires mesh convergence checks: refine until results change less than a set percentage between mesh densities.

Model validity depends on assumptions (linear elastic, small deflection, steady state); results outside the assumption envelope must not be reported as design proof.

Units discipline: carry units through every equation and convert before substitution; dimensional mismatch is a leading cause of analysis error.

Thermal and fatigue checks are separate load cases: fatigue life uses alternating and mean stress with an S-N curve or damage accumulation rule.

Prototype fidelity is chosen by question: low-fidelity (foam, cardboard, sketches) tests form and concept; high-fidelity tests performance, tolerance, and manufacturing.

Test plan elements: objective, specimen ID, instrumentation, procedure, acceptance criteria, data sheet, and operator sign-off.

Measurement system analysis precedes design testing; gauge repeatability and reproducibility (R&R) below roughly 10–30% of tolerance is acceptable.

Design of experiments varies factors systematically to isolate effects and interactions rather than one-factor-at-a-time testing.

Acceptance testing compares measured output to requirement thresholds; durability testing accelerates time using validated acceleration factors.

Test data must record conditions (temperature, humidity, specimen lot) because untraceable data cannot support a verification claim.

Failure during test is information: classify failure mode, mechanism, and site, then feed back to design or to the test method itself.

Correlation compares prototype test results to model predictions; a mismatch beyond a set band triggers model revision, not just hardware revision.

Material selection maps function to property: stiffness (E), strength (yield/ultimate), density (ρ), thermal conductivity, and corrosion resistance.

Specific stiffness E/ρ and specific strength σ/ρ drive lightweight selection; a material chart plots these on log axes with selection lines.

Ashby-style screening applies a hard constraint filter first, then ranks survivors by a material index derived from the objective.

Process selection depends on lot size, tolerance, surface finish, and material: casting for complex shapes, machining for precision, injection molding for high volume.

Cost model: unit cost = (material + labor + tooling amortized over volume + overhead) + scrap allowance; tooling dominates at low volume.

Design for manufacturing reduces part count and fasteners; design for assembly reduces handling time and orientation difficulty.

Heat treatments (annealing, quenching, tempering) trade hardness against ductility; welding alters heat-affected-zone properties near the joint.

Tolerance costs rise steeply as tolerance tightens; specify the loosest tolerance that satisfies function to control cost.

Hierarchy of controls: eliminate, substitute, engineer (guards, interlocks), administrate (procedures, training), then personal protective equipment - PPE is last.

Risk = severity × probability; residual risk after controls must be documented and communicated, and acceptable risk thresholds come from standards or law.

Informed consent and whistleblowing duties: an engineer must report a hazard that endangers the public even when it conflicts with employer interest.

Standards are consensus technical documents; codes are standards adopted into law; both define minimum, not optimal, performance.

Conflict of interest, data fabrication, and misrepresentation of test results are core research and professional misconduct categories.

Failure modes and effects analysis ranks risk priority number = severity × occurrence × detection; high RPN items require action first.

Lockout/tagout isolates hazardous energy before maintenance; machine guarding and emergency stops are engineering controls, not administrative ones.

Intellectual property: patents protect novel, useful, non-obvious inventions for a limited term; trade secrets require active confidentiality measures.

A design report structure: problem statement, requirements, alternatives considered, selection rationale, analysis, test results, conclusions, and recommendations.

Technical drawings follow projection and dimensioning conventions; a title block carries scale, units, tolerance default, and revision level.

Revision control assigns version identifiers and change logs so every team member works from the same released document.

Team role clarity (owner, reviewer, approver) prevents orphaned tasks; a RACI matrix maps responsibility, accountability, consult, and inform.

Meeting records capture decisions, owners, and due dates; an action without an owner and a date is not a commitment.

Constructive critique addresses the design, cites evidence, and proposes an alternative; personal attribution blocks technical dissent.

Graphical integrity: axes labeled with units, consistent scales, and no truncated baselines that exaggerate differences.

Stakeholder communication tailors depth to audience - executives need risk, cost, and schedule; technicians need procedure and acceptance criteria.