Engineering Design Process for Mechanical Engineers

How real mechanical designs progress from vague 'I want to make this' to fully fleshed out hardware. Focus on decisions, iteration, and real-world constraints that junior engineers face moving from theory into actual practice.

Most guides to the engineering design process paint a clean and linear picture. Reality is far from it. Design starts with a hazy understanding of requirements, making key decisions before all of the relevant information is available. It then iterates over time, gradually producing something that works, can be produced, and will actually stand up under use.

⚙️ What the Engineering Design Process Actually Is

The engineering design process is not something you complete and put in your trophy case. Instead it is an ongoing decision process that evolves with your project as you learn new things. While it is desirable to work on idealized problems where everything can be defined perfectly at the beginning, most real-world problems require revisiting prior decisions as new information comes to light. This information might come from a test not previously performed, a supplier that is unable to meet a specified requirement, or manufacturing realities that were previously unknown but are revealed only when prototype hardware is inspected.

Your design goal is not the "perfect design" that equals ideal performance in all characteristics. In fact, the better goal is for a design that works as needed, is practical to build, and safely withstands the conditions under which it will be used. There are always trade-offs required involving performance, cost, schedule, manufacturability, and reliability. What you want to do is make these decisions on a rational basis rather than by luck.

Key Insight: Iteration is your friend, not the enemy—it is the default mode for product design and iteration is not a sign of failure. Your initial design decisions will invariably be made with incomplete information, and subsequent testing, analysis, and supplier input will then be used to iterate on those initial decisions to arrive at the best final solution. Design Process is how you manage your ignorance before arriving at a solution.

🎯 Start With the Problem, Not the CAD Model

⚠️ Common Mistake: Opening CAD before understanding the problem you are trying to solve. Geometry can be seductive, looking very active in producing something which appears to be progress, but it is not until much later that you realize you have redesigned your solution to the wrong problem around the wrong parts—thus redesigning the same thing three times in total.

Before any sketches, answer these questions:

✓ What needs to work?
Define the core function before anything else
✓ Who uses it?
User context and operating conditions
✓ What constraints exist?
Space, cost, schedule, manufacturing, maintenance
✓ What is unknown?
Investigation needed before committing

💡 Example: "Make it lighter" is worthless to tackle. However, "Reduce weight by 20% while maintaining current stiffness and keeping cost under $500 per unit at 10,000 units per year" is something that you can design to and is a guard rail against misdirected effort.

📋 Turn Requirements Into Engineering Targets

Now that we have the problem well articulated, let's turn it into some engineering targets before the scope creeps out of control and design goes off the rails.

Functional Requirements
What the design must do: structural support of the load, accommodate available space, and compatibility with operating temperature
Targets & Limits
Size, weight, cost, expected life, performance benchmarks
Priority Hierarchy
"Must have" vs "nice to have"—when deciding how to compromise, knowing the difference can make all the difference

One of the challenges of writing specification is having a response ready for the inevitable "why A not B?". However, the best response is simply to cite the requirements. They can also define what finished means, a question usually answered by schedule pressure rather than engineering judgment.

💡 Generate Concepts Before You Optimize Anything

⚡ Critical Point: The largest gains in design come from selecting the initial architecture rather than continuing to tweak a mediocre design.

A common pitfall that many engineers fall into is screening too few concepts and subsequently spending too much time optimizing a design that had little chance of success.

Effective Concept Screening:
  • Sketch multiple approaches—compare architectures, not tiny variations
  • Evaluate different load paths, mechanisms, attachment methods
  • Use rough calculations: free-body diagrams, envelope checks
  • Check packaging conflicts, manufacturing feasibility early
  • Assess standard part availability and major risks

Fast but accurate screening of design alternatives is more valuable than false precision in detailed FEA models. For example, we can quickly determine that a cantilever beam design will not work, but a supported beam design will, without ever creating a full FEA model. Therefore, it is beneficial to make a directionally correct design decision as quickly as possible, saving the high resolution FEA for once the design has been steered into a promising direction.

🔨 Build the First Feasible Design

FEASIBLE
NOT OPTIMIZED • NOT FINAL • NOT PERFECT

Once we have a concept direction, we need to build the first actual design. This design should have rough sizing done, and the key load paths and interfaces defined out so that other components can start development. We select candidate materials and manufacturing strategies based on what is really available to us.

✓ DO THIS
  • Build and test rough prototypes
  • Expose problems early
  • Get concrete feedback
  • Iterate based on learning
✗ AVOID THIS
  • Perfecting before testing
  • Endless CAD tweaking
  • Optimizing too early
  • Assuming success

The first feasible design for your product is not the end of the story, but rather the beginning. You will gain far more insights from testing a simple but viable prototype than from refining many parameters in CAD software.

📊 Move From Concept to Detailed Design

As the design matures, convert rough layouts into controlled geometry. Lock key interfaces and critical dimensions so manufacturing can quote the job and other teams can design mating parts.

📏
Lock Interfaces
Critical dimensions
🎯
Tight Control
Where it matters
📝
Add Detail
For review & analysis

Gradually add geometric detail and firm up key interfaces and dimensions. Lock down the critical dimensions, so manufacturing can determine a quote for the part, and other parties can design for fit with your engineering's components.

Validate the Design Before You Trust It

⚠️ Validation is not the same as decoration, validation is how you find out that your design actually works before you waste money producing production tooling.

1. Quick Hand Checks
Is the design any good?
2. Targeted Simulation
Use analysis where it makes a difference, not where it confirms your preconceived notions
3. Test Highest Risks First
Prototype the biggest uncertainties: mechanisms, joints, assemblies, loads
4. Compare to Requirements
Not wishful thinking—real evidence against real targets

Tests can contradict your model. Often this means that your assumptions and/or your model were wrong. Investigate the cause of failure before moving on. See How Things Fail for some general principles on failure analysis.

🔁 Iterate Without Losing Control

Projects always involve some iteration. After testing you find problems. Production uncovers issues that were hiding in your CAD model. Suppliers can't supply components to the specifications you assumed they could. But uncontrolled iteration devolves into chaos.

Controlled Iteration Framework:
  • If test results change, update the information — Don't ignore the results!
  • Make individual changes one at a time — Limit effects to single revision
  • Re-check key assumptions — After meaningful revisions
  • Stop Iterating — No impact on key decisions? Then stop.
Good Iteration
Driven by new learning
Bad Iteration
Driven by perfectionism

If changing a fillet radius from 3mm to 3.5mm makes no difference to strength, cost or manufacture, stop worrying about it. There will come a point where you have to release the design to production.

⚠️ Common Failure Points in the Process

Most failures in the design process are predictable. Most failures in design come from following established patterns in process. Even though failures are predictable, most designers fail to notice them because the failures fit within an established paradigm that feels effective.

❌ Starting CAD Too Early
Even when I don't quite get the problem I feel like I am making progress by starting to draw parts. That's really bearing out in a complete redesign already.
❌ Treating Assumptions Like Facts
Products fail because designs were based on the designer's incomplete understanding of loads, material, or usage, not because they were bad designs.
❌ Optimizing Details Before Concept Risk
Spending time making something good that it is, is a waste of time. Instead spend the time choosing the right architecture in the first place.
❌ Ignoring Manufacturing Until Late
Design for Manufacturing creates parts that cannot be reasonably produced or even made at all. Design for manufacturing needs to start very early.
❌ Confusing Analysis Output With Truth
Green light from FEA is incredibly powerful even when the initial boundary conditions were flawed; always trust but verify.
❌ Declaring "Done" Without Evidence
Shipping on time is often success regardless of whether or not any testing has been done on the design.

Your best chance to catch these mistakes is during the design process. Take advantage of it.

🏁 What "Done" Actually Means

You'd think it would be easy to tell when a design is complete, but it's a far harder line to draw than you might imagine
⏱️
Schedule Pressure
Release Early
🔍
Perfectionism
Refine Endlessly
Both are wrong.
✓ A Design Is Done When:
Requirements Are Traceable
This policy provides a clear, concise overview of all state required design elements
Major Risks Are Understood
The things that you think could fail and why you think they will not fail
Design Is Buildable
With the intended manufacturing process at acceptable cost
Key Assumptions Are Checked
Through analysis, testing, or supplier confirmation
Documentation Is Sufficient
You feel someone else could do setup, test and validate without constantly needing to refer to manual instructions for answers

"Ready enough?" I asked. Turns out, "ready enough" means different things at different times for different releases. And in the end, the test of readiness is always whether there's enough documentation, whether it's "ready enough" for end users. A prototype release is never "ready enough" for end users to read documentation for it, but a production release is always too early to be "ready enough" to release unless it has enough documentation.

Quick Process Checklist for Junior Mechanical Engineers

💡 Use this checklist throughout your design process

1
Do I understand the real problem?
It's not just what the user asked for, but what actually needs to work and why.
2
What assumptions am I relying on?
Functionality, potential uses, possible scenarios in which the design would be utilized, or any real world instances where this device could be used. Factors that could influence the design of the project such as use limitations, ease of repair/maintenance, etc. Write them down.
3
Which concept risks were actually screened?
Is the work optimally good? Could I have arrived at an even better solution?
4
What could still break this design?
Manufacturing errors, Assembly problems, Misuse, Environmental stresses, Gradual degradation over time.
5
What evidence says this design is ready?
Numbers, whether calculated, analyzed or drawn from test data, confirmings from suppliers, and signatures from Design Review.

Why these questions: 1. These questions force you to think design critically instead of just following motions. 2. People (especially management) can use these questions to understand your design decisions and you can use them to deflect questions with "why did you do it that way?"

Other Related Topics: Understanding Load Cases & Assumptions * Failure Modes * Design for Failure * Design for Manufacturing (DFM) while your design is still cost effective to make changes.