Computational Engineering / The computational engineer in practice
Judgement, and what remains
What none of the sixteen steps covers, and what this course has deliberately not taught.
Lesson review: expert verification remains open
This lesson has received a full-context editorial review and a check of its technical explanations. Selected arithmetic has been checked separately. This is not independent engineering sign-off.
Checks needed for project use
- A computational engineering reviewer must check algorithms, convergence claims, statistical assumptions and numerical examples beyond the cases explicitly recorded.
- Independent engineering review of capstone assumptions, service coordination, fire/acoustic requirements, structural scheme feasibility and whole-life carbon methodology remains required. Illustrative calculations do not establish a project recommendation.
These are educational models and exercises. Check their assumptions before using the results. The full technical content and every diagram have not yet been independently verified.
Concrete and steel calculations retain their stated first-generation Eurocode teaching basis. Use the standards and amendments specified for your project; do not mix generations without an appropriate design basis.
Read the lesson register and outstanding review notesWhy this matters
A course that ends by implying its procedure is sufficient would be teaching the wrong lesson — and would contradict Module 1, which argued that the engineer's contribution is exactly the part no procedure covers.
So this lesson is about the residue: what the sequence cannot do, what this course has not taught, and what a reader should now go and learn elsewhere.
By the end of this lesson you should be able to
- Name what the sequence cannot supply
- State this course's boundaries honestly
- Identify what to learn next and where
- Say what remains unverified in the course itself
What the sequence cannot supply
Knowing what the structure does. Every check in Module 12 and 13 depends on having an expectation. Where does it sag, where does it hog, which members are in tension, is this symmetric? That comes from Structural Behaviour and from having looked at a lot of structures, and no procedure substitutes for it. A checklist in the hands of someone with no expectation produces ticks.
Knowing what is unusual. The most valuable diagnostic instinct is that looks wrong, and it is trained by seeing what right looks like many times. The Model Doctor is a substitute for a small part of that experience and not for the experience.
Knowing when the question is wrong. Step 1 says to define the problem. It does not say how to notice that the brief has already given away the answer, which is what happened in the last lesson's project. That noticing is judgement.
Deciding how much checking is enough. Module 13 said the depth of checking should follow the consequence. Judging the consequence is not on any list.
Knowing when to stop. Module 16 gave four stopping criteria and said each gives something up. Choosing between them is a decision about what you are willing to miss.
What this course has not taught
Stated plainly, because a course that implies completeness is misleading:
Any design resistance. This course stops at the analysis output. Member and section design to the Eurocodes is Steel Design and Reinforced Concrete Design, which have separate standards references and review requirements. Check the current requirements and the stated verification status before applying them.
Dynamics. Modal analysis, response spectra, time-history and damping are Structural Dynamics. Module 10 covered only what a dynamic analysis needs from the model.
Soil behaviour. Module 12 covered spring supports as a modelling decision; the soil mechanics behind the stiffness is Soil Mechanics.
How to write a solver. Module 9 opened the box far enough to model well and no further.
Particular software packages. The concepts here — node, element, restraint, release, geometric stiffness, load path — are the profession's vocabulary, and terminology varies between packages. Reading your own package's documentation for what it does and does not do is a validation activity, and Module 13 said so.
Current capability claims. Nothing here says what any AI system, cloud platform or quantum computer can currently do. Those claims date faster than the page.
What is unverified in the course itself
Consult the current currency register for the status, source and review date of time-sensitive claims. Carbon factors remain illustrative teaching values and need verification against suitable project data. Capability claims, standards editions and data-protection obligations also need current sources. A lesson review does not replace an independent engineering review.
What to carry
Check the model independently of the way its results are presented:
A wrong model produces neatly formatted results, plausible contours and a converged solution, and nothing in the presentation distinguishes it from a right one.
Everything else — the equilibrium check, the prediction, the sensitivity study, the three-stage check, the model record, the deterministic-check rule — exists because of that one fact. Use several checks with different failure sensitivities; presentation and convergence alone cannot establish model validity.
Check yourself
Which of these is the sixteen-step sequence least able to help with?
Practice
A course's currency register holds 16 claims, of which 8 are stable principles. How many are time-sensitive in a way that requires tracking?
Check yourself
What does this course claim to have taught, and what does it explicitly not?
Check yourself
Which habit directly helps detect a model that does not represent the intended structure?
Summary
- The sequence cannot supply an expectation, and every check depends on having one
- It cannot tell you the question is wrong, and that is where the last project turned
- This course gives no design resistance, no dynamics, no soil behaviour and no software
- Consult the current currency register for outstanding verification and dated sources
- A wrong model is presentationally identical to a right one — everything else follows from that