Computational Engineering

What has changed since 2022

The syllabus behind this course was set from a reference published in July 2022. Its engineering principles hold; several of its subject areas have moved. This page says which is which.

16 time-sensitive claims are tracked. 8 are stable principles that do not date. 7 require verification against a current authoritative source, and are marked as such wherever they appear. 1 is reported as a 2022 position rather than as current practice.

Nothing on this page is softened. A claim marked requires verification is one this work has not checked against a current authoritative source, and it is shown rather than quietly omitted — because omitting it would leave a reader unaware that there was ever a question.

Stable principle (8): A statement about mechanics, arithmetic or engineering judgement that does not have a date on it.
Verified (0): Checked against a current authoritative source on the date shown, and the source is named.
Requires verification (7): Not checked within this work. It is shown so it is not silently omitted, and it must not be relied on until checked.
As at 2022 — not checked since (1): True of the state of practice the source describes. It is reported as a 2022 position, not as today's.
Superseded (0): Known to have changed since 2022. What replaced it is stated.

Analysis methods

Static linear finite element analysis assembles element stiffnesses into a global matrix, strikes out the restrained degrees of freedom, and solves for the free displacements.

Status: Stable principle
Basis: Mechanics. Unchanged since the method was formulated and not subject to revision.
As at: 2026-07-26
What to do: None.

Second-order effects

A compression load amplifies lateral deflections by α_cr/(α_cr − 1) and initial imperfections by 1/(α_cr − 1).

Status: Stable principle
Basis: Algebra, derived in Module 10 and checked by tests.
As at: 2026-07-26
What to do: None for the algebra. Any design threshold built on it is a code matter and is not quoted in this course.

Meshing

Refining a mesh towards a re-entrant corner or a point support increases the reported peak stress without limit.

Status: Stable principle
Basis: A property of the elasticity solution, not of any software.
As at: 2026-07-26
What to do: None.

AI and machine learning

A trained model fails outside its training distribution, reflects bias in its data, and cannot be relied on to signal when it is extrapolating.

Status: Stable principle
Basis: A property of fitting a function to samples. It is demonstrated numerically in Module 18 rather than asserted.
As at: 2026-07-26
What to do: None.

The source's account of what machine learning systems could and could not do describes the state of the field in 2022.

Status: As at 2022 — not checked since
Basis: The guide is version 1.3, July 2022. The field has moved substantially since, in ways this course does not attempt to summarise.
As at: 2022-07-01
What to do: Treat any capability claim from that era as historical. This course makes no claim about the capability of any present-day system, because such a claim would date faster than the page.

The qualified engineer remains accountable for a safety-critical decision, whatever tool informed it.

Status: Stable principle
Basis: A professional-obligation statement, not a technical one. It has not been weakened by any development in the tooling.
As at: 2026-07-26
What to do: None.

Explainability techniques for large models have developed since 2022, and the practical position on what can be explained about a specific model's output has changed.

Status: Requires verification
Basis: Not checked against a current authoritative source within this work. The course therefore teaches the requirement — a safety-critical recommendation needs evidence a reviewer can inspect — and does not claim what any particular technique now achieves.
As at: 2026-07-26
What to do: Check the current position before relying on any explainability claim. The requirement in Module 18 does not depend on the answer.

Cloud and connected working

Moving computation off your own hardware moves the questions of where data sits, who can reach it, what happens when access is lost, and who is liable — it does not remove them.

Status: Stable principle
Basis: A statement about responsibility, not about any platform.
As at: 2026-07-26
What to do: None.

Specific claims about solver throughput, pricing or availability on any cloud platform.

Status: Requires verification
Basis: This course makes none. Any such figure is out of date before it is published.
As at: 2026-07-26
What to do: Get current figures from the vendor, dated, at the time you need them.

Quantum computing

Quantum computers are not in practical use for structural analysis, and the problems they are expected to help with are combinatorial rather than the linear systems that dominate finite element work.

Status: Requires verification
Basis: The first half is a statement about current practice that this work has not verified against a current authoritative source. The second half is a statement about problem classes and is stable. The course marks the difference on the page.
As at: 2026-07-26
What to do: Treat the practical-use claim as unverified. Do not repeat any projection about timescales from any source, including this one.

BIM and interoperability

Named information-management standards, their editions and their current status.

Status: Requires verification
Basis: This course teaches interoperability as concepts — federation, exchange, information loss and gain, responsibility at each handover — and does not quote a standard's clause or status.
As at: 2026-07-26
What to do: Consult the published standard for anything you intend to rely on contractually. Its edition matters and this page does not track it.

Design standards

Any design resistance, partial factor or code rule.

Status: Requires verification
Basis: This course gives none. It stops at the analysis output and points to the Steel Design and Reinforced Concrete Design courses, which carry their own verified standards profiles.
As at: 2026-07-26
What to do: Use the published standard and its current National Annex.

Embodied carbon

Numerical embodied-carbon factors for structural materials.

Status: Requires verification
Basis: The factors built into the carbon tool are illustrative teaching values chosen to make the comparison behave correctly. They have not been verified against a current published dataset within this work, and the tool says so on its own face.
As at: 2026-07-26
What to do: Replace them with dated factors from a current published source before quoting any number outside the course. The *shape* of the comparison — that scheme choice dominates material substitution — does not depend on the exact factors, and that is what the module teaches.

Structural mass alone does not determine environmental performance: the factor per tonne, the transport, the durability and the reuse potential all enter.

Status: Stable principle
Basis: A statement about what the calculation contains.
As at: 2026-07-26
What to do: None.

Security and data protection

Specific regulatory obligations on handling project data.

Status: Requires verification
Basis: The course teaches the questions to ask — where is it, who can reach it, what is the consequence of losing it, who is liable — and does not state any obligation.
As at: 2026-07-26
What to do: Take advice on the obligations that apply to your project and jurisdiction.

Digital fabrication

Model-to-fabrication workflows shift the tolerance question rather than removing it: geometry that a machine can cut still has to be transported, fitted and adjusted.

Status: Stable principle
Basis: A statement about construction, not about any machine's capability.
As at: 2026-07-26
What to do: None.

← Computational Engineering