Module 12
Diagnosing computational models
The largest module in the course: what goes wrong, how each fault announces itself, and the discriminating check that separates it from the others.
What this module covers
- Run the four checks that should precede reading any result
- Trace a unit error to its source by comparing which outputs moved and which did not
- Distinguish a restraint problem from a release problem from a connectivity problem
- Recognise ill-conditioning and name its three usual causes
- Tell a converging mesh sequence from a singular one
- Diagnose a model by naming the discriminating check before naming the cause
Lessons
Four things to do before reading any result, none of which takes longer than a minute.
Start lesson →The most common serious error in structural modelling is a factor of a thousand — and the technique for finding which input it is in.
Start lesson →Three faults that look alike in the output and need completely different fixes.
Start lesson →Where the precision goes, and how to tell a mesh that is converging from one that never will.
Start lesson →- 12.5
The Model Doctor
The method, applied: name the discriminating check before you name the cause.
Start lesson →
Module checkpoint
Check what you have taken in
3 questions
Question 1
A model's moments agree with hand calculation and its deflections are ten times too large. Where is the error?
Question 2
A stress at a re-entrant corner reads 62, 85, 117 and 161 N/mm² over four successive halvings of the element size. What is the ratio between the last two values, to two decimal places?
Question 3
A model applies 5 600 kN and reports 5 599.9994 kN of reaction. Express the residual as a fraction of the applied load, in scientific notation — give the coefficient when written as x × 10⁻⁷.