The package is due out, and nobody is sure what has been checked
A calculation package for a four-storey steel frame needs to go to building control on Friday. It has been checked, mostly. The senior engineer marked up the PDF of the first version a fortnight ago. The designer has since updated the transfer beam design after an architectural change, and the stability section after a new wind assessment. Some of the mark-ups are answered, some are not, and one comment was answered by email rather than in the PDF.
Now someone has to piece together whether every section in the package has been checked in its current form. The honest answer takes an afternoon to reach.
Why checking is so hard to evidence
Checking itself is careful, skilled work. The problem is the record around it.
- Calcs are produced in several tools, such as Tekla Structural Designer, spreadsheets and hand calculations scanned to PDF, so there is no single file to mark up.
- Mark-ups are made on a PDF of one version, and the calcs move on underneath them.
- Responses to comments happen in conversation, in email or in the next version, rarely against the comment itself.
- The checker's sign-off is a signature on a cover sheet, with no record of which revision of each section it covered.
- Your quality system asks for evidence of checking, and the evidence has to be assembled after the event.
Revisions make it worse. A package might go through several internal versions before issue, and a checker who signed off section 5 on the second version has no way of knowing that section 5 changed on the fourth unless someone tells them. Most of the time nobody does, because nobody realises it matters until the question is asked.
What an unclear check record costs
| Situation | Effect on the practice |
|---|---|
| Unsure which sections are checked | Package held back, or issued with uncertainty |
| Comments lost between versions | The same issue raised twice, or not closed |
| Sign-off not tied to revision | Weak evidence if a design is questioned later |
| Checking evidence rebuilt for audit | Senior time spent on paperwork |
Senior engineers are the scarcest people in most practices, and they are the ones pulled into reconstructing check histories. Every hour spent that way is an hour not spent checking.
How we build a calc check tracker
- We map your package structure: sections, elements or members, whichever level your checking procedure works at.
- Each section is linked to its source file and revision, whichever tool produced it.
- The checker records comments against a section, with the revision they checked, from a simple web screen or by uploading a marked-up PDF that we split into comments.
- The designer responds to each comment in place, and the checker closes it or reopens it.
- When a section's source file changes after checking, the tracker marks the check as needing review.
- The package status shows, at a glance, which sections are unchecked, open, or checked at the current revision.
- A package cover sheet with the check record is generated for issue, from the same data.
The tracker does not decide whether a calculation is correct, and it does not set your checking category or rules. It records who checked what, against which version, and what was said. That is the part that currently gets lost.
We start by mapping one recent package with your senior engineer: how it was split, who checked what, and where the record fell apart. That usually shows the level of detail the tracker needs. Too fine and nobody keeps it up; too coarse and it cannot answer the questions your quality system asks.
What checking looks like with a tracker
The checker opens a list of sections waiting for them, not a stack of PDFs. The designer sees exactly which comments are still open. When the transfer beam design changes, its check status resets and both engineers know without being told.
When the package goes to building control, the cover sheet reflects the actual state of checking, and the record is there if anyone asks about it later.
Recognise any of these?
- Checkers mark up PDFs that are out of date within days.
- Comment responses happen in email or in person.
- Nobody can say which revision a sign-off covered.
- Checking evidence is assembled before audits.
- Packages are delayed while someone works out what is still unchecked.