Status Scope of services. Nothing on this site is a record of completed work.
Phases Five, A to E
Stages Eleven
Returns Two
QA/QC Four layers and a release gate
Revision rules Four
How we work
The process, published in full: five phases and eleven stages, who owns each one and what leaves it, the four seats QA/QC is read from, and how revisions are controlled.
Continuous line: Seimora Dashed line: Both
Seimora is new and publishes no completed projects, so there is no portfolio here to judge us on. What there is instead is this page: the process written down at the level a fabricator would need to argue with it — who owns each stage, what leaves it, how many times the work goes round before steel can be cut, and what has to be true before anything is issued.
Everything on it is a rule we work to and a rule you can hold us to, which is a harder thing to publish than a case study and a more useful one to read.
The work itself is on the services pages. This is the shape a job takes across them, and the parts of it that are limits rather than features are on it too.
The workflow
Five phases and eleven stages, and two of them go round. A stepper would draw this as a straight line with eleven stations on it, which is the one thing about the work that is not true. Both returns are drawn below rather than left to be inferred, because a fabricator reading a process does not count the stages — they count the trips.
A — Set up the job
Nothing is drawn in this phase. What comes out of it is the material every later argument gets settled against: a written scope boundary, and the conventions the drawings will be read under. Agreeing either of them at stage 6 instead means re-doing work that was correct against the wrong assumption.
| Number | Document | Rev | State |
|---|---|---|---|
| DEMO-SEI-ZZ-XX-RP-S-0001 | Design basis: code, load cases, materials, criteria | P01 | WIP |
| DEMO-SEI-ZZ-XX-RP-Z-0002 | Information basis: origin, grid, levels, naming, development | P01 | WIP |
| DEMO-SEI-ZZ-XX-SH-Z-0005 | Coordination basis: who answers what, and under which conventions | P01 | WIP |
1 — Project intake and scope review
Owned by: Both
What leaves it: A scope list: what is ours, what is yours, what is nobody's yet
2 — Kick-off
Owned by: Both
What leaves it: Four agreed documents — design basis, document register, drawing conventions, quality plan
B — Qualify the inputs
The phase most processes do not have as a phase at all. Stages 3 and 4 are about the inputs rather than the output — what arrived, what revision it is at, whether its parts agree with each other, and which of the questions it leaves open belong to whom.
3 — Design and document review
Owned by: Seimora
What leaves it: A written check of what you sent — revision, completeness, agreement
4 — RFIs and clarifications
Owned by: Both
What leaves it: A numbered register of what the documents do not settle
WHY THIS PHASE EXISTS
Almost every expensive detailing error starts as a decision somebody downstream had to take quietly because nobody upstream would. Stages 3 and 4 exist to make those decisions loud and numbered before material is cut.
C — Produce and check
The only phase that loops inside itself: production, a review of it, and then the four-layer check below. A failure at the check does not go back to the beginning, and it does not skip the review on the way out again.
5 — Engineering, modelling, detailing
Owned by: Seimora
What leaves it: The model and the package at draft revision
6 — Internal review
Owned by: Seimora
What leaves it: A review record; comments numbered against the package
7 — QA/QC
Owned by: Seimora
What leaves it: Sign-off across the four layers below
A layer that fails returns the package to stage 5 — and it re-enters the check at stage 6, not at stage 7. Work that was corrected is reviewed again before it is checked again.
D — Get it approved
The phase a programme is usually lost in, and it loops too. What goes out is a submission; what comes back is comments; what goes out next is a revision that names the comment behind it.
Shown at full size. Scroll sideways for the whole sheet.
- Drawing
- General arrangement, grid 1
- Number
- DEMO-SEI-ZZ-XX-DR-S-0001
- Rev
- P02
- Status
- S2 Shared
- Note
- Sample. Not a Seimora project.
8 — Client and EOR coordination
Owned by: Both
What leaves it: A submission to you and your engineer of record, and comments returned numbered
9 — Revisions
Owned by: Seimora
What leaves it: A revised issue; every revision names the comment that caused it
A revision returns to stage 6, not to stage 8. The revised issue is reviewed internally before it goes back out, which is why a second submission is not the same document with one detail changed on it.
E — Issue and support
The package is issued at its final revision and indexed against the document register agreed at stage 2 — so that a field query raised long after the drawing office has moved on has a document to be answered from rather than somebody's memory.
| Number | Document | Rev | State |
|---|---|---|---|
| DEMO-SEI-ZZ-XX-SH-S-0009 | Release package, lot 1: C1 and BP1 | P01 | WIP |
| DEMO-SEI-ZZ-XX-M3-S-0010 | IFC and the native file at a stated revision; the as-built model | P01 | WIP |
| DEMO-SEI-ZZ-XX-M3-Z-0011 | As-built model | P01 | WIP |
10 — Final deliverables
Owned by: Seimora
What leaves it: The issued package at final revision, indexed against the document register
When fabrication is in scope, stage 10 branches. Coordinating a release into a shop, and assembling the records that come back out of it, are separate scopes with pages of their own: fabrication coordination and closeout documentation.
11 — Construction support and closeout
Owned by: Both
What leaves it: Answers to field queries, written against the register; a record set at closeout
The stages marked Both are not marked that out of politeness. A scope boundary one party wrote is not a boundary that was agreed; a kick-off one party attended has settled nothing; and a question whose answer belongs to your engineer of record is not a question that can be closed from this end. Those are the stages that fail when only one side does the work, and they are where a job comes apart long before anything is drawn.
Who does what on your projectQA/QC
A reviewer reading from one seat will pass work that fails from another. A connection can be correct on paper, unweldable in position, and unreachable for a field bolt-up — and one check finds one of those three.
So the check is layered by seat, not by topic. Four layers and a release gate, each with a question it exists to answer, the seat it is read from, and the stage a failure sends the package back to.
1 — Basis
Your seat
Does it match the job?
- Project requirements and client standards
- The design criteria and codes the job is governed by
Fails back to: Stage 4 — it becomes an RFI rather than an assumption
| Number | Document | Rev |
|---|---|---|
| DEMO-SEI-ZZ-XX-RP-S-0001 | Design basis | P01 |
2 — Technical
The engineer's seat
Is the engineering right?
- Drawing and model consistency
- Member sizes and dimensions
- Connections, bolts and welds
- Joist and deck coordination, where applicable
Fails back to: Stage 5
3 — Fabrication
The shop floor's seat
Can it be made?
- Paint, finish and galvanising requirements, where applicable
- Piece marks and drawing completeness
Fails back to: Stage 5
| Number | Document | Rev |
|---|---|---|
| DEMO-SEI-ZZ-XX-DR-S-0101 | Shop drawing, C1 | P01 |
| DEMO-SEI-ZZ-XX-SH-S-0102 | Plate schedule, BP1 and BP2 | P01 |
4 — Field
The erector's and the driver's seat
Can it be delivered and erected?
- Field welds, constructability and erectability
- Shipping and transport limitations
Fails back to: Stage 5
Gate — Release
The reviewer's seat
Can it be issued?
- Incorporation of the approval comments your engineer of record returned
- Final technical and document review
Fails back to: Stage 9
- Drawing
- General arrangement, grid 1
- Number
- DEMO-SEI-ZZ-XX-DR-S-0001
- Rev
- P02
- Status
- S2 Shared
- Note
- Sample. Not a Seimora project.
WHAT THESE FOUR LAYERS ARE NOT
None of these four layers is a physical inspection. Layers 3 and 4 are read on paper, from the shop's and the erector's seat, before anything is cut. Inspection of fabricated steel is a separate scope with its own inspection and test plan.
Three of the five send a failure back to stage 5, which is the work itself. Layer 1 does not, and the difference matters: a basis that does not match the job is not a drafting error, it is a question, so it goes back to stage 4 to be numbered and owned rather than quietly corrected by whoever noticed.
The release gate is not a fifth layer either. Layers 1 to 4 are about whether the work is right; the gate is about the issue. Its subject is the last round of approval comments your engineer of record returned, and the completeness of the document set against the register agreed at stage 2 — neither of which is a question about the engineering. Work can be correct and still not be issuable, and that is why the gate has a seat of its own.
Communication
Two things decide whether communication on a project works, and neither of them is a response time: whether a question can be found again later, and whether an open one can be seen now.
Acknowledged is not answered. Many RFIs are closed by your engineer of record, or by you. We are not going to promise you a date for a decision we do not make, and a hedged version of that promise would be worse than either extreme — a client reads “typically” as an undertaking and a lawyer reads it as a way out.
What is inside anybody’s control is the register the question sits in. An RFI register has three states, and this site marks them the way an inspection and test plan is marked, because they are the same idea:
- Open— raised, owned, not yet answered
- Closed— answered, against its number
- Overdue— past the date the answer was needed by
A register whose third state is invisible is a register that reports only good news, which is the failure mode of every RFI log that has ever been kept in a spreadsheet nobody opens.
Assumptions are raised, not taken
Where a drawing does not settle something — a bolt grade, a camber, an edge distance at a skew — it becomes a numbered question rather than a quiet decision. Quiet decisions are the single largest source of detailing rework, and they are invisible until material is cut.
Where an assumption is unavoidable, it is printed
Some work cannot wait for an answer. When it proceeds on an assumption, the assumption is stated on the drawing or the calculation, in a place a reviewer cannot miss, rather than held in someone's head.
Comments are traced, and open items stay visible
The revision that results from a review comment says which comment caused it, and an open item is visible as an open item.
Document and revision control
Four rules, and the third one is the one most often broken.
| Number | Document | Rev | State |
|---|---|---|---|
| DEMO-SEI-ZZ-XX-DR-S-0001 | General arrangement, grid 1 | P01 | Archived |
| DEMO-SEI-ZZ-XX-DR-S-0001 | General arrangement, grid 1: duct re-routed by its author (CL-001) | P02 | WIP |
Every issue carries a revision, and a reason
A drawing leaves at a revision mark. What changed and why it changed travel with it, so a reviewer comparing two issues is reading an explanation rather than performing a diff.
Superseding is not deleting
A superseded revision is retained and marked superseded. It is never overwritten and never removed, because the question 'what was the shop working to in March' has to stay answerable after the job is finished.
The model is the source
Drawings are issued from the model rather than edited beside it. A change made on a sheet and not in the model is a change that will be lost at the next issue and found in the shop.
One answer, everywhere
When a question is answered, the answer goes into the model. That is what stops the same ambiguity being resolved one way on one sheet and another way on the next.
How resources are selected
Seimora owns no fabrication facility, no shop and no site presence, and no facility is named anywhere on this site. So the fair question to ask here is who would actually do the work.
The honest form of an answer to that is a standard that can end in a no, rather than a list of criteria every competitor could publish word for word. Criteria are a claim. A disqualifier is a commitment with a cost, and a buyer reads the difference instantly.
Certification belongs to a fabricating shop
EN 1090, ISO 3834 and AISC certification are held by a fabricating shop against its own systems — Seimora holds none of them and claims none. No facility is named on this site.
HELD — NOT PUBLISHED
The standard applied at selection — what is tested, in what order, and the condition that ends an evaluation at each step — is not published here. It is held until the selection process is confirmed, and it is named rather than written around.
What is not on this page
A response-time commitment for questions and RFIs. A rework policy or delivery guarantee. A downloadable sample deliverable package.
None of the three is here, and not because they were forgotten. The first two are commitments this business has not made, and a number invented for a website is a number that would be discovered to be invented at the worst possible moment. The third would have to be real, redacted work — there is none yet, and borrowing somebody else’s is exactly what this site refuses to do on the work page for the same reason.
The first of those is why there is no clock anywhere above: not in the workflow, not in QA/QC, not in the section on communication. If a turnaround statement is ever published at the top of this page, this paragraph is amended in the same change rather than quietly shortened around it.
Ask for any of the three against a live enquiry and you get a straight answer rather than a page.
Hold us to it
The fastest way to test a process is against a real scope. Send one, and the answers you get back are the evidence this page is asking to be judged on.