← All stories
Purpose BuiltNovember 19, 2026

Two Books, One Building

A permit set is drawings plus a separate book of structural math — CrossBeam reads every member in one against the other.

A thick structural plan set open on a desk beside a bound calculation package, one beam mark circled on each.
WHERE THE MATH MEETS THE DRAWING

Every permit set for a building with real structure arrives in two parts, and the drawings are only one of them.

The other is the calculation package — the structural engineer's book of math. Page after page of load paths and member sizes and the arithmetic that proves the building will stand up, bound separately and handed across the counter alongside the plans. The drawings show the building. The calcs prove it. And the two are supposed to describe the same house, down to the last beam.

Here's the thing: they're written at different moments, often by different hands, and almost nobody reads them line against line.

The second book

A plan set agreeing with itself is its own kind of problem — the roof plan matching the elevations, the floor area reconciling from page to page. But that's one book, arguing with itself. The calculation package is a separate book, and it has to agree with the drawings too.

That reconciliation is slow, eye-glazing work. The calcs live in one document and the beams live in another, and matching them means flipping between the two, mark by mark, confirming that the shape sized in the math is the shape drawn on the plans. A person checks a handful — pulls three or four members, confirms the sizes line up, and moves on. Checking every single member by hand against a separate book is the kind of task that eats a whole morning and rarely gets one.

So it gets sampled. And a sample is exactly where a mismatch hides.

Member by member, both directions

CrossBeam now lines up every structural member as it was calculated against how it was drawn — not a sample, the whole roster. It matches them by their marks, FB16 to FB16, and compares the values that matter: the size, the number of spans, the lengths. When it can't confidently line two up, it says so rather than assuming they're the same beam.

And it reads both directions. Every member the calcs size, checked against how that member is drawn — and every member on the drawings, checked against how it was sized. A beam that's drawn but never shows up in the math gets caught. So does a beam that's in the math but drawn another way.

Take one beam. Treat the numbers below as an illustration of the kind of mismatch the review catches — not a promised result, and not a standing accuracy claim.

ONE MARK, TWO BOOKS   (illustrative)

  FB16

  On the sheet ...... W10×68, one 30-ft span
  In the calcs ...... W10×45, two spans (10 ft + 20 ft)

  Same mark. Two different beams.
  Flagged — with the page in each book.

On the drawings, FB16 is one beam carrying a single long span. In the calculations, the same mark is a lighter section broken over two shorter spans with a support somewhere in the middle. Both can't be the building. And on a pass that happened to sample three other intersections, everything checked out — because the member that disagreed wasn't one of the three.

That's the quiet danger of sampling: it only ever finds the mismatch it happens to land on.

Following the column down

A member read in isolation isn't the whole story, so the review doesn't read it that way. It holds every structural sheet at once and follows the load the way the building does — a beam hands its reaction to the column beneath it, the column carries it down, and the footing or the pile at the bottom passes it into the ground. When a beam drops its load onto the member that supports it, the review follows it there and checks that member too.

A single beam can check out and the building still be wrong — because the thing holding it up doesn't. So the review follows the load all the way down.

To do that it reads the sheets at full resolution — nothing shrunk to fit — so a pile number stays a pile number and a half-inch mark stays legible. And when a sheet is simply too unclear to read for scale, the review reports that check as incomplete rather than guessing at it.

Round one, not round two

None of this decides a permit. The local jurisdiction does, and a finding is a flag to verify at the counter, not a ruling. What changes is when the mismatch turns up.

A beam drawn one way and sized another is the kind of thing that normally surfaces in round two — in a correction letter, after the city's plan checker is the one who finally lined the two books up. By then it's weeks of calendar and another lap through the queue.

Caught in the pre-check, it's a fix before anything is submitted — the drawings and the calcs reconciled while the set is still on the engineer's desk, both books, at last, describing the same building.

CrossBeam Stories

CrossBeam reviews permit plan sets against the current California code — every citation, every cycle. For the running log of what changes mid-cycle, read Code Watch.