Tutorial 27: Monte-Carlo Pick-Sequence Analysis

Warning

Crane stations and the slab edge — read this before copying any station shown here.

A crane station must sit entirely outside the wall-clearance standoff, on prepared subgrade, or entirely inboard of the closure strip, on the poured slab. Never between the two, and never spanning the boundary. The test is the machine’s bearing footprint — its tracks or outriggers — not the centre of a drawn circle. One track on poured slab and one on unpoured subgrade gives differential support under the mat; the slab edge spalls or the subgrade settles, and the machine goes out of level under load.

TCA Safety Guideline v13.1 (June 2013) §2 item 6, p.8: “Install entrance and exit ramps for the crane to position itself onto the floor slab. Do not allow the crane to exert its weight on the extreme edge of any portion of the slab.” See also OSHA 29 CFR 1926.1402 — ground conditions; the operator must be notified of bearing capacity, and the controlling entity is responsible for adequate site preparation.

Station placement in this release is a known open defect (CV-2714). The perpendicular standoff is a fixed value that every catalogued machine’s bearing footprint exceeds, so a generated station can straddle the boundary. Do not copy a station position from this lesson, or from its video, onto a real site. Have the setup reviewed against the machine’s bearing footprint and the site’s own closure-strip and brace-line dimensions.

And on this release the drawn frame can be mirrored, which makes an inside/outside verdict untrustworthy whenever it happens. It was measured in coordinates on a saved post-draw drawing rather than inferred: after Site Drawing Options is entered twice, wall-line ends land at the exact Y-mirror of their shipped positions – same X to the hundredth, Y negated – 622 feet apart at the widest, while a point at the origin separates by zero, which is what distinguishes a mirror from a constant shift. Two points were left unmoved, so this is a partial redraw and not a blanket transform of every coordinate. The no-flip control has since been run, and the frame flip IS the cause. An otherwise identical take with no Site Drawing Options entries drew ZERO mirrored points, with all four shipped points unchanged, while the flipped take mirrored two of them. The control was not a null run – it drew ten points against the shipped four, and left fourteen wall-line ends open against the shipped two – so the zero means something. The mirror is therefore not present on an untouched drawing – but the action that brings it on is smaller than it sounds: opening Site Drawing Options twice and then drawing is enough. Measured on the dialog itself, BOTH openings were closed with Esc – no OK, no edit, nothing typed. Backing out of the dialog still sets it. That is still ordinary use and still a defect, but it is narrower than the previous wording implied. The discriminator is clean and the sample is small – four shipped points, on staged copies rather than customer files. An earlier version of this note said the draw refused and therefore nothing reached the sheet; part of it drew, and what drew is mirrored.) So a station reported as clear of the closure strip may not be. Treat every crane position here as illustrative of the workflow only, never as a sited answer.

Scope, measured 2026-09-16: the footage on this page shows cv-web Cast-mode capacity rings in the casting yard, not stations placed on a site plan – 28 sampled frames across the four crane videos, none showing a slab edge or closure strip. The placement defect above is in the site-plan sequencer, so it is a caution about the workflow you are learning, not a statement about a picture on this page.

Let the numbers pick the order — simulate hundreds of crane pick sequences on your panel set in cv-web, compare the construction strategies against a random baseline, and read which lift ordering is statistically the most robust.

The core curriculum drew a panel and stood its set on the wall lines; Tutorial 16 carried that project into cv-web in the browser. This advanced lesson uses one of cv-web’s preview capabilities: on a casting layout, ConstructiVision can run a Monte-Carlo pick-sequence analysis — it plays out the crane’s day across many simulated pick orders, scores each against real construction-planning metrics, and surfaces the ordering that holds up best. It is a planning aid, not a rigging plan: it ranks orderings, it does not certify a lift.

Note

Preview / pilot-gated feature — U.S. Patent Pending. Monte-Carlo pick-sequence analysis is a preview capability in cv-web: it is working software, but it is gated behind pilot enrollment or an entitlement and requires an active membership (it is one of cv-web’s AI/analysis features). Access, exact controls, and labels are still being finalized, so this lesson stays conceptual where the UI is not yet documented, and describes what the analysis produces rather than pixel-exact steps. The pick-sequence and crane pick/brace simulation work is U.S. Patent Pending — described here as “Patent Pending” only.

Note

This lesson also exists in the browser. Work it in ConstructiVision on the web — same steps, same values, so you can follow either and end up with the same panel.

What you will build

From a tilt-up project already carried into cv-web (Tutorial 16), on its casting layout:

  • a completed Monte-Carlo run over your panel set — hundreds of simulated pick sequences plus a handful of named construction strategies (perimeter, nearest-neighbor, heaviest-first, lightest-first, mark-sequential);

  • a strategy-vs-metric comparison table — arm rotation, arm extension, crane moves, rig changes, max radius, total weight, critical picks, and total swing distance;

  • the random-shuffle baseline (the Monte-Carlo median with its spread) that every named strategy is judged against;

  • a recommended ordering — the strategy that scores best on the metrics you choose to weigh — ready to apply back to the layout’s pick order.

Before you start

  • This lesson stands on the finished project thread carried into the browser: complete Tutorial 16: From CAD to cv-web first so your project (its panels and their weights, pick points, and brace points) is available in cv-web. The casting-layout idea itself is Tutorial 12 in AutoCAD (Tutorial 12: Construction Layout) — here the same set is laid out in the browser’s cast view.

  • You need a signed-in SimpleStruct account with an active membership — the analysis features are membership-gated. If your membership has lapsed they lock until you reactivate; see Membership & Tokens.

  • You need pilot access to the preview. If the pick-sequence analysis is not visible in your account, it is behind the pilot/entitlement gate — request enrollment; preview access does not spend tokens.

  • Have the project open as a cloud project in the cv-web editor (Work with cloud projects), with its panel set placed on a casting layout so there is a pick order to simulate.

Tip

The analysis reads what the panels already carry from CAD — each panel’s weight, its pick points, and its brace points. A set that skipped those (Tutorial 8’s hardware, Tutorial 9’s pick and brace points) gives the simulation less to reason about. The richer the panel data, the more the ordering it recommends is worth.

Lesson

Step 1 — Open the casting layout in cv-web

In the cv-web editor, open your cloud project and switch to its casting (cast) layout — the browser view that lays the panel set down flat in the yard, in the order the crane will pick them. This is the web counterpart of the construction layout you built in Tutorial 12: Construction Layout: a pick order you can see and rearrange. The crane sits at a set position with an exclusion radius around its body, and the layout can show capacity rings — the geometry the simulation reasons over.

Step 2 — Set the crane and the pick order you want to test

Before simulating, confirm the layout reflects the job:

  1. Place the crane and set its exclusion radius so panels that would fall inside the crane’s own footprint are counted as requiring a move (re-mobilization), not a swing.

  2. Leave the current pick order as your starting point — the analysis will compare your order (and other strategies) against the field of random orderings.

Everything from here is read-only reasoning over the layout; the simulation does not alter your panels.

Step 3 — Run the Monte-Carlo pick-sequence analysis

Start the analysis from the cast layout’s simulation control. cv-web then does two things at once, entirely in your browser:

  • it evaluates a set of named construction strategies — walking the perimeter (your current order), always taking the nearest remaining panel, heaviest-first, lightest-first, and strict mark-sequential order; and

  • it runs the Monte-Carlo pass — hundreds of random pick orders (the preview computes on the order of five hundred iterations) to map the full range of outcomes a crew could stumble into.

For each ordering it walks the crane panel by panel and tallies the same construction-planning metrics. A short “running” state shows while the iterations compute; a one-layout run finishes quickly.

Step 4 — Read the comparison table

The result is a strategy × metric table. Each named strategy is a row; the columns are the metrics the simulation measured for that ordering:

  • Arm rotation — total crane swing, in radians, summed across the sequence (less is better).

  • Arm extension — total change in pick radius as the boom reaches in and out (less is better).

  • Crane moves — how many picks force a re-mobilization inside the exclusion zone (less is better).

  • Rig changes — how often the rigging must change between consecutive picks — pick count, brace count, or pick spread shifting (less is better).

  • Max radius — the farthest pick radius the sequence demands (less is better).

  • Total weight — the tonnage handled (context, not a rank you minimize).

  • Critical picks — picks whose panel weight approaches the crane’s rated capacity at that radius (less is better; this is a flag for the lift engineer to review, never a determination).

  • Swing distance — total travel of the loads, in feet (less is better).

A final, italicized row shows the Monte-Carlo random baseline — the median of all the random runs, written with a ± spread (its inter-quartile range). That row is the anchor: it tells you what an arbitrary order would typically cost, and how widely random outcomes scatter.

Step 5 — Choose what matters and read the recommendation

You decide which metrics to weigh. Toggle the metrics that matter for this job (arm rotation and rig changes are sensible defaults) and the table scores each named strategy by its combined rank across only those metrics — lowest combined score wins. The best-scoring strategy is highlighted and marked with a star. That is the analysis’s recommended ordering for the priorities you set; change the priorities and the recommendation can change with them.

When a recommended strategy is one you want to adopt, apply it — cv-web reorders the casting layout’s pick sequence to match, and the crane’s day now follows the ordering the simulation favored.

Check your result

There is no automated checker here — this result lives in the browser, over cloud data, and the “right answer” is a reading, not a single graded value. Judge the run against what a sound Monte-Carlo report should let you say:

  • The baseline is populated. The italic random-baseline row shows a median and a ± spread for the metrics. A spread that is essentially zero means the set is too small (or too uniform) for the ordering to matter much — a legitimate finding, not a failure.

  • A named strategy beats the random median. For the metrics you care about, at least one named strategy should sit at or below the random median. If every named strategy is worse than a coin-flip order, re-check the crane position and exclusion radius — the geometry, not the ordering, is likely driving the numbers.

  • The recommendation moves when you change priorities. Toggle a different metric on and off and confirm the starred strategy re-scores. A recommendation that never changes means only one metric is effectively in play.

  • Critical picks are read as prompts, not verdicts. Any non-zero critical picks count is a list for the lift engineer to review against the rigging plan — the analysis flags proximity to rated capacity; it does not clear or certify a pick.

Read the whole table the way Analyze panel sets frames its placements: these are decision-support reference points across simulated outcomes, not engineering determinations. The analysis narrows the field of orderings and shows the trade-offs; the construction engineer still signs the lift.

Watch this lesson

The video shows the Preview feature as it stands today; the pilot gate and patent notice above still apply.

Transcript (20 captions)

0:01 — Tutorial 27 — you can argue about the best construction order, or you can measure several hundred of them.

0:07 — Start from a sample building — a full multi-panel project, so there is something real to work with.

0:13 — ConstructiVision Sample Building — 60 panels. Nothing you do while looking around changes anyone's data.

0:19 — The cv-web editor with the project open.

0:22 — Cast mode — the casting layout this analysis runs against.

0:26 — Real beds, real panels, real weights — the analysis is only as good as the layout.

0:31 — Your current pick order is one of the candidates — the one you would have used.

0:38 — Set the crane and the order you actually intend. That is the entry the others get compared against.

0:44 — Start the analysis from the layout's simulation control.

0:47 — Two things happen at once: named strategies are evaluated — perimeter walk, nearest remaining, heaviest-first, lightest-first, mark-sequential — and a Monte-Carlo pass runs hundreds of random orders.

0:56 — It all computes in your browser. Nothing is uploaded.

1:00 — One row per strategy.

1:02 — The same metrics across every row — crane moves, rig changes, radius, weight.

1:08 — Including critical picks: the lifts with the least margin under each ordering.

1:13 — The Monte-Carlo row is the random baseline — median and spread rather than a single number, and no score, because 'pick at random' is not a strategy anyone would choose.

1:23 — It is there to answer a different question: how much does sequencing actually buy you over stumbling into an order?

1:30 — Score ranks the named strategies against each other.

1:33 — The starred row is the most robust of the set — best on the metrics that cost real money, not just the fastest on paper.

1:42 — Apply writes that ordering back onto the layout — the analysis becomes the plan.

1:48 — Read it, disagree with it, pick the one that matches your yard. The point is that the trade is now visible before the crane shows up.

Where next

  • The statistical framing in depth: Analyze panel sets — how cv-web presents distributions, medians, and “review” flags across a corpus, and why placements are reference points rather than determinations.

  • Getting a project into cv-web first: Tutorial 16: From CAD to cv-web — export the panel data from AutoCAD and carry the set into the browser.

  • The casting layout it simulates: Tutorial 12: Construction Layout — build and record the panel set’s casting positions in AutoCAD.

  • The cloud project this runs over: Work with cloud projects — team projects in the cv-web editor, and what membership unlocks: Membership & Tokens.