Chapter 35 Quiz

B. LOD describes reliance, not appearance. A manufacturer's model with every louver and bolt head can be worthless if it is not the unit that was bought; a grey box at the exact outside dimension with its exact service clearances at its exact coordinates can be entirely reliable. Detail is a property of the picture. Level of development is a property of the promise.

C. A partition modeled for coordination carries correct thickness and location because that is what coordination needs. It frequently does not carry layer count, run-to-deck versus run-to-above-ceiling, stud gauge, or rating. The model reports wall surface area; the estimate needs board area. On Northgate that difference was 17.8 percent and $306,700.

C. The crane and the panels never occupy the same space at the same time, so there is no geometry to test. Note that B is findable — but only if somebody modeled the clearance zone and wrote the rule that tests for intrusion into it. That is the difference between a soft clash and a 4D clash, and most coordination programs are quietly mediocre at the first one.

B. One 60-foot duct main crossing one sprinkler main reports as fourteen element-pair clashes and is one decision. A coordinator who reports raw counts is either inexperienced or hiding. Report the filtered count, the funnel that produced it, and the rules — which belong in the execution plan with a signature on them.

C. The convention allocates space in order of decreasing constraint. The system with the fewest options claims its space first, because it is the one least able to recover from a bad decision. You cannot negotiate with gravity.

C. Not one of the four differences is a modeling error. Every partition is exactly where and as thick as it should be. What failed is the assumption that a model built to answer "does it fit?" can also answer "how much board do I buy?" A precise wrong number is more dangerous than a rough right one, because nobody checks it.

D. 4D finds spatial and sequence conflicts that CPM logic cannot express, and nothing else. It will also animate a schedule that is complete fiction, smoothly and confidently. Garbage in, cinematic garbage out.

B. Fifty-seven percent of the program is money paid to trade contractors to model. Only $46,000 of the $643,500 sits on a line that says anything about BIM — which is exactly why people believe it is free, and exactly why it fails on projects where nobody added it up.

C. It is worse than no model, because it looks like compliance. The coordination never happened; you bought an expensive record of a process that did not occur. The other two responses at least tell you the truth in month one.

B. Not because rung 5 is fake — hospitals, campuses, airports, and data centers do run real twins — but because rung 5 requires a facilities organization staffed and funded to consume it. Selling a live twin to an owner with four facilities staff and a work-order spreadsheet is selling a monument.

C. And the reason it matters is that "but the mechanical drawing shows it here" stops being an argument in February. The drawing showed a single line in plan. The model showed a 46-inch insulated envelope at a specific elevation, and eight companies signed it.

False. It can be worthless for coordination if it is not the model number that was actually purchased. Detail is not development.

False. Three different artifacts, made by different parties, for different purposes, with different legal responsibility. The design model is a wish, the construction model is a plan, the fabrication model is an instruction.

False. A machine cannot test for intrusion into a clearance that nobody modeled. Somebody has to model the access zone and write the rule — which is why an experienced coordinator is worth what they cost.

True. Program cost has a floor and scales with parties and congestion; return scales with the rework available to avoid. On a fit-out with an existing grid and two systems there is not much. What does pay: a scan above the ceiling, coordination of the two or three congested spots, layout if you already own the instrument, and a 4D for the one hard week.

True. Northgate's extraction returned 984.6 tons against a canonical 985 — because at LOD 400 the model is the fabrication instruction. There is no gap between the representation and the thing. That is what makes this the one place 5D works cleanly.

False. Every activity had a defensible duration and a defensible predecessor. Every document in the standard set was correct. There is simply no CPM relationship that can express "the machine will not fit through the hole" — that sentence lives in three dimensions plus time.

You may not extract a quantity from, or coordinate a clearance from, an element whose level of development does not support it.

Example: AHU-5 in Case Study 1. The LOD matrix required owner equipment at 400 with vendor data; the vendor package was nine weeks late; the LOD 300 placeholder survived sign-off carrying no coil-pull envelope. Coordination ran clean because there was no zone to test against. Cost: $31,200, found at startup when somebody tried to pull a coil.

Correct fix: a stated coordinate system, a project base point with real-world coordinates, a survey point, stated units, project north versus true north, and a small shared origin-test file — three cubes at known coordinates — that every party inserts and confirms before modeling anything else. About an hour, if written in week one.

Dangerous fix: a detailer who drags their model into place by hand. It now looks right, and every coordinate it will ever produce — every layout point, every sleeve, every hanger insert — is wrong by the amount of the drag. Detect it by checking the model against the shared origin-test file, not by looking at it.

Because the ratio hides where the money came from. One line — the crane trap, $261,250 — is 30 percent of the gross savings and 41 percent of the entire program cost, and it came from a single afternoon in June. Two lines are 51 percent. BIM's return is not a steady trickle of small efficiencies; it is a small, certain, budgeted cost that buys down a low-probability, high-consequence risk. You evaluate that the way you evaluate an insurance premium — by what the tail looks like, not by the average.

Coordinating the model resolves conflicts among what the parties intend to build. Verifying the building confirms that what actually exists matches it. The level-three multi-trade rack proves the gap: the slab deflected about 1¼ inches high over forty feet under wet lightweight topping, between scan three and scan four. The rack had already released against scan three. It was built correctly, to a model that was correct, for a building that had moved — $18,400 and four days. A verification program has a frequency, and things happen between scans.

1. Which rung? Do you want an as-built model, an as-built model with structured asset data, or that data imported into the maintenance-management system you actually use? (§35.8 rungs 2, 3, 4.)
2. Which data fields, for which assets? Give me the list — tag, type, location, manufacturer, model, serial, install and startup dates, warranty dates and contact, service intervals, spares and filter sizes, responsible subcontractor, document links. Or tell me your system's import format and I will build the list from it.
3. Who consumes it, and are they staffed to? If the answer is "nobody yet," you are buying a deliverable, not a capability — and you should buy the smaller one.

The follow-on is the one that saves the job: whatever the answers, capture the data as it arrives — at the submittal, verified at the nameplate, on the daily report. Reconstructing it in the last six weeks from equipment already above a hard ceiling is how you deliver a spreadsheet with 40 percent of the serial numbers wrong.

This week. Three things, none of which requires anyone's permission. (1) Publish a coordinate origin and an origin-test file, and require every party to confirm alignment in writing before the next exchange — one hour, and it is the prerequisite for everything else. (2) Get in writing, by RFI, what level of development the design models are represented to carry, by system, and whether they are represented as suitable for construction coordination. (3) Freeze fabrication release on every area not yet released, and put that in writing to each subcontractor the same day.

This month. (1) Rewrite the execution plan around the four sections that actually decide anything — the LOD matrix, the origin, the exchange schedule with dates, and sign-off authority with a stated obligation. (2) Price a detailing allowance for each MEP trade and take it to the owner as a change or to your own executive as a cost, because you will not get modeling you do not pay for. (3) Scope coordination to the congested areas only — you no longer have the calendar for the whole building, so spend it where the systems concentrate. (4) Get the governing-document sentence into the subcontracts by amendment.

What you can no longer achieve. Level one is gone. Sheet metal is cut; the model cannot change what is already fabricated, and pretending otherwise wastes the one thing you still have. Say so plainly: "Level one will be coordinated in the field, and I expect field rework there. Levels two and three I can still protect, and here is what it costs." An owner who hears that in month 6 can plan. An owner who hears it in month 11 has been managed badly.

(a) The conflict. Roofing "around the curbs" is scheduled in week 37 and the units land in week 38. You cannot complete the membrane around a curb before the unit is on it, and rigging four rooftop units across finished membrane is how you buy a roofing repair and a warranty argument. There is also a second-order conflict: the crane pick requires the gym roof area clear, and in week 37 the roofing crew is standing on it.

No clash test finds this because the geometry never intersects. The membrane and the units are never in the same place at the same time; the RTU sits on the curb in the finished building and will clash-test clean forever. What conflicts is access, work area, and sequence — which exist only in time.

(b) The arithmetic. The window is 9:15 a.m. to 2:30 p.m. = 315 minutes. Setup and teardown consume 90 + 90 = 180 minutes, leaving 135 minutes of picking. At 55 minutes per pick that is two picks per day, with 25 minutes of spare — not three. Four units therefore take two crane days, not one, and both must fall on days when school is in its normal schedule. Check the school calendar before you pick the dates; an early-dismissal day shortens the window and costs you a pick.

(c) The corrected sequence. Weeks 34–36 unchanged. Week 37: set RTU-1 through RTU-4 across two crane days. Week 38: complete the roofing membrane, including the flashing around the curbs. The procurement fact to confirm first is the RTU delivery date — you have just pulled the set a week earlier, and if the units are not on site in week 37 you have made the schedule worse, not better. Get the delivery confirmed in writing from the manufacturer before you re-logic anything.