Murray State University · Part four · Example projects

Construction Scheduling

Murray State University
Demo 10 of 17 · Part four · Construction Management · Stage 1

Read the schedule

Where is this job’s float, and which chain has none?

Before you start. A network diagram is the job written as activities with predecessors and durations. Total float is how many working days an activity can slip before the whole project slips; the chain with none is the critical path. Nothing here is declared critical — it is computed, every week, from the live state of the job.

Predict

Before you open the Gantt: which chain do you think has zero float — and how much cushion does the job have against its contract?

Try

Type a seed, press Start the job, then read the Gantt and Network views before you commit anything.

Takeaway

Float is computed, not declared, and the chain with none is the one that sets the finish date.

Check yourself

An activity shows 9 days of total float and 1 day of free float. What may you safely spend?

This is MEP rough-in on Level 1 at baseline. Read both numbers before you answer.

Bryant Harrison · Murray State University
One HTML file · runs offline

Guidehow to use this demo

You are the general contractor and the clock is the contract. What was that decision worth on the Monday you made it?

Where this sits: Demo 10 of 17, Part four. Part four applies the same habit — predict, run it, let the result correct you — inside a real discipline. This is the scheduling case, and it needs network diagrams, not any of the AI demos.

  1. Read the schedule. Everyone enters the same seed code, so the weather and every roll are identical for every team. Press Start the job, then read the float column before deciding anything.
  2. Run a week. Predict which activity will be critical, then Commit the week. No undo, ever.
  3. Make a call. Answer what is on the table. Not answering is answering — every call defaults to its free, most exposed option.
  4. Debrief. Grade the decision, not the outcome.

Three ways to look at the same job

4D cutaway — a section through the building driven by real progress. The finished job is drawn as a faint dashed ghost and work fills it in; crews stand on the floor they are working, so a stacked floor looks crowded and turns red. Gantt — bars from early start to early finish, a thin tail out to late finish for float, critical work in red, cure and inspection holds hatched, today and the contract date both marked. Network — the precedence diagram, with relationship type and lag on every arrow and the driving path thick and red. It is computed every week, never declared, so it moves and the game says so when it does.

Three things that cost teams the game

Getting around

The stage tabs switch stages; stages 2–4 open once there is a schedule to run. Details opens the evidence and limits for the current stage. Guide reopens this card; Settings has Open Presenter Notes, Presentation mode and Reset. On a phone the job folds into SITE and MEETING tabs. Tap any underlined term for what it means.

Time

A full session is 75 minutes: 11 minutes of tutorial, 33 of competitive run, 12 of debrief. The tutorial alone is about 10 minutes. The printable session guide has the running order; the CPM appendix has the background.

Settings

Presenter notes

The session guide on screen — the same document as the printable sheet and the PDF, not a summary of it. Running order, the four stages, the five misconceptions and the presenter controls.

Presentation mode

Projector mode: large UI, the schedule and the computed critical path readable from the back row, and a Notes button in the header. Turn it on for THE NEWS each week so the whole room reads the same card, and for the leaderboard at minute 59.

Reset

Back to the setup sheet in place, with no page reload: no run in progress, seed and team cleared, every prediction echo and check answer cleared, stages 2–4 re-locked, stage 1 selected. Presentation mode and the leaderboard are not demo state and survive it, the same way they survive a reload. Mid-run it asks first.

Printable guide and appendix

The session guide as a printable sheet, with a PDF beside it; and the CPM appendix — the derivations, the full calls deck, every debrief question, the measured seeds, the sources and the vocabulary.

Printable session guide CPM appendix

Engine self-test

36 assertions run live — the tutorial CPM against the hand calculation, determinism, the non-compressible holds, the weather boundary in both directions, and the two checks that hold the calls fair. If a student says the software is wrong, run this in front of the class.

Leaderboard

Scores are held on this machine and survive a reload and a Reset. Clear them between sections so a second section starts on an empty board.

DetailsRead the schedule

What is live

What the page actually computes

Live the schedule, the float column and the critical path

A precedence-diagram CPM is solved from src/data.js every turn. Nothing is marked critical in the data file; criticality falls out of the arithmetic below.

Illustrativea plausible small-commercial pro forma, not a real project’s numbers the contract, the money and the durations

Racer Commons is invented: 38 activities, a $4,450,000 contract, 134 working days. The figures are chosen so that schedule rather than cost-cutting dominates the score.

How it works

Forward: ES = max over predecessors (FS: EF+lag · SS: ES+lag · FF: EF+lag−duration)  |  EF = ES + duration
Backward: LF = min over successors (FS: LS−lag · SS: LS−lag+duration · FF: LF−lag)  |  LS = LF − duration
Total float = LS − ES   (shared along the chain)
Free float = earliest successor ES − your EF   (yours alone; always ≤ total float)

Walk the network in topological order for the forward pass and in reverse for the backward pass. The largest early finish is the project duration. Total float and free float are different quantities, and students collapse them into “spare time” at their own cost: at baseline MEP rough-in on Level 1 carries 9 days of total float and 1 of free float.

Assumptions and simplifications

  • Float is measured to the project’s own computed finish, so it is never negative. A job twelve days behind shows no −12; performance against the contract is reported separately in calendar days. Students who have seen negative float in P6 need this said out loud.
  • Days are numbered from 0 and finish is exclusive, so EF = ES + duration. Textbook tables numbering days 1–n (start day inclusive, EF = ES + duration − 1) read one higher on ES and LS; EF and LF come out identical. A three-day activity at the start is ES 0 / EF 3 here and ES 1 / EF 3 there.
  • Cure and inspection holds are modelled as activities rather than lags, so you can see them sitting on the critical path.
  • Durations are crew-days of work, not fixed calendar durations: they shorten with crews and stretch when a trade is starved.

Sources

  • The CPM itself is standard precedence-diagramming method, not a cited result. The engine is checked against a hand calculation instead — all sixty numbers of the ten-activity tutorial network, printed in the CPM appendix §1.
  • Appendix §8 is the printable vocabulary; the hover glossary in the game is the authoritative copy.
Boundary: this is a teaching schedule, not a scheduling tool. It does no resource levelling, no cost loading, no per-activity calendars and no holiday calendars, and it will not tell you anything about a real project.

Presenter Notesthe printable session guide, on screen

Session Guide Construction Scheduling
Topping Out — a week-by-week scheduling game — Racer Commons, 3-story commercial, 38 activities
Sheet IG-01
Demo10 of 17 · Part four
Course levelFirst-year CM
Session75 minutes
Teams3–4 students
PrereqNetwork diagrams

Every team schedules the same building, week by week, from the same seed code. The seed fixes the project, the sequence of events and the roll behind every judgment call before anyone touches anything, so two teams that finish $200,000 apart did not get different weather and did not get different luck — they bought different exposure. That is the whole pedagogical trick. You need: one laptop per team (phones work), a projector, and a seed code on the board.

This sheet is the session. The CPM derivations, the full calls deck, all eleven debrief questions, the seed table, the overtime literature and the vocabulary live in the CPM Appendix — linked from the Guide and from Settings, and not reproduced here. Nothing was dropped in the split; it moved.

1 Learning goals

Not stated objectives — what the activity itself demonstrably shows. One per stage, and these are the four Takeaway cards the learner reads on screen, verbatim:

  1. Float is computed, not declared, and the chain with none is the one that sets the finish date.
  2. The critical path is re-solved every Monday against what actually happened, so it moves under you.
  3. A call is a priced bet with its odds printed on it; every team faces the same roll, so what differs is the exposure you bought.
  4. A decision is graded on what it was worth the Monday you made it, not on how the roll landed.

2 Where this sits on the path

Demo 10 of 17, Part four — example projects. Part four shows the same habit of mind applied in a real discipline: state what you expect, run the thing, and let the result correct you. This one is the scheduling case, and the prerequisite it leans on is the network diagram — activities, predecessors, durations. It does not need any of the AI demos.

3 One question for the session

What was that decision worth on the Monday you made it?

“Nobody in this room is going to be graded on whether the roll went your way. You are going to be graded on what you knew, what it cost, and where the float was when you decided.”

4 Seventy-five minute running order

The tutorial project, Shoe Tree Annex, is 10 activities, runs 8–10 turns and carries three calls of its own. Do not skip it. Teams that meet the interface on the graded run spend six weeks clicking instead of thinking, and the debrief gets thin.

Running order — 75 minutes
MinutesLengthSegment
0–55 minSetup (Stage 1). Seed code and difficulty on the board; teams enter both and name themselves. Presentation mode up front. Confirm every team reads the same project start date, Mon 7 Sep 2026, before anyone commits a week.
5–116 minCPM at the board. Forward pass, backward pass, total float, on the ten-activity tutorial network — the hand calculation is in the Appendix, §1. Land one sentence: float is not slack in your day, it is what a specific activity can slip before it moves the finish. Leave free float for the debrief.
11–2211 minTutorial run — 8–10 turns. Walk the room. Push teams to read the float column before deciding anything, and make sure every team answers the subgrade call instead of clicking past it.
22–264 minBrief the competition. Same seed, same difficulty, score is profit, no undo. Say two things out loud: calls default to the free option, so not answering is answering; and crews take two weeks, so manpower is a forecast.
26–5933 minCompetitive run (Stages 2–3). Twenty-four to thirty turns. Roughly half the weeks are one click; the rest carry a call, a weather decision or a crew order. Checkpoint at minute 44 — teams should be near week 12, and anyone behind should commit the quiet weeks fast.
59–634 minLeaderboard and the do-nothing baseline. Show the spread, then what absorbing everything would have scored on this seed. Worst team against baseline is usually more interesting than first against second.
63–7512 minDebrief (Stage 4). Open on the calls table; it is the fastest route into the room. Then the five misconceptions below and two or three questions from Appendix §5. Finish on the week the critical path moved and ask who noticed at the time.
If you are short on time

Call time at minute 59 wherever teams are and score projected profit — the CPM finish date priced out with general conditions, liquidated damages and the early bonus. It is a legitimate score and it ranks teams the same way. Alternatively, assign the tutorial as ten minutes of homework and give the whole 44 minutes to Racer Commons. Do not buy time by cutting the debrief: the calls table is where the session pays off.

5 Stage by stage

Stage 1 — Read the schedule

Question. Where is this job's float, and which chain has none?
Predict prompt. Before they start the job: which chain has zero float? The prediction is captured on screen and echoed beside the computed answer.
Activity. Seed and difficulty in, Start the job, then read the Gantt and Network views before committing anything. 4–5 minutes.
Expected observation. Unimpeded, the network runs 89 working days against a 134-day contract — 45 working days, 63 calendar days, of cushion. One chain carries zero total float and everything else has some.
Answer key. The float column, not the bar length, identifies the critical chain. A bar can be long and have float; a two-day cure can be short and have none.
Teaching limit. Float here is measured to the project's own computed finish, so it is never negative. Students who have seen negative float in P6 need that said explicitly.

Stage 2 — Run a week

Question. What does one week actually change?
Predict prompt. Which activity is critical after this week? — asked before Commit and captured, because the engine re-solves the network afterwards and will answer it. This is the demo's signature prediction; make the room commit out loud too.
Activity. Read THE NEWS, allocate crews, Commit the week, watch five working days run in 4D. 2 minutes for the first one, seconds thereafter.
Expected observation. The prediction appears beside the computed result. Sooner or later they disagree, and the game announces the move and names the activity it moved to.
Answer key. Completed work is pinned to the dates it actually happened, work in progress is anchored to its real start, and nothing not yet started can begin before today. Criticality falls out of that.
Teaching limit. One turn is one week of five working days; the engine has no clock, no network call and no unseeded randomness.

Stage 3 — Make a call

Question. What are you buying when you answer a call?
Predict prompt. Before choosing: what do you think the free option costs you, in dollars?
Activity. Answer the call on the table — each option shows its price today and its exposure, the probability the deferred consequence fires — then Commit. 16 calls in Racer Commons, 3 in the tutorial.
Expected observation. Nothing is hidden: the card prints the exposure as a percentage and names the activity where the consequence will land. Buying exposure down is buying a smaller bar for the same roll.
Answer key. Every call defaults to its first option, which is always free and always the most exposed. Clicking Commit through the job takes the 45%, the 40% and all the rest without deciding to.
Teaching limit. Exposure is a stated probability against a seeded roll, not a model of real construction risk. One option is deliberately ⚠ UNSAFE and is not a trade-off — see the box below.

Stage 4 — Debrief

Question. Was that a good decision, or a good outcome?
Predict prompt. Before opening it: will your calls table read mostly “covered” or mostly “got away with it”?
Activity. Open the Debrief. One row per call: Week / Call / You chose / Cost / Exposure / The roll / Outcome. 12 minutes with the room.
Expected observation. The verdicts are a two-by-two, and the diagonal is the lesson: got away with it is where the luckiest team lives, and unlucky: you covered it and it still bit is where a team that did everything right can be sitting.
Answer key. Same seed, same roll. Put a team that left a call at 45% beside a team that bought it down to 3% and price both on the Monday they decided.
Teaching limit. Profit is the score, but the cash curve is the second set of books — a team can project a healthy profit while sitting $1,000,000 into the bank's money.

The one call that is not a trade-off

Around week six, backfilling against green unbraced foundation walls is offered three ways. Waiting costs three days; engineered kickers buy them back for $6,400; and backfilling high against one side now is free, saves the three days and is marked ⚠ UNSAFE, with the reason printed before anybody clicks. An unbraced, uncured wall loaded on one side only is an OSHA excavation and structural-stability violation before it is a schedule decision. It is in the deck deliberately, and if a team takes it the debrief opens with a red block above their profit — whether or not the wall stood. If the roll went their way it says so: and it held. That is the most dangerous possible outcome, because it teaches you it was fine. Read the block, then the number, then ask whether they would sign that decision with their own name on the citation.

6 Five misconceptions, and the moment that exposes each

These are the on-screen refutation bank too: each stage's Check yourself card offers the flawed model as a real option and names why it is wrong. State the misconception before the correction — a correction alone does not displace a model the student already holds.

(a) “Float is free time you can spend casually.” — Stage 1 Total float belongs to the whole chain and is consumed by whoever spends it first. MEP1 carries 9 days of total float but only 1 day of free float: slip it three days and the finish has not moved, but Level 2 rough-in cannot start when it wanted to. Then the switchgear card takes two days and the failed rough-in inspection takes three, on a branch that had four.
(b) “Overtime always accelerates.” — Stage 2 It genuinely does at first — a 50-hour week delivers 1.25× and a 60-hour week 1.50× in week one, and that is real acceleration, not a concession. But the cost premium applies in full from week one and never decays, while efficiency does: by week 5 a 50-hour crew produces 0.938 of a straight-time week at 1.375× the labour cost. Overtime is a burst tool for a defined push. And it does nothing at all to a cure or a hold.
(c) “The critical path is fixed at baseline.” — Stage 2 It is recomputed every Monday. ROOF sits off the path at baseline with six days of float; two rain days plus the membrane-backorder card and it is critical — at which point the glazing card that arrives later matters twice as much. The debrief question is not whether they saw the announcement, it is whether anyone changed their allocation the week before.
(d) “Weather delays everything equally.” — Stage 3 Rain days stop only weather-flagged work: earthwork, concrete placement, roofing and envelope. The same “ground is soup” card costs three critical days in week 3 and exactly nothing in week 18 with the building dried in. A cure keeps curing through a washout. The forecast is not the decision — the float column is.
(e) “A good outcome means it was a good call.” — Stage 4 The gambler's fallacy in a hard hat, and the most expensive habit on this list. A team pours to plan at column line 4, the roll comes in at 0.61, the inspector takes the pads, and they leave convinced they read the geotech. They took a 45% chance of a three-day hold and $8,500 — and the roll every other team was handed landed above the line.

7 Presenter controls

Settings (⚙ in the header) holds Open Presenter Notes, Presentation mode and Reset, in that order, then this demo's own controls: the Guide, the printable guide and the CPM Appendix, the engine self-test, and clearing the leaderboard.

Presentation mode — projector mode — scales the page up and puts a Notes button in the header. Turn it on for THE NEWS each week so the whole room reads the same card, and for the leaderboard and calls tables at minute 59. It survives Reset deliberately; a presenter resets between rooms.

Reset puts the job back to the setup sheet in place, with no page reload: no run in progress, seed and team cleared, every prediction echo and check answer cleared, stage 1 selected. Presentation mode and the leaderboard are not demo state and survive it. Mid-run it asks first, because a reset throws away a class's work.

Engine self-test runs 36 assertions live — the tutorial CPM against the hand calculation, determinism, the non-compressible holds, the weather boundary in both directions, and the two checks that hold the calls fair. If a student says the software is wrong, run it in front of them; it is a surprisingly good two minutes of teaching.

If a score is disputed or a browser dies. Same seed plus same decisions produces the same profit to the dollar, and that is one of the 36. Replay the seed with the team's decision log.

Construction Scheduling — Session Guide Sheet IG-01 Racer Commons — 38 activities — 134 working days

Open the printable copy (teacher-guide.html) — the same document, laid out for print; the PDF beside it is a render of that file. The CPM appendix (appendix.html) is the background sheet: derivations, the full calls deck, every debrief question, the measured seeds, the sources and the vocabulary.

Debrief

Engine self-testCPM vs hand calculation · determinism · model bounds

Live every assertion below runs in your browser when you press the button — nothing here is a stored result
Bryant Harrison · Murray State University