CPM Scheduling on the PE Construction Exam: Critical Path, Float & Crashing

Scheduling questions are the closest thing the PE Construction exam has to guaranteed points. The algorithms never change — a forward pass works the same way on every network NCEES can draw — and there's no judgment call, no code lookup, no ambiguity in the answer. Yet candidates routinely drop these questions, not because the method is hard, but because they're slow at it: a 10-activity network run hesitantly eats twelve minutes, twice the per-question budget for the whole exam.

The fix is mechanical fluency. Critical path method (CPM) questions reward the candidate who can run the forward and backward passes without stopping to re-derive the rules, who knows the difference between total float and free float cold, and who can rank crash candidates by cost slope in one read of the table.

This guide covers the full scheduling scope on the Construction exam — network analysis, float, the critical path, and time–cost trade-off — with a complete worked network you can use as a pacing benchmark.

Why scheduling matters on the Construction exam

The NCEES specification allocates 7–11 questions to Project Planning and Scheduling — up to 14% of the exam — spanning construction sequencing, activity time analysis, network analysis and CPM, resource scheduling and leveling, and time–cost trade-off. Only two knowledge areas on the Construction spec are larger.

What makes this area unusual is the points-per-hour-of-study ratio. Unlike formwork or soil mechanics questions, scheduling problems rarely require a reference lookup — they're procedure questions you run from skill. That cuts both ways: there's nothing to look up when you're stuck, so partial knowledge converts to zero points. The candidates who treat CPM as a drilled procedure, like long division, reliably bank this entire domain. The full domain-by-domain breakdown of where the other 69–73 questions come from is in our complete guide to the PE Civil Construction exam.

Core concepts you must master

Activity-on-node networks

Modern exam problems use activity-on-node (AON) diagrams: each activity is a box holding its duration, and arrows show logic — an arrow from A to C means C cannot start until A finishes (finish-to-start, the default relationship). Read the network before computing anything: find the start activities (no predecessors), the end activities (no successors), and any activity with multiple predecessors, because those merge points are where candidates make errors. Some questions present the logic as a predecessor table instead of a drawing; sketching the network from the table should take under a minute.

The forward pass: earliest start and earliest finish

The forward pass computes how early each activity can happen. For each activity, the early start (ES) is the latest early finish among all its predecessors — every predecessor must finish before it begins — and the early finish (EF) is:

The project duration is the largest EF in the network. Use the day-0 convention (first activity starts at time 0) and apply it consistently — mixing conventions mid-network is a classic self-inflicted error.

The backward pass: latest start and latest finish

The backward pass computes how late each activity can happen without delaying the project. Set the late finish (LF) of the final activity equal to the project duration, then work backwards: each activity's LF is the smallest late start among its successors, and:

Forward pass takes the max at merges; backward pass takes the min at bursts. Candidates who can state why — early dates are constrained by the slowest predecessor, late dates by the most urgent successor — never mix the two up under pressure.

Total float vs free float

This distinction is the most-tested concept in the domain. Total float (TF) is how long an activity can slip without delaying the project:

Free float (FF) is how long it can slip without delaying any immediate successor's early start: FF = min(ES of successors) − EF. Free float is never larger than total float. An activity can have several days of total float and zero free float — meaning any slip immediately pushes its successor, even though the project end date survives. When a question asks "how long can activity X be delayed without affecting activity Y," it's asking about free float; "without delaying the project" is total float. Read the stem twice.

The critical path

The critical path is the longest path through the network — and equivalently, the chain of activities with zero total float. Its length is the project duration. Two facts the exam likes to probe: a network can have more than one critical path (after crashing, it usually does), and a near-critical path one day shorter than the critical path limits how far crashing one path can shorten the project before the other takes over.

Schedule crashing and the cost slope

Crashing buys schedule with money: each activity has a normal duration and cost, and a crash duration and cost. The exchange rate is the cost slope:

Least-cost crashing follows three rules: only crashing a critical activity shortens the project; among critical activities, crash the one with the lowest cost slope first; and after every day removed, re-check whether a parallel path has become critical, because once two paths tie, you must shorten both to gain another day.

The four problem types you'll see

Type 1: Project duration and critical path

Given a network or predecessor table, find the duration and identify the critical path. Pure forward pass: the answer is the largest EF, and the critical path is traced through the zero-float chain. The full worked network below is this problem type — budget four minutes for a seven-activity network.

Type 2: Float of a flagged activity

"What is the total float of activity E?" requires both passes, but only along the paths touching E — on a sprawling network you don't need every box filled in, just E's ES (forward, through its predecessors) and LS (backward, from the project end through its successors). Distractors are built from the free float, from a single-convention error, and from the float of a neighboring activity.

Type 3: Cheapest activity to crash

Given a normal/crash table, shorten the project by one or two days at minimum cost. Compute cost slopes for critical activities only — the table will include tempting low-slope non-critical activities precisely to catch candidates who skip the network analysis.

Type 4: Effect of a delay

"Activity C is delayed 6 days; what happens to the project?" Compare the delay to C's total float: with TF = 4, the project slips 6 − 4 = 2 days. The follow-up trap is the successor question — if C's free float is 0, downstream activities start moving on day one of the delay even while the project end date is still protected.

A multi-concept worked problem

A project has seven activities. Durations in days, with predecessors: A = 3 (—), B = 5 (—), C = 4 (A), D = 6 (A, B), E = 2 (C), F = 5 (C, D), G = 3 (E, F). Find the project duration, the critical path, the total and free float of every activity, and the cheapest way to finish 2 days sooner given crash data for the critical activities.

Step 1 — Forward pass

Step 2 — Backward pass

Step 3 — Float and the critical path

Step 4 — Least-cost crash to 17 days

Common errors that cost points

Taking the min at a merge on the forward pass

At a merge point, ES is the largest predecessor EF — the activity waits for its slowest predecessor. Grabbing the smallest EF shortens the project duration and lands you exactly on a distractor. In the network above, taking min at D gives ES = 3 instead of 5 and a 17-day "duration" — which is answer choice (B) for a reason.

Confusing total float with free float

Activity C above has TF = 4 and FF = 0. A question asking "how many days can C slip without delaying activity E" is answered by free float (0), not total float (4). The exam writes both numbers into the answer choices every time.

Crashing a non-critical activity

Shortening activity E in the network above costs money and saves nothing — E has 7 days of float. The crash table will happily list a cheap slope for a non-critical activity; the cost slope ranking only applies to activities on the critical path.

Forgetting the parallel path after crashing

Every day removed from the critical path closes the gap to the next-longest path. The single most common crashing error is buying a day that the parallel path immediately takes back. Re-compute path lengths after each crash increment — in a four-path network this takes thirty seconds and saves the question.

Mixing day-0 and day-1 conventions

Whether the first activity starts at time 0 or on day 1 doesn't change durations or floats — but mixing the conventions mid-problem shifts every date by one and generates off-by-one distractors. Pick the convention the problem's diagram uses and stay in it.

How to study scheduling effectively

Phase 1: Drill the passes until automatic

Build or find ten small networks (five to eight activities) and run both passes on each, timed — the PEwise Construction course's scheduling module animates the procedure network by network if you want the mechanics demonstrated first. Target: a seven-activity network, both passes plus floats, in under four minutes. The procedure should feel like arithmetic, not problem-solving — that's the level the exam pays for.

Phase 2: Add float interpretation

For each practiced network, answer three questions: which activities can slip without moving the project, which can slip without moving any neighbor, and what happens if a specific activity is delayed by more than its float. This converts mechanical fluency into the conceptual reads the exam actually tests.

Phase 3: Crash scenarios

Work crashing problems with at least three paths, where the cheapest-slope activity is sometimes non-critical and where two paths tie partway through. Those two complications are the entire difficulty of exam crashing questions.

Phase 4: Mixed timed sets

Fold scheduling questions into mixed practice with estimating and earned-value problems — on the exam they arrive interleaved, and earned-value questions share the cost-and-time vocabulary, so drilling them together prevents formula bleed between the two domains.

Quick reference: key formulas and values

Quantity	Expression	Notes
Early finish	EF = ES + D	ES = max of predecessor EFs
Late start	LS = LF − D	LF = min of successor LSs
Total float	TF = LS − ES = LF − EF	Slip available before the project moves
Free float	FF = min(successor ES) − EF	Slip available before any successor moves; FF ≤ TF
Critical path	TF = 0 along the path	Longest path; its length = project duration
Cost slope	(crash $ − normal $) / (normal D − crash D)	Crash lowest slope on the critical path first
Delay effect	slip = delay − TF (if positive)	Project moves only after float is consumed

Connecting this to your overall Construction exam strategy

Scheduling sits in a cluster of cost-and-time domains that share vocabulary and often share question stems. Estimating questions (6–9 on the spec) use the same crew-day and productivity arithmetic — our guide to cost estimating and quantity takeoff on the PE exam covers that side, including the earned-value indices that NCEES files under estimating rather than scheduling. Sequencing logic also feeds the operations questions — equipment cycle times and production rates set the activity durations that CPM consumes — covered in our breakdown of construction operations and methods problems. For how all eleven knowledge areas fit together, start from the PE Civil Construction exam guide.

Final thoughts

Scheduling is the one Construction domain where you control every variable: no reference navigation, no judgment calls, no unfamiliar standards — just a procedure that rewards repetition. Treat the forward and backward passes the way you once treated unit conversions: drill until the mechanics disappear and only the reading of the question remains. Locking down 7–11 questions before exam day isn't a study strategy so much as a standing decision, and it frees your review time for the domains that genuinely need it.