Construction scheduling,
term by term.

Monte Carlo simulation

Also called: Monte Carlo, monte-carlo, Monte Carlo analysis

A Monte Carlo simulation runs the schedule thousands of times, each pass drawing a different duration for every activity from a range you set, and reports the spread of finish dates that come out. Instead of one date it gives you a curve: how often the job finishes by a given day.

GanttAI models change deterministically: you ask what happens if steel slips three weeks and it reflows the network and tells you. It does not run probabilistic simulations.

Schedule risk analysis

Also called: SRA, risk analysis, quantitative risk analysis, QSRA

Schedule risk analysis puts duration ranges and risk events onto a CPM schedule and simulates it, usually by Monte Carlo, to find how likely the planned date really is and which activities drive the uncertainty. The output is a confidence curve and a risk-driver ranking, not a single date.

GanttAI verifies that a schedule is sound and models specific what-ifs against it. Probabilistic risk analysis is not something it does.

P50

Also called: P80, P90, P50 P80, confidence level, probability of completion

In a risk analysis, P50 is the date the job finishes by half the time, P80 the date it beats four times in five. Owners often ask for a plan built to P50 and a commitment made at P80; the gap between the two is the contingency you are carrying.

Critical chain

Also called: CCPM, critical chain project management

Critical chain strips the padding out of individual activities and pools it into buffers: a project buffer before the finish date, feeding buffers where paths join the chain. Work is managed by how fast the buffers are being consumed rather than by activity-level dates.

Schedule margin

Also called: schedule buffer, contingency time, float ownership

Schedule margin is time held deliberately against risk, owned by whoever put it there and shown as its own activity rather than hidden inside durations. Making it visible is what stops the argument about who owns the float when the job runs late.

Earned value management

Also called: earned value, EVM

Earned value management measures progress in the currency of the plan: budgeted cost of work performed against what was planned and what was spent. It gives you a cost and a schedule variance from the same set of numbers, so progress is a measurement rather than an opinion.

Schedule performance index

Also called: SPI

SPI is earned value divided by planned value. Below 1.0 you are behind the plan in value terms. It is blunt about one thing: work completed off the critical path still counts, so a healthy SPI does not mean a healthy finish date.

Cost performance index

Also called: CPI

CPI is earned value divided by actual cost. Below 1.0 the work is costing more than it earned. Paired with SPI it separates a job that is late from a job that is expensive, which are different problems with different fixes.

Earned schedule

Also called: earned schedule method

Earned schedule fixes EVM's blind spot at the end of a job, where SPI drifts back to 1.0 no matter how late you are, by measuring the variance in time rather than in cost. It answers how many weeks behind you are, not how many dollars.

Time impact analysis

Also called: TIA, time impact evaluation

A time impact analysis inserts a modelled delay, a fragnet, into the schedule as it stood just before the event and reruns CPM. The difference in the completion date is the entitlement. It is the method most contracts name because it is prospective: it uses what was known at the time.

Fragnet

Also called: fragment network, sub-network

A fragnet is a small piece of network, a handful of activities and their logic, modelling one event: a change order, a delay, a redesign. It gets inserted into the schedule so the effect can be measured against the schedule that existed before it.

Windows analysis

Also called: window analysis, contemporaneous period analysis

Windows analysis cuts the job into periods, usually update to update, and measures what drove the finish date in each one. It is slower than the alternatives and harder to argue with, because it follows the critical path as it actually moved rather than assuming it stood still.

As-planned versus as-built

Also called: as planned vs as built, as-built analysis, as planned as built

As-planned versus as-built lays the schedule you intended over the one that happened and reads off the differences. It is the simplest delay method and the weakest, because it shows what changed without showing what caused it.

Concurrent delay

Also called: concurrency, pacing delay

Concurrent delay is two delays running at the same time, one the contractor's and one the owner's, either of which would have pushed the date on its own. It usually means time is granted but money is not, which is why the analysis is fought over so hard.

Liquidated damages

Also called: LDS, delay damages

Liquidated damages are a per-day sum fixed in the contract for finishing late, agreed up front so nobody has to prove actual loss later. They are what makes the completion date on a schedule a financial instrument rather than a target.

Work breakdown structure

Also called: WBS

A work breakdown structure is the hierarchy the scope hangs on: the job split into deliverables, then into packages, then into the activities you actually schedule. It decides what can be rolled up and reported, so it is worth settling before the first activity is typed.

Leads and lags

Also called: lag, lead, negative lag, lag time

A lag delays a successor after its relationship is satisfied: pour, then wait seven days to strip. A lead, written as a negative lag, lets a successor start before its predecessor finishes. Leads are the ones that draw scrutiny; a negative lag on a finish-to-start link is usually a missing activity in disguise.

Finish to start

Also called: relationship type, FS SS FF SF, start to start, finish to finish, dependency type, logic type

Four relationship types tie activities together. Finish-to-start is the default and should be most of your logic. Start-to-start and finish-to-finish run work in parallel with an offset. Start-to-finish is rare enough that finding one usually means finding a mistake.

Predecessor

Also called: successor, driving relationship, driving predecessor

A predecessor is what has to happen before an activity can proceed; a successor is what waits on it. The driving predecessor is the one actually setting the start date, as opposed to the others that have slack and are just along for the ride.

Open end

Also called: dangling activity, dangler, missing logic, open ends

An open end is an activity with no predecessor or no successor, so nothing pushes it and nothing depends on it. Open ends are why a schedule can pass a glance and still be unable to tell you what a delay costs: the date has nothing holding it in place.

Constraint

Also called: date constraint, hard constraint, must finish on, start no earlier than

A constraint pins an activity to a date rather than letting logic decide it. Soft ones (start no earlier than) are ordinary. Hard ones (must finish on) override the network and can hide negative float, which is why every checking standard counts them.

Negative float

Also called: negative slack, behind the constraint

Negative float means the network needs more time than the constraint or the completion date allows: minus ten days says the path is ten days late before anyone has picked up a tool. It is a symptom, either of real slippage or of a constraint that should not be there.

Near-critical path

Also called: near critical, secondary critical path

The near-critical paths are the ones with little float left, often set at ten days or less. They matter because a critical path is only critical until something moves: manage the top path alone and the second one becomes your problem the following week.

Longest path

Also called: longest path value, LP

The longest path is the chain that actually drives the finish date, traced backwards from completion through driving relationships. On a schedule with constraints or multiple calendars it is a truer read of what is critical than a plain total-float threshold.

Milestone

Also called: milestones, key date, interim milestone

A milestone is a zero-duration marker for an event rather than a piece of work: permit issued, building watertight, power on. Contract milestones often carry their own damages, so they are worth logic-linking as carefully as the work itself.

Level of effort

Also called: LOE, hammock activity, hammock

A level of effort activity covers work that runs alongside the job rather than advancing it: supervision, site facilities, quality management. It stretches to fit whatever it spans, so it should never sit on the critical path.

Rolling wave planning

Also called: rolling wave, progressive elaboration

Rolling wave planning details the near work fully and leaves the far work in bigger blocks, filling it in as design and procurement firm up. It keeps a two-year schedule honest instead of pretending week 84 is knowable today.

Data date

Also called: status date, as of date, time now

The data date is the line between actual and planned: everything to the left has happened, everything to the right is forecast. Nearly every argument about a schedule update starts with someone reading it against the wrong data date.

Retained logic

Also called: progress override, out of sequence progress, out-of-sequence

When work happens out of sequence the software has to choose. Retained logic holds the remaining work behind its predecessors; progress override ignores the broken link and lets it run. Override usually produces the earlier date and the less defensible one, and the setting is easy to change without anyone noticing.

Percent complete

Also called: physical percent complete, duration percent complete, progress

Duration percent complete is time elapsed. Physical percent complete is work actually installed. They part company the moment a job runs slowly, so which one an update is reporting decides whether it is measuring progress or measuring the calendar.

Remaining duration

Also called: original duration, RD, OD

Original duration is what the activity was planned to take; remaining duration is what the person doing it now says is left. Remaining duration is the number that drives the forecast, and it is the one worth challenging in an update meeting.

Calendar

Also called: work calendar, shift calendar, non-work days, working days

A calendar defines what counts as a working day for an activity: shifts, weekends, holidays, seasonal shutdowns. Multiple calendars on one schedule are normal and are also where float goes strange, because a day of float on a five-day calendar is not a day on a seven.

Weather days

Also called: weather day, adverse weather, weather allowance

Weather days are non-working days built in for the weather a normal year brings, usually by month and by activity type. Allowed for in the calendar, bad weather is a plan; left out, it becomes a claim.

Schedule basis

Also called: basis of schedule, basis memo, narrative

The basis of schedule is the memo saying what the schedule assumed: productivity rates, crew sizes, shift patterns, procurement leads, what was excluded. Without it a schedule is a set of bars nobody can check, and a year later nobody can remember.

Schedule of values

Also called: SOV

A schedule of values breaks the contract sum into line items to bill against. Tied to schedule activities it drives cash flow forecasting; left untied it is a separate document that slowly stops matching the work.

Notice to proceed

Also called: NTP

Notice to proceed is the owner's instruction to start, and usually the point the contract clock begins. Most schedules key their first activity to it, which makes an NTP date change one of the more expensive things that can arrive by email.

Request for information

Also called: RFI

An RFI asks the design team to resolve something the documents do not answer. Logged against the activities they hold up, RFIs turn into a measurable delay; kept in a separate log, they turn into a surprise.

Submittal

Also called: submittals, shop drawings, submittal register

A submittal is the shop drawing, sample or data sheet approved before fabrication starts. The review cycle is real duration on the critical path for anything long-lead, and it is one of the most common things missing from a schedule that looks complete.

Long lead item

Also called: long-lead, procurement, lead time, long lead

A long-lead item takes longer to procure than the work in front of it takes to build: switchgear, chillers, curtain wall, lifts. Once the fabrication and delivery are in the schedule as activities, procurement usually turns out to be driving more of the job than the site work is.

Substantial completion

Also called: practical completion, beneficial occupancy, turnover

Substantial completion is the point the owner can use the building for its purpose, even with items outstanding. It typically stops liquidated damages and starts warranties, which is why it is the date most schedules are really built around.

Punch list

Also called: snag list, punchlist, deficiency list

The punch list is the outstanding defects and unfinished items found at walkthrough. Scheduled as a single bar at the end it is always optimistic; scheduled by area, following the trades out, it tends to survive contact with the job.

Commissioning

Also called: CX, systems commissioning, testing and commissioning

Commissioning proves the installed systems work together and to specification: static checks, energisation, functional testing, integration. It needs the building weathertight and powered, so it is the phase most exposed to everything that ran late before it.

4D BIM

Also called: 4D, 4D scheduling, BIM scheduling, 4D model

4D ties schedule activities to model objects so the sequence can be played back as the building going up. It is unusually good at catching what a bar chart hides: two trades in the same space, or access that closes before the work behind it is done.

S-curve

Also called: s curve, progress curve, cumulative curve

An S-curve plots cumulative progress, cost or hours against time, and takes its shape from a job that starts slowly, runs hard through the middle and tails off. Planned against actual, the gap between the two curves is the earliest honest warning you get.

Gantt chart

Also called: bar chart, Gantt

A Gantt chart draws activities as bars on a timeline. It became a schedule rather than a picture when CPM logic was added underneath: without the relationships it shows when work was meant to happen but not what happens when it doesn't.

Activity code

Also called: activity codes, coding, activity id

Activity codes are the labels you filter and group by: area, level, trade, phase, responsibility. They decide what views the schedule can produce later, which is why coding is worth agreeing before the schedule is built rather than after.

Cost loading

Also called: cost loaded schedule, resource loading

Cost loading attaches money to activities so the schedule produces a cash-flow forecast and a basis for progress payments. It also raises the stakes on the logic: once payment follows the bars, the sequence stops being a planning question.

Mandatory logic

Also called: discretionary logic, hard logic, soft logic, preferential logic

Mandatory (hard) logic is physics and code: you cannot roof before you frame. Discretionary (soft) logic is preference: this crew does the east side first. Knowing which is which is what makes acceleration possible, because only one of the two is negotiable.

Zone

Also called: zones, area, work area, sector, phasing

A zone is a physical piece of the job trades move through in turn: a floor, a wing, a pour. Scheduling by zone rather than by trade is what keeps crews out of each other's way and makes hand-offs something you can plan instead of referee.