---
title: "Resource loading, levelling and schedule maintenance"
description: "Logic is not enough A logically perfect schedule can still be impossible if it needs three tower cranes on a site that has room for one, or the same…"
url: https://optimizeall.com/learn/project-controls-with-ai/resources-and-schedule-maintenance
updated: 2026-10-05
---

Project Controls in the AI Era · Foundations of project controls · lesson 4 of 22 · 14 min

# Resource loading, levelling and schedule maintenance

## Logic is not enough

A logically perfect schedule can still be impossible if it needs three tower cranes on a site that has room for one, or the same commissioning engineer in two cities in the same week. **Resource loading** assigns labour, equipment and materials to activities; **resource levelling** adjusts timing so demand does not exceed availability.

## Step by step: resource-loading a schedule

1. **Pick the resources that matter.** Load the constrained ones: specialist crews, critical equipment, key engineers, long-lead materials. Loading every individual is rarely worth it.
2. **Estimate quantities and productivity.** Duration should come from quantity ÷ productivity ÷ crew size, not from a guess. Example: 1,200 m of cable tray ÷ 40 m per crew-day ÷ 2 crews = 15 days.
3. **Assign resources to activities** with units per day.
4. **Review the histogram.** Plot demand per week against availability.
5. **Level.** Delay non-critical activities within float first; only then consider extending the project or adding resources.
6. **Check the result.** Levelling consumes float and may create a new critical path. Re-run CPM and review.

## Resource histogram template

```
Resource: Electrical crew (available: 6 crews)
Week        W1  W2  W3  W4  W5  W6  W7  W8
Demand       4   6   9   8   5   3   6   4
Over-limit   -   -   3   2   -   -   -   -
Action: shift 'Lighting L3' (float 8d) to W5–W6; shift 'Small power L4' (float 12d) to W6–W7
```

## Productivity-based durations: worked example

*Illustrative.* An Abu Dhabi mid-rise residential project (fictional, "Corniche Heights") planned blockwork at 20 m² per mason-day with 10 masons, for 12,000 m²: 12,000 ÷ (20 × 10) = 60 working days. After three weeks, actual productivity was 15 m² per mason-day. The remaining 9,000 m² would need 9,000 ÷ (15 × 10) = 60 more days instead of the 45 the original plan implied. Because blockwork was on the critical path, the forecast finish moved by 15 working days unless crews were increased. The planner modelled adding 3 masons (9,000 ÷ (15 × 13) ≈ 46 days) and checked that site logistics could support them. This kind of quantity-based updating is far more reliable than adjusting percent complete.

## Schedule maintenance: the weekly and monthly rhythm

| Frequency | Task |
|---|---|
| Weekly | Status actual dates, remaining durations, look-ahead (3–6 weeks) with site teams |
| Weekly | Constraint log: permits, drawings, materials, access needed before work can start |
| Monthly | Full update, critical path review, float erosion analysis, earned schedule |
| Monthly | Compare to baseline; document changes to logic with reasons |
| As needed | Incorporate approved changes into a new baseline version |

## Look-ahead planning and constraint removal

Short-interval planning (look-aheads) bridges the master schedule and the field. For every activity due in the next few weeks, ask: are drawings approved, materials on site, permits in hand, predecessor work done, labour available? An activity with an open constraint is not ready; plan to remove the constraint or re-sequence. This practice, associated with lean construction approaches, reduces the gap between what was planned and what actually gets done each week.

## Schedule change control

Keep a **schedule change log**: every logic change, duration change beyond a threshold, or constraint added, with who, when and why. Forensic delay analysis in claims (common on large projects in the Gulf, UK and US) depends on a trustworthy record of how the schedule evolved.

## Baseline, current and target schedules

- **Baseline:** approved; changes only through change control.
- **Current (forecast):** updated each period; shows where you are heading.
- **Target or recovery:** a scenario showing how you plan to recover; clearly labelled so it is never confused with the forecast.

## Common mistakes

- Durations not derived from quantities and productivity.
- Levelling that silently pushes the finish date without anyone noticing.
- No look-ahead, so the master schedule and the field drift apart.
- Logic changes without a record.
- Presenting a recovery schedule as if it were the forecast.

## AI in resource planning

Optimisation tools can generate many levelling scenarios and suggest the one with the least delay or cost. They work only if logic, calendars and resource limits are accurate, and a planner must check the proposal is buildable in the real world, with safety and site access in mind.

## Hands-on: productivity-based forecasting in Excel

Track one quantity-driven activity per row and let the sheet re-forecast from actual productivity:

```text
Columns: A Activity | B Total qty | C Installed to date | D Crew-days used | E Planned rate | F Crew size
G Actual rate          =IFERROR(C2/D2, E2)
H Remaining qty        =B2-C2
I Remaining days       =ROUNDUP(H2/(G2*F2), 0)
J Planned remaining    =ROUNDUP(H2/(E2*F2), 0)
K Slip (days)          =I2-J2
L Crew needed to hold  =ROUNDUP(H2/(G2*J2), 0)
```

For Corniche Heights: B 12,000, C 3,000, E 20, F 10 and an actual rate of 15 give I = 60, J = 45, K = 15 and L = 14 masons (with 13, remaining ≈ 46 days, one day of slip). Column L is only a starting point; site logistics and safety decide what is feasible.

## Hands-on: a weekly look-ahead constraint log

| Activity | Planned start | Drawings | Materials | Permit | Predecessor | Labour | Ready? |
|---|---|---|---|---|---|---|---|
| Blockwork L5 | W23 | ✓ | ✓ | n/a | ✓ | ✓ | Yes |
| MEP first fix L3 | W24 | Rev C pending | ✓ | n/a | ✓ | 2 of 4 crews | No |

```text
Ready?  =IF(COUNTIF(C2:G2,"✓")+COUNTIF(C2:G2,"n/a")=5,"Yes","No")
```

Track the weekly **percent plan complete** (activities completed as planned ÷ activities planned for the week). A rising trend shows the look-ahead is working.

## How to measure success

- Share of critical activities whose durations have a documented quantity and productivity basis.
- Percent plan complete trending upwards over several weeks.
- Every levelling or logic change recorded in the schedule change log with its effect on the finish date.

## Video lecture: Resource loading, levelling and schedule maintenance

Lecture coming soon · 9 chapters · about 8 minutes. Read the full transcript below.

1. A perfect schedule that cannot be built
2. Why it matters
3. The concept: a restaurant kitchen
4. Worked example one: cable tray
5. Worked example two: Corniche Heights blockwork
6. Watch me do it: a levelling histogram
7. The maintenance rhythm and look-aheads
8. Common mistakes
9. Recap and try this now

## Lecture transcript

### A perfect schedule that cannot be built

Here's a schedule I once reviewed. Beautiful logic. No open ends. Critical path clearly marked. And in week fourteen it needed three tower cranes on a site with room for exactly one. The schedule was logically perfect and physically impossible. In this lecture you'll learn how to fix that: how to resource-load the activities that really matter, how to level them without silently pushing the finish date, and how to derive durations from quantities and productivity instead of guesswork. Then we'll look at the weekly and monthly rhythm that keeps a schedule alive, including look-ahead planning and constraint removal. By the end you'll be able to update a forecast from real productivity data in a few minutes.

### Why it matters

Why does this matter? Three reasons. First, logic alone ignores capacity. If two activities need the same specialist crew at the same time, one of them will wait, whatever the Gantt chart says. Second, durations that were guessed can't be tested. If you don't know the quantity and the expected productivity behind a duration, you can't tell whether today's progress is good or bad. And third, a schedule that isn't maintained forecasts nothing. The master programme drifts away from what's happening on site, and people stop trusting it. When that happens, the site runs on whiteboards and the schedule becomes a monthly reporting ritual.

### The concept: a restaurant kitchen

Think of a restaurant kitchen on a Saturday night. You might have plenty of chefs, but only one pizza oven. The oven is the constraint, and the order in which dishes go in decides when tables get fed. Resource planning works the same way. First, load only the constrained resources: specialist crews, critical equipment, key engineers, long-lead materials. Loading every individual is rarely worth it. Second, derive durations from quantity divided by productivity times crew size. Third, look at the histogram of demand against availability. Fourth, level: delay non-critical activities within their float first, and only then consider extending the project or adding resources. And fifth, here's the key idea, re-run the critical path afterwards, because levelling consumes float and can create a brand new critical path.

### Worked example one: cable tray

Let's start with a simple calculation. You need to install one thousand two hundred metres of cable tray. Your historical productivity is forty metres per crew per day, and you have two crews. So the duration is twelve hundred divided by forty, divided by two. Fifteen working days. Now, why is this better than someone saying 'about three weeks'? Because it's testable. After three days you can measure how many metres are installed. If you're at one hundred and eighty metres instead of two hundred and forty, you know productivity is running at seventy-five per cent of plan, and you can forecast the remaining duration with real data rather than hope.

### Worked example two: Corniche Heights blockwork

Now a realistic scenario from the lesson text. Corniche Heights is a fictional mid-rise residential project in Abu Dhabi. Blockwork is planned at twenty square metres per mason-day, with ten masons, for twelve thousand square metres. That's sixty working days. After three weeks, actual productivity is fifteen, not twenty. So the remaining nine thousand square metres will take nine thousand divided by fifteen times ten. Sixty more days, instead of the forty-five the plan implied. Blockwork is on the critical path, so the forecast finish moves by fifteen working days unless something changes. The planner models adding three masons: nine thousand divided by fifteen times thirteen, roughly forty-six days. Then comes the crucial human question. Can site logistics, access and supervision actually support thirteen masons safely? Only then does it go into the forecast.

### Watch me do it: a levelling histogram

Let me show you how I level a single constrained resource. I've got a simple sheet: weeks across the top, activities down the side, crews per week in each cell, and a total row. Above the chart I draw the limit: six electrical crews available. Weeks three and four are over, at nine and eight. Now I look at float. Lighting on level three has eight days of float. Small power on level four has twelve. So I shift lighting into weeks five and six, and small power into six and seven. The histogram now sits under the line. But I'm not done. I apply the same moves in the scheduling tool, re-run the critical path, and check the finish date. If it moved, levelling wasn't free, and that has to be reported, not buried.

### The maintenance rhythm and look-aheads

Next, maintenance. Every week, record actual dates and realistic remaining durations, and run a three to six week look-ahead with the site or delivery teams. For every activity in the window, ask: are drawings approved, materials on site, permits in hand, predecessors done, labour available? An activity with an open constraint isn't ready, so you either remove the constraint or re-sequence. That practice, associated with lean construction, dramatically narrows the gap between planned and actual weekly work. Every month, do a full update: review the critical path, float erosion and earned schedule. And keep three schedules clearly distinct. The baseline, which changes only through change control. The current forecast. And any recovery or target schedule, clearly labelled so nobody confuses what you hope with what you forecast.

### Common mistakes

Let's cover the common mistakes. Durations that aren't derived from quantities and productivity. Levelling that quietly pushes the finish date without anyone noticing. No look-ahead, so the master schedule and the field drift apart. Logic changes made without a record, which becomes a serious problem if there's ever a delay claim, whether that's in the Gulf, the UK or the US. And, perhaps most damaging, presenting a recovery schedule as though it were the forecast. A recovery schedule shows what you'll do if actions succeed. The forecast shows what you honestly expect. Mixing them up destroys trust very quickly.

### Recap and try this now

To recap. A schedule has to respect capacity as well as logic. Load the constrained resources, level within float first, and always re-run the critical path afterwards. Derive durations from quantity and productivity so progress can be tested and forecasts updated with real data. Keep the schedule alive with weekly look-aheads and constraint logs, a monthly full update, and a clear separation between baseline, forecast and recovery versions. A quick note on AI: optimisation tools can generate dozens of levelling scenarios, but they only work if logic, calendars and limits are accurate, and a planner must check the result is buildable and safe. Your try-this-now: build a simple eight-week histogram for one constrained resource on a project you know, find the over-allocations, and propose levelling moves that use float.

## Key takeaways

- Load and level the constrained resources; logic alone can produce an impossible schedule.
- Derive durations from quantity ÷ productivity ÷ crew; update forecasts using actual productivity.
- Weekly look-aheads with constraint checks connect the master schedule to the field.
- Keep baseline, forecast and recovery schedules distinct and log every logic change.

## Try it

Build a simple 8-week histogram for one constrained resource on a project you know, identify over-allocations and propose levelling moves that use float.

- [Previous: Scheduling and the critical path](https://optimizeall.com/learn/project-controls-with-ai/scheduling-and-critical-path)
- [Next: Cost engineering: estimating methods and estimate classes](https://optimizeall.com/learn/project-controls-with-ai/cost-engineering-and-estimating)
- [All lessons of Project Controls in the AI Era](https://optimizeall.com/learn/project-controls-with-ai)