Just provide the required project information, such as blocks, storeys, floor areas, structural and wall systems, to estimate site and factory man-days, programme and crew needs. For a crew sized from an IFC take-off, use the Precast Manpower Estimator.
Result
Two columns, one set of inputs. As specified follows the schema: 0.6 m²/manday and 70% / 30% of all site labour scaled by the structural / wall index. Code-anchored starts from the Code's own Annex C baseline and scales only the structural and wall trades — their share of a cast in-situ building in the trade mix, about 33% and 15%. The gap between them is the uncertainty in the method, not a rounding difference.
What the law requires on site
MOM requires these people to be appointed before a crane lifts, and on every worksite. They are the estimator’s own rules, from the Workplace Safety and Health Regulations, counted for the Code-anchored column. The lifting team is compared with the “crane and riggers” trade the answer above contains; the site appointments must be among the workforce, not added to it.
How it works, and why
In plain words, before any arithmetic. Every step below is a row in How each figure was reached, where its numbers are shown.
- Labour follows floor area. A bigger building takes more work, and floor area is the only size a project knows before it has been designed. It is also what BCA measures site productivity against, so it is the honest place to start.
- A published rate turns area into man-days. The Code's Annex C gives, for each kind of development, how much floor area one worker completes in a day — 0.439 m² for public housing, 0.319 for private. Area divided by that rate is the man-days the building needs. Housing beats a hospital because the work repeats.
- How you build moves labour; it does not delete it. Casting concrete on site is labour on site. Precast makes the same components in a factory, and PPVC makes whole fitted rooms. Fewer people are needed on site and more in a factory — the work does not vanish, which is why site and factory man-days are reported separately.
- Repetition earns a discount. The tenth identical floor is quicker than the first: the crew has learned the cycle and the formwork already exists. The more identical storeys, the larger the discount.
- The site itself costs labour. A steep plot with one narrow gate means more double-handling, more waiting for the crane and more standing about. The building is unchanged; the labour to build it is not.
- People are man-days divided by time. The programme is piling, then each block's floors at the cycle you choose with the blocks started a stagger apart, then fit-out. Spreading the man-days over the working days in that programme gives how many people must be on site at once — which is why a slower floor cycle lowers the peak: the same labour spread over more weeks is fewer people at a time.
- The trade table divides the total; it does not change it. The mix says what share belongs to formwork, rebar, M&E, finishes and the rest. It is a breakdown of the figure above, not a second opinion on it.
- The range exists because the inputs are estimates. Piling overruns more often than it finishes early, and a basement area at concept stage is a sketch. The simulation varies each input across its own declared range thousands of times, so the spread you see is the spread your own assumptions imply.
What to trust, and how far. The productivity baseline is published — the Code's own Annex C, read from the dataset rather than copied here. Most of what adjusts it is not: the labour-saving indices, natural cycles, off-site shares, repetition bands, site multipliers and PPVC figures come from the older Buildable Design Appraisal System, because the 2022 Code publishes points per system rather than indices — and none of them has been checked against measured site returns. The trade mix and its phase windows are a separate proposal's, rescaled to add up. So the shape of the answer is informative and the exact figure is an estimate carrying that uncertainty. That is why two columns are shown side by side: the gap between them is the size of the doubt, not a rounding difference.
The figures behind the answer — change any of them
Every figure the calculation uses that is not a project input is here, grouped by what it describes. Each group says what its figures mean; click a figure to see what it is, why it is needed, and how far this project’s answer moves if it goes up by a tenth. Change one and the whole page recalculates. Tinted figures were set by the building type; outlined ones are yours. Your changes go with a saved scenario and into the workbook’s Tables sheet.
Compare scenarios
Save what is on screen as a scenario, change something — the method, the building type, a figure — and compare. Each scenario keeps its inputs and its changed figures, never its answers: every column below is recalculated now, so they differ only for reasons they contain. Scenarios stay in this browser; Save to a file shares them with a colleague.
How sure is that? — the same model, run many times
Every input above is a guess inside a range. This draws each of them from a declared distribution and runs the whole calculation again — method index, repetition, PPVC, the phase programme, the trade curve and the weekly peak — so the answer is a range with a shape rather than a single figure computed from the middle of every guess. There is no peak factor: the peak is the busiest week of a modelled curve, which is why it is not a fixed multiple of the average. The Annex C baseline is never sampled — it is published, and giving it error bars would invent uncertainty its source does not have.
How far each input may vary — change any range
Each run draws every input below from this range, as a percentage of the value you entered on the left: 85 to 130 means anything from 15% less to 30% more, most often the value itself. Wider ranges give a wider answer; the published rates are never varied. Change a figure and the simulation re-runs.
The inputs have changed since this simulation ran. While this section is open it re-runs on them by itself (at most 2,000 runs); press Run for 10,000.
How each figure was reached
Week by week, and by trade
The weekly figures as a table
The same weeks, stacked by trade
Who is on site each week: the trades in seven groups, stacked in the order they arrive — ground works at the bottom, general labour and management on top. Point at a week to see its groups and busiest trades.
The weekly figures by trade group as a table
Check against a finished job
Every table this page uses is a proposal or an assumption until a real job says otherwise. Enter a finished job's blocks, storeys, floor plate and method above, then open its headcount: two columns, week (or month) and workers, or the estimator's manpower-records template. It is drawn over the prediction and compared below. The file is read in this tab and sent nowhere.
Against the Code's productivity baseline
Site productivity (area ÷ site mandays) against Annex C Table 4, at the 2010 level. The Code's open-solution route asks for at least 25% better, with a Project Productivity Improvement Plan to show how.
Against what has been measured in Singapore
Site productivity on real projects, from published sources. Each column is how this estimate compares: an estimate far above the best of these is claiming more than has been built.
What this rests on
The full method, every table's source and the open questions are in the method note.
Where every number comes from
- The method tables (labour-saving indices, natural cycles, off-site shares, repetition bands, site multipliers, PPVC) are the schema's. The 2022 Code publishes points per system in Annex A Tables 3 and 4, not labour-saving indices; these follow the older Buildable Design Appraisal System. None of them has been checked against measured site returns.
- The Annex C baselines are the Code's, read from the dataset at load time, not copied into this page.
- The programme is built from its phases: substructure (C5), then each block's floors at the cycle with the blocks a stagger apart (C7), then fit-out (C6). The defaults — 6 months, 4 weeks, 12 months — are this tool's reading of a Singapore high-rise job, not a published figure; change them to the project's.
- The calculation decides how much labour each part of the building takes; the trade mix only divides it. The structural mandays pay for formwork, rebar, concrete, precast installation and the crane crew; the wall mandays for masonry and plastering; the rest for the finishing trades. The structure gang per block is the structural trades' labour over each block's superstructure days, so it is the same labour the chart draws. In the As-specified column the schema's 70% is everything that is not walls, so it pays for the structural and finishing trades together.
- The trade mix, the direct/subcontracted split and the phase windows are a second proposal's: 20 trades for cast in-situ, precast and PPVC. Its columns added up to 1.03, 0.96 and 0.92, so each is rescaled. Its 42-month template is stretched so its superstructure (months 12–30) lands on this programme's. The weekly curve, the peaks and the trade table come from it.
- Two rows are this tool's, not the proposal's. Piling and earthworks takes 4% of the typical floors' labour, the whole-site model's share, through the substructure. Basement, podium and roof carries the labour of the other floor area (A6), built at the baseline rate with no method, repetition or PPVC adjustment. Masonry and plastering are named for the wall method: dry partition makes them drywall partitions and skim coat.
- A check that the Code-anchored column is in the right place: a conventional building — two-way beam and brickwall — comes out within about 7% of the Code's own 2010 baseline for private housing. The labour-saving methods then improve on it.
- The peak is about 2½ times the average. The piling months and the last months of fit-out carry few people, and the trade-mix proposal's windows put most trades inside the superstructure and the first half of fit-out. Only measured headcount returns from finished jobs can say whether that is too peaky.
What changed from the schema, and why
- Off-site labour under PPVC is 40% of the method mandays before the 40% site cut. The pseudocode took it after, which makes it 24%.
- Tables 1 and 2's off-site percentages are applied (share × that component's site labour). The pseudocode defined them and never read them, so precast without PPVC had no factory labour at all.
- Steps 5–7 were cut off in the schema and are completed here: terrain × access multiply (so steep and very limited is 1.5625, not 1.25); the programme is substructure + superstructure + fit-out; crew = site mandays × 8 ÷ hours a day ÷ working days.
- Each block has its own structure gang, started a stagger after the one before. A job that builds its blocks strictly one after another is a stagger as long as a block's superstructure.
- Table 4's last row reads "40 18%"; it is taken as more than 40 levels.
- From the trade-mix proposal, three things were left out: a subcontractor multiplier of 1 ÷ (1 − 65%) applied on top of a crew that already counts everyone on site (it nearly triples the answer); a fixed peak of 1.4 × average; and a monthly profile with fixed crews (30 on piling, M&E rising 5 a month) that does not add up to the mandays.
Open questions for the schema's authors
- 0.6 m²/manday. Annex C Table 4 gives 0.439 (public) and 0.319 (private) at the 2010 level. If 0.6 is a current measured HDB figure, its source should be cited. If it is a target, the tool is predicting a target, not a demand.
- Whole-site scaling. Scaling 70% of all labour by the slab system also changes the M&E and the finishes, and gives them a 1.3 m²/manday productivity — four times the Code's baseline. Which trades should the structural index reach?
- PPVC on top of B1. PPVC replaces the structural and wall systems, but the schema applies its 40% cut on top of their indices, so it is counted twice. Should B1/B2 be ignored, or fixed to their PPVC values, when B3 is Yes?
- Other floor area. A6 now counts basement, podium and roof at the plain baseline rate. Should any of it take a method factor — a precast podium, say?
Quick estimate — the rule of thumb
Peak workers from a floor area in five steps: floor area ÷ a productivity norm gives the man-days; spread over the days actually worked gives the average; times a peak factor gives the peak. Type a site and a plot ratio, or leave them blank to use the project above. Every field starts from the page; type over it to change it, clear it to go back.
Three ways to the peak
The rule of thumb above, the detailed model’s two columns and the simulation’s range, on one axis.