A person employed on 40 hours a week does not deliver 40 hours of plant work every week of the year. Leave, public holidays, sickness, training and unallocated time all come out of the same 2,087 nominal hours. This tab works out the productive hours one person actually delivers, and from that the availability factor — the fraction of a nominal full-time equivalent that one hire is worth.

• Annual leave — the entitlement your staff are actually able to take. If leave is not taken it becomes a liability and a fatigue problem, not a saving.
• Public holidays — 12 in New Zealand. Staff still attend and are still on call on these days; the days are removed from ordinary productive time, not from on-call availability.
• Sick leave — plan for expected usage, not zero. A plan with no sickness allowance is brittle.
• Training and development — competency is a regulatory expectation, not a discretionary extra.
• On-tools factor — the share of attended time that is real productive work once breaks, hand-overs, travel to depot, paperwork and interruptions are removed. Set it in Settings (gear icon).

Getting this wrong is the single most common cause of understaffing: a workload of 6,000 hours divided by 2,087 says "you need 3 people", when the honest answer is closer to 4.

On-call cover is a separate constraint from workload. Even a small supply with very little routine work needs enough competent people to hold a sustainable after-hours roster. Very often the roster, not the workload, is what determines the minimum team size.

• 1 week in N — the on-call frequency each rostered person experiences. 1-in-1 means always on call. Good practice for a sustained roster is 1 week in 4 or better.
• A 1-in-4 roster does not need 4 people. People take leave, get sick and go to training, and the roster still has to be filled 52 weeks a year. Allowing for that, a 1-in-4 roster with one person on call typically needs about 4.5 to 5 people in the pool.
• Stand-down — rest after a night call-out. If a person is out at 2 am, they should not be driving to a plant and handling chlorine at 8 am. Honouring stand-down costs productive hours; not honouring it transfers the cost to safety.
• Off-roster response — if staff who are not on call are still expected to answer the phone, the roster on paper is not the roster in practice. Nobody is ever truly off duty.

This tab sets the operating policy that shapes the roster: which tasks need two people, how after-hours two-person tasks are handled, lone-worker monitoring, and the fit-for-duty rule for anyone called in.

Some plant activities should never be done by a person on their own. A chlorine drum or cylinder change is the classic example: if the person is overcome, there is nobody to raise the alarm, isolate the leak or pull them clear. A lone worker at 2 am at an unstaffed rural plant has no such backstop.

Set the policy for each activity below. The list is a starting point, not exhaustive — use Add activity to capture anything specific to your plants. Two things then happen in the model:

• Hours — a two-person task adds the second person's time to the workload. It is real work and it has to be staffed, whether it is done on the day or deferred to the next.
• Risk — the model flags a high-hazard activity marked as permitted for a lone worker. It also checks how two-person tasks that arise after hours are handled, and whether anyone called in off duty is fit for it.

Two-person tasks on call-out. Most can be made safe and left until the next working day, when a second person is available in normal hours; escalation covers the rare case that cannot wait. That is normal, safe practice and does not require a second person on call every night. A formal second on-call roster is only needed where a two-person task genuinely cannot be deferred — set that on this tab and the model will size for it.

Events per year is how often the activity happens across the whole supply. Hours per event is the duration of the task, not the whole visit — enter the second person's attendance time. The primary responder's time is already counted in routine plant hours, so it is not double-counted here.

Two independent numbers are calculated, and the team must satisfy both:

• Workload requirement — total annual work divided by the productive hours one person actually delivers.
• Roster requirement — the pool of competent people needed to fill the on-call roster at the target frequency, every week of the year, including while people are on leave.

The recommended headcount is the larger of the two. Which one is binding is worth knowing: if the roster is binding, hiring for workload alone will not fix the fatigue problem, and vice versa.

The flawed planning number is shown deliberately. It is what you get if you divide the workload by contracted hours and forget that people take leave. It is almost always the number an understaffed organisation is unknowingly using.

The same workload can be staffed three very different ways. This tab runs the model three times so the trade-off is explicit.

• Minimum crew — the thinnest plausible team: 1 week on call in 2, no stand-down, staff answer the phone on their days off, lone working accepted, minimal training. It is cheaper, and the risk is carried by the staff and by public health.
• As configured — your inputs.
• Best practice — 1 week on call in 4 or better, leave taken and covered, stand-down honoured after night call-outs, no lone working on high-hazard tasks, overtime used only to bridge vacancies, and a generous training allowance. Two-person tasks that arise after hours are made safe and deferred to the next working day.

Best practice does not mean a second person on call every night. Most two-person tasks can be made safe and left until normal hours, with escalation for the rare case that cannot wait. A formal second on-call roster is reserved for genuinely non-deferrable tasks; where even that is unaffordable, the honest options are to remove the hazard (for example, move away from chlorine gas) or share cover with a neighbouring supplier and document the residual risk.

Findings are generated from your inputs against recognised good practice. Each is tagged with who carries the consequence:

• Personnel — fatigue, injury, chemical exposure, burnout, retention.
• Public health — delayed or failed response to a treatment failure; disinfection lost; a critical control point left out of range because nobody could get there.
• Organisational — regulatory exposure, health and safety duties, unmanaged overtime cost, key-person dependency, loss of competency.

This is a screening tool, not a formal risk assessment, and it is not legal or employment advice. Findings should be tested against your own hazard register, employment agreements and current legislation.

This is the only place in the tool that exports or saves anything.

• PDF report — opens your browser print dialog. Choose Save as PDF. For a clean result, set margins to Default and turn Headers and footers off in the print dialog.
• CSV — all inputs and results as a flat table for spreadsheets.
• Save file — a JSON file containing every input and setting. Keep it, share it, and load it back later to pick up exactly where you left off.