Day 1

Workday Timeline: Queue, Fatigue, Cobweb (2D)

A 24 h swimlane of one worker: tasks queue up at work, fatigue builds and recovers. Twin of the 3D desk.
Switch to 3D
Utilisation rho-
Leisure time-
Effective sleep-
Mean queue (theory)-
Mean wait (theory)-
Waiting now / average so far-
Fatigue on waking (repeating day)-
Peak fatigue today-
Backlog growth per day-
Day map slope A-
Mean queue L = rho/(1-rho) (violet); this setting (green dot)
Fatigue on waking, day by day from a rested start (amber) and the repeating-day fixed point F* (green dashed)
The science
Illustrative model: the fatigue coefficients are round numbers chosen for sensible shapes, not fitted to data. Nothing here is medical or HR advice.

Work as a queue (M/M/1). Tasks arrive during working hours as a Poisson stream with rate lambda; you finish them one at a time with exponential service times of rate mu. With utilisation rho = lambda/mu < 1 the mean number of tasks in the system is L = rho/(1 - rho) and the mean time in the system is W = 1/(mu - lambda); Little's law links them, L = lambda W. At rho >= 1 there is no steady state: the backlog grows by about (lambda - mu) tasks per working hour. L rises steeply near rho = 1: a "busy but fine" 90 % load already means about nine tasks waiting.

A 24-hour time budget and fatigue. The day is split into commute, work, commute, leisure and sleep. Fatigue F follows dF/dt = a - r F in each phase: it rises during work (faster when the queue is long) and commuting, and relaxes exponentially during leisure and (faster) sleep. Because every phase is linear the day is an affine map F_next = A F + B, so the repeating-day value is F* = B/(1 - A). If the activities exceed 24 h the model takes the missing hours from sleep.

Cobweb (new in this 2D view). Plot F_next against F: the day map is a straight line with slope A, the fixed point F* is where it crosses the diagonal, and the staircase of arrows shows each day's fatigue converging there at rate A^n. Slope A close to 1 means recovery is nearly absent and fatigue accumulates for many days.

Simplifications: real work is not Poisson (meetings and deadlines cluster), tasks differ in size and priority, people multitask, and several people share load. Recovery is not just the length of rest but its quality. The model shows a mechanism (queues blow up near full utilisation; short sleep leaves a residue) rather than predicting any individual.

Independent numerical checks
Made with 🟠 Blender