This simulator follows a single motor-pump asset that runs, fails, waits for repair and runs again. A seeded random-event simulation draws each operating and repair interval, so you can compare what a short trial actually records with what the long-run availability formula predicts.
• A 3D motor-pump asset, repair workstation, maintenance clock and a strip showing recent operating states. • Sliders for mean operating time to failure (5-100 h), mean repair time (1-20 h) and a repeatable trial seed (1-99). • Six live readouts: steady-state availability, probability up at the current time, observed uptime fraction, observed failures, completed repairs and accumulated uptime in hours. • Two presets: Frequent failures (long-run availability 0.5, although a short sample can differ greatly) and Faster repair (long-run availability 20/21).
With failure rate lambda = 1/MTBF and repair rate mu = 1/MTTR, the long-run availability is A = mu / (lambda + mu) = MTBF / (MTBF + MTTR). The probability of being up at time t, starting up, is P_up(t) = A + (1 - A) exp[-(lambda + mu) t]. The simulation draws each interval as -ln(U) divided by the relevant rate, where U is a uniform random number.
The model is a two-state continuous-time Markov process with independent exponential up and repair intervals, starting up, and unlimited repair resources for this single asset. MTBF here means mean operating uptime between failures, excluding repair time, and one animation second equals one hour. A short trial has few events, so its observed uptime fraction can differ widely from the formula; change the seed to see that sampling variability.
MTBF is the mean operating time between failures, a measure of how often the asset fails. MTTR is the mean time to repair, how long it is down each time.
No. The formula describes the long-run average. A short trial with only a few failures shows sampling variability, and changing the random seed shows this directly.
Yes. Repair changes the downtime per failure, not the distribution of the time to the first failure, so availability rises while the initial survival curve stays the same.
It is a single asset with exponential intervals and unlimited repair resources. Wear-out, spares logistics, repair crews and scheduled maintenance are not modeled.