# Expiring audiobook allowance: reset-phase sensitivity model This is an original exploratory fluid scheduling model, not observed Spotify usage, a tested app counter, a catalogue audit or a completion-date prediction. Source rules were reviewed on 31 October 2026. Chosen allowance values are NOT a list of current plans or a universal 15-hour allocation. Actual included hours vary by country and account. Top-ups, add-ons, purchases and replay are excluded. L is normal-speed content length; H is a chosen allowance; w is chosen wall-clock hours available per elapsed day; v is chosen playback speed; C is a chosen constant cycle length; d is whole days until the first reset. The ENTIRE chosen H is unused at the book’s start, including when d=1. Daily availability is a UNIFORM CONTINUOUS FLOW, not a discrete 1-hour session. A fractional elapsed day is a model quantity, not a consumer calendar promise. Each interval consumes min(content remaining, quota remaining, w*v*interval). If complete, stop immediately. Otherwise advance to reset, expire any unused quota, replenish H and continue over intervals of length C. The first interval has length d. Active duration=L/(w*v). Elapsed duration=active+quota-induced wait. Expired unused quota is retained separately from waiting. The recipe rejects nonpositive values and d>C rather than imputing an unsupported input. Design: L=[5,10,15,20,25,30,40]; H=[7,10,15,20]; w=[1/4,1/2,1,2]; v=[3/4,1,3/2,2]; C=[28,30,31]; d=1..C. There are 1344 profiles and 39872 phase cases. Every chosen phase is retained. The profile spread is max(elapsed)-min(elapsed) across its whole-day phases. It is NOT a confidence interval or an observed consumer range. Case counts cannot estimate prevalence because no inputs have population weights. A sufficient no-quota-wait condition is H>=w*v*C. The first interval also satisfies it because d<=C and H is fully unused. Other cases can finish without waiting too, particularly for short books. The condition is not necessary. Example: L 20,H 15,w 1,v 2,C 30. At d 7, consume 14 hours before renewal and 6 after, finish in 10 modeled days. At d 1, consume 2 before renewal,15 over the next cycle and 3 after a second renewal, finishing in 32.5 modeled days. At d 30, consume 15, wait 22.5 days and finish 5 after renewal, also 32.5 modeled days. Run `python -I -S reproduce.py --out reproduced` with the Python 3 standard library. Compare phase-grid.csv, phase-sensitivity.csv, worked-example.json, worked-example.csv and summary.json byte for byte. Full source HTML is not redistributed. The model does not recommend a purchase, establish eligibility or compare monetary value.