STAGE 2 · EXPLORATORY ARCHIVE ANALYSIS · VERSION 1
Shorter recorded dryer runtimes.
A question still to validate.
Analysis snapshot: September 6, 2026. Measurement period: August 18–September 5, 2026. This public edition summarizes an existing archive analysis, not a new controlled experiment or peer-reviewed paper.
Dataset and accounting
14 ZIP exports contained 116 exported run records. Removing 14 exact repeated records left 102 unique runs: 49 classified for this public report as shake-dry, 47 as direct-dry, and 6 invalid measurement. The 96 valid reviewed runs enter comparisons; invalid runs are excluded.
Method
Compare conditions within archive/measurement-date sets, a provisional proxy rather than verified person/dryer identity. Do not pool individual runtimes across sets. Exact duplicate start/stop/runtime signatures were counted once and their canonical labels agreed. Legacy review labels were normalized; comments were not parsed as reviews. Public condition names are shake-dry and direct-dry; the private source labels and measurements are unchanged.
Within-set median reduction (%) = 100 × (direct-dry median − shake-dry median) / direct-dry median. Medians were checked using integer-millisecond input. Displayed seconds use half-up rounding; differences are computed before rounding displayed medians.
Per-set observations
Anonymous set IDs preserve the source analysis order. Exact timestamps, export filenames, raw recordings, and private records are not included in this public edition. These IDs do not establish independent participants.
| Set | n | Shake-dry median | direct-dry median | Difference | Reduction | Invalid | Caution |
|---|---|---|---|---|---|---|---|
| S01 | 5 / 5 | 7.880 | 10.880 | 3.000 | 27.57% | 0 | — |
| S02 | 3 / 1 | 8.518 | 15.942 | 7.424 | 46.57% | 0 | Small sample |
| S03 | 3 / 3 | 12.093 | 19.400 | 7.307 | 37.66% | 0 | — |
| S04 | 4 / 5 | 7.270 | 11.399 | 4.129 | 36.22% | 1 | — |
| S05 | 5 / 6 | 11.000 | 16.620 | 5.620 | 33.81% | 0 | — |
| S06 | 2 / 2 | 12.960 | 22.648 | 9.688 | 42.78% | 0 | Small sample |
| S07 | 1 / 1 | 9.081 | 15.860 | 6.779 | 42.74% | 0 | Small sample |
| S08 | 2 / 2 | 10.400 | 14.691 | 4.291 | 29.21% | 0 | Small sample |
| S09 | 1 / 1 | 12.299 | 14.140 | 1.841 | 13.02% | 0 | Small sample |
| S10 | 2 / 2 | 7.530 | 12.442 | 4.912 | 39.48% | 0 | Small sample |
| S11 | 4 / 4 | 5.861 | 11.461 | 5.600 | 48.86% | 2 | — |
| S12 | 6 / 6 | 10.358 | 18.459 | 8.102 | 43.89% | 1 | — |
| S13 | 5 / 4 | 7.419 | 9.819 | 2.400 | 24.44% | 1 | — |
| S14 | 1 / 1 | 8.378 | 18.119 | 9.741 | 53.76% | 0 | Small sample |
| S15 | 5 / 4 | 7.180 | 10.809 | 3.629 | 33.57% | 1 | — |
Secondary energy scenarios — estimated savings
The equal-weight average of the 15 within-set arithmetic-mean differences is 5.646 seconds. This is not an average of the median percentages and is not a population estimate. The ratio of the equal-set mean runtimes corresponds to 38.02% lower runtime. It does not demonstrate that percentage of real-world energy savings.
Illustrative energy difference (Wh/use) = assumed power (W) × mean runtime difference (s) / 3600. At an assumed 1,500 W, this yields 2.353 Wh/use. The power is neither measured nor an established fleet average.
- 500 W assumed: 0.784 Wh/use.
- 1,000 W assumed: 1.568 Wh/use.
- 1,500 W assumed: 2.353 Wh/use.
- 2,000 W assumed: 3.137 Wh/use.
- 2,300 W assumed: 3.607 Wh/use.
Small-sample sensitivity: Restricting to the eight sets with at least three runs per condition gives a mean difference of 4.967 seconds. This filter does not resolve the missing controls or establish independence.