Sietch Cave-System Engineering: A Moisture-Exchange and Thermal Model of Sietch Tabr and Four Registered Sietches, Validated Against Chamber Logs and Tracer Measurements of 10196–10236 AG

1. Introduction

A sietch is a Fremen cave community cut into the rock ridges of Arrakis, sealed against the surface and organised around the hoarding of water. Accounts of sietch life credit the rock with its unmoving temperature, and much of that credit is deserved: rock walls absorb and return heat on a daily scale, and thick overburden excludes the conducted surface wave. Yet large sietches such as Sietch Tabr also run windtraps, dew precipitators, stills and workshops. What remains open is how much the engineered features add to the rock's own damping, and how the work of keeping water divides between structure and machinery.

In a reserve model of twenty-three registered sietches, published in this journal, the operating margin of the median sietch depends on ration compliance and on collector yield, which moves the compliance threshold by several points per 10% change. Vapour that escapes the habitable volume never reaches the operating cisterns, and a climate-chamber study of Fremen adults has measured how much occupants give off at rest.

We separate three interventions, entry sealing, chamber geometry and moisture capture, and two outcomes: the diurnal temperature amplitude of occupied chambers, and structural water loss in litres per day. The ordering tested is that seals and geometry govern stability while capture governs recovery. Writing from the Arrakeen Planetary Ecology Institute in 10243 AG, we draw on Fremen Planetological Survey chamber logs of 10196–10232 AG and a tracer campaign completed in 10236 AG.

2. System Description

Sietch Tabr houses a census population of about 1,120. Its occupied chambers lie beneath 12 to 40 m of rock and are reached through entrances closed by door seals, the moisture-tight hatches that Fremen close behind every passage. Survey plans give each chamber a count of seal stages to open air, from one to three, and a gallery path length to the nearest opening, from about 15 m to 140 m. Deep chambers at the end of long, dead-end galleries are the geometric intervention; shallow chambers near an entrance approximate the unmodified cave.

Ventilation cannot be eliminated, since breath and cooking foul still air, so the sietch draws a controlled exchange through vent shafts to the ridge. Dew precipitators in the inner galleries condense vapour from chamber air and drain to the operating cisterns, which also receive ridge windtrap output; windtraps on the vent outlets strip part of the vapour from outgoing air. The hidden catchbasins of the greening cache are fed only through the standing tithe of about 12% of collector yield recorded in the reserve study, which applies to recovered breath water as to any other collector output.

The chamber study cited above recorded a resting water loss of about 123 mL/h in Fremen adults at 38°C and 15% relative humidity. Interior air is cooler and wetter, so we take 25 mL/h (range 15–40) per occupant, or 672 L/day for Tabr. Cooking, open basins and workshops add an estimated 288 L/day, for a total generation of 960 L/day.

3. Analysis / Model

Vapour obeys a two-sink balance. Generation G is removed by exchange with the surface at a rate E, in air changes per day, or by precipitator recovery at a rate R, so the escaping fraction is E/(E + R). Outlet windtraps recover a share η of that stream, giving net loss G × E/(E + R) × (1 − η). Exchange is entry leakage, set by seal stages, plus vent exchange, set by path length. Calibrated on Tabr, leakage is 0.12 per day, vent exchange 0.30, R 1.25 and η 0.30.

Heat follows a different path. In Tabr rock the daily wave decays by a factor of e within roughly 0.2 m, so overburden ceases to matter once it exceeds a metre or so. The residual swing is carried in by exchanged air and absorbed by walls that act, on the daily scale, as an unlimited heat store. With wall coupling h of about 20 per day, interior amplitude is surface amplitude times E/√((E + h)² + ω²), where ω is 2π per local day. Capture does not enter.

Each intervention was removed in turn. Unmodified geometry raises vent exchange to 0.95 per day, the Survey mean for chambers within 20 m of an opening; removing capture sets R and η to zero. The seal counterfactual is the least secure. Hygrometer-decay records from seven episodes in which a Tabr entrance was held open for work parties give leakage of 1.30 per day (SD 0.34), our base case. That implies a steep step from no seal to one, whereas the regression in Section 4 prices each additional stage at only 0.14 per day. A first hatch that stops bulk flow may be worth far more than later stages, but no unsealed chamber was measured. We therefore also report a linear case, with unsealed leakage of 0.40 per day extrapolated along the regression slope from Tabr's typical two stages. Contributions were apportioned by Shapley decomposition over all eight on–off combinations.

As built, Tabr loses about 169 L of vapour a day (range 120–230 under parameter uncertainty), 7.5% of a daily requirement of 2,240 L at the reserve model's ration of 2.0 L per person-day. Deep chambers are predicted to swing 0.75°C (0.5–1.1°C) against about 38°C at the surface. Wall storage does most of this: with all three interventions removed the interior still holds to 3.70°C, about 92% of the total reduction. Of the remaining 2.95°C, seals account for 1.90°C (64%) and geometry for 1.05°C (36%) in the base case; in the linear case the gain is 1.56°C, split 30% to seals and 70% to geometry. Of the 791 L/day of loss avoided, capture accounts for 645 L/day (82%), sealing 90 L/day (11%) and geometry 56 L/day (7%); in the linear case capture's share is 88%. A 10% cut in exchange saves about 13 L/day, a 10% gain in recovery about 12 L/day.

4. Validation Against Field Data

Between 10233 and 10236 AG an Ixian team released an inert tracer during low-occupancy intervals in 64 chambers, 22 at Tabr and 14, 11, 9 and 8 at the registered sietches S-7, S-12, S-3 and S-15, and timed its decay (Marn, 10236 AG). Exchange averaged 0.46 per day (SD 0.29); the nine chambers within 20 m of an entrance averaged 1.02 (SD 0.19), close to the modelled 1.07. Calibrated on Tabr alone (r² 0.79), the model predicted exchange in the 42 chambers of the other sietches with r² 0.71 and a root-mean-square error of 0.11 per day.

A linear mixed-effects regression of measured exchange, with sietch as a random intercept, tested the model's structure. Each additional seal stage lowered exchange by 0.14 per day (95% CI −0.19 to −0.09; p < 0.001), and each further 10 m of path by 0.05 (95% CI −0.07 to −0.03; p < 0.001), a slope that reproduces the 0.65 separating the geometry counterfactual from the as-built value. Overburden had no detectable effect (β −0.02 per 10 m; 95% CI −0.05 to 0.01; p = 0.19), as expected from an exchange model in which only seals and path set the rate. An outlet windtrap made no difference (β +0.03; 95% CI −0.04 to 0.10; p = 0.40), compatible with capture acting on what leaves without changing the rate of leaving. Dominance analysis of the marginal R² of 0.61 gave seal stages 0.34, path 0.22, overburden 0.04 and windtrap 0.01. With five sietches the intraclass correlation of 0.08 is only indicative; a fixed-effect refit moved neither slope by more than 0.01.

Survey temperature logs gave the second test (Fremen Planetological Survey, 10196–10232 AG). Deep Tabr chambers showed a median amplitude of 0.8°C (interquartile range 0.6–1.1°C) against the predicted 0.75°C. Near-entrance chambers did not fit: given their measured exchange the model predicts about 1.8°C, below the lower quartile of the logged median of 2.9°C (2.2–3.8°C), and matching the logs would need exchange near 1.7 per day. Tracer runs avoided the working hours and storm seasons in which seal traffic admits surface air in bursts, and rock near a portal may not behave as an unlimited store. We cannot separate these, and confine the thermal validation to deep chambers. Across all 64 chambers it explained 66% of the variance. The ridge station recorded a surface amplitude near 38°C, ranging 29–46°C by season (Straka, 10229 AG).

If the logged 2.9°C represents unmodified geometry, geometry's share of the stability gain becomes 52% in the base case and 82% in the linear case, so the seal share spans 18–64% across assumptions. The data rank neither above the other; they do establish the dominance of wall storage and the absence of any thermal role for capture.

For the water ledger, the model predicts precipitator returns of about 719 L/day and outlet windtrap returns of 72 L/day, 791 L/day in all. Survey ledgers of operating-cistern returns from these devices, entered before the tithe, run at 690–770 L/day in seasonal means, an overprediction of roughly 3–15% that cistern evaporation and unmetered draw could explain.

5. Failure Modes

Survey logs record seventeen door-seal breaches at Tabr between 10196 and 10232 AG, repaired within a median of two days (range one to six). The whole-sietch seals-removed configuration gives an upper bound of 208 L/day extra loss and a deep-chamber amplitude of 2.7°C. Survey plans show five sealed entrances; if leakage divides evenly, one breach adds about 62 L/day and lifts deep chambers to about 1.2°C. A median breach costs 120–420 L, under 0.1% of Tabr's annual requirement of about 818,000 L.

Coriolis storms matter more. Pressure pumping as a front crosses the ridge roughly triples exchange for one to four days, while blown sand fouls outlet windtraps (Vashtel, 10212 AG). With exchange at 1.26 per day and no outlet capture, loss reaches about 482 L/day, 313 L/day above normal and 21.5% of daily requirement, in the same event that cuts windtrap yield.

Discovery is a failure of another order, since vent outlets and windtraps are the only surface features a sietch cannot hide. Oral accounts describe vents stopped with sand during Harkonnen patrol seasons before 10191 AG (Sietch Tabr Council of Elders, 10214–10231 AG). Stopping vents lowers water loss at a cost in air quality, and since sietch locations were withheld from Imperial surveys then (Office of the Imperial Planetologist, 10185–10191 AG), that regime has no instrumental data.

6. Conclusion

Our results order the interventions as follows. For stability the rock does most of the work, damping a 38°C surface swing to under 4°C before any intervention is counted. Seals and dead-end placement together bring deep chambers below 1°C; how that remainder divides between them rests on an untested seal counterfactual and a thermal miss near entrances, and we do not rank them. Capture adds nothing to stability but carries 82–88% of the avoided water loss, because at steady state every litre not recovered eventually escapes.

A structural loss near 7.5% of requirement is not trivial, and storms raise it roughly threefold. Recovered breath water enters the operating cisterns less the tithe, so retention failures appear in operating accounts as reduced yield, the quantity to which the compliance threshold is most sensitive, while the cache is touched only through its fixed share. Greening-programme outposts would gain from overburden only up to the first metre or so.

Four limits apply. Occupant generation is extrapolated from a chamber study at higher temperature. Tracer data cover five sietches, and only Tabr has long ledgers. The first seal stage rests on seven work-party episodes. The model omits air quality, and so cannot price the ventilation that sietches must keep.