The Guarino Infrastructure Dependency Metric, Paper III: The Water Edge at the Tap and the Price of an Hour of Autonomy

Paper III of the Guarino Infrastructure Dependency Metric (GIDM). Preprint, revision 1, 21 September 2026. Continues Paper II (doi:10.5281/zenodo.22867011) and Paper I (doi:10.5281/zenodo.22866438). Abstract. Papers I and II of this series derived the buffer group ΠB = Ta/Tx on every edge of an infrastructure dependency graph and tested it on the February 2021 Texas chain, leaving two things open: the water edge below the tap, which Paper I could value only from a statute, and the cost of a buffer, which neither paper computed. This paper does both. With the value dimension restored to the working system, the five variables of a costed edge reduce to two groups, ΠB and a cost group ΠC = ca/(vCd), the marginal cost of an hour of autonomy over the loss rate of an hour of outage, with the rule that an edge is bought while ΠB < 1 and ΠC < 1. The water edge is read from the TCEQ boil water notice record for 2,038 community systems serving 17.3 million people, joined to EPA's inventory of every Texas community system so that the shares are statewide: about 56 percent of the state's community-system population was behind a notice and about 44 percent behind an edge that bound before the load shed order was lifted. Counted from the load shed order, the population-weighted median system issued its notice on the second day, so the statutory 24-hour buffer was a floor and realised ΠB was about 0.83 at the median; but 80 percent of the population under notice was behind an edge that bound before the order was lifted. The exposure at the tap was a population-weighted median of six days, at most weakly related to the freezing days at the nearest station, falling with system size, and outlasting the grid by a median of four days that belong to pressure recovery and sampling rather than to the storm. The hour of autonomy is then priced on each edge from the public record. On the weather edge, winterisation at the operator's own figure of about 4,400 dollars per MW for a seven-day design gives ΠC of 0.003 against the event's 9,000-dollar value of lost load, and the four published estimates, spanning three orders of magnitude, keep it between 0.0001 and 0.1. Firm gas transportation from Railroad Commission tariffs gives 0.008 to 0.03. Another day of coal stock gives 0.0001 and nothing to prevent, since the edge was slack. On the water edge the buffers are priced from projects on the Texas revolving fund list, and the cost group separates two buffers for the same edge: an hour of elevated storage must be used 53 to 340 times to pay for itself at the measured cost of a water disruption, while a standby generator at the plant pays for itself after 76 to 1,200 hours of outage over its life, a few days at a large system and a few months at the smallest. Neither pays for one event alone. The first dollar goes to the weather edge because its hour is the cheapest in the chain, the coal pile gets none, and the water plant gets a generator and not a tank, which is the ordering the post-storm rules adopted. The claims are stated as ranges with their charging conventions; the water ratios turn on a value of water the paper does not measure, so the water edge is ranked within itself and not against the power chain. Contents of the deposit. The manuscript of record (GIDM_Paper_III.pdf, two-column, 12 pages) with its LaTeX source, the abstract as printed, a metadata sheet, and a manifest with SHA-256 checksums for every file. Figures 1 and 2 in PDF and 300 dpi PNG with standalone sources. Source data as downloaded: the Tomko, Nittrouer, Sanchez-Vila and Sawyer dataset (doi:10.5281/zenodo.7447637), EPA's SDWIS Water System Summary export for Texas (active systems, 2026 Q2), TCEQ's lifted-notice and active-notice lists of 24 March 2021, and the 756 DR-4586 rows extracted from the OpenFEMA Public Assistance Funded Projects Details dataset. Three Python scripts with their outputs as run: water_edge.py reproduces every figure in Section 5, Table 1, Figure 1 and the statewide shares; cost_group.py carries the rank check for Equation 1 and every cost-group ratio in Section 6, Table 2 and Figure 2, each price a named constant with its source, under both charging conventions; tceq_lists.py parses TCEQ's two lists and recovers their printed totals. Derived results as CSV: the 2,038-system analysis set, Table 1, the Figure 1 data, the cost-group ratios, the SDWIS join, the parsed TCEQ lists and the FEMA water-distribution worksheets. Licences. The manuscript, abstract, figures, tables, manifest and derived results are released under Creative Commons Attribution 4.0 International; the code is released under the MIT licence. Third-party files are redistributed under their own terms: the Tomko dataset under CC BY 4.0; the EPA SDWIS export and the OpenFEMA extract are works of the United States Government in the public domain; the TCEQ lists are public records of the State of Texas. This product uses the Federal Emergency Management Agency's OpenFEMA API, but is not endorsed by FEMA. The Federal Government or FEMA cannot vouch for the data or analyses derived from these data after the data have been retrieved from the Agency's website(s).