The Oversize Audit

Every plumber learns the fixture-unit method: assign each fixture its weight from the code table, add them up, read the demand off Hunter’s curve. Roy B. Hunter built it at the National Bureau of Standards in 1940, around two assumptions that were true in the hotels and apartment houses he measured: people queue for fixtures during a morning rush, and fixtures flow hard. Both assumptions are false in a modern home, which is why the 2018 Uniform Plumbing Code added Appendix M, the peak demand method built on the Peak Water Demand Study (IAPMO Study 4-2024). What we could not find published anywhere is a clean, reproducible comparison: how far apart are the two methods on the same ordinary buildings? So we computed it, and this page is the full record.

The results

The modern figures for the starter home, the 2.5-bath home, the 12-apartment case, and the 40-unit building are IAPMO’s own published outputs (Appendix B Example 1, Table 15, and Table 16 of the Peak Water Demand Study); our engine reproduces them exactly, and that reproduction is part of the build gate this site deploys under. The 12-apartment row is the study’s own worked case: one pipe serving 12 of a 40-unit building’s apartments.

Three things the table shows

1. The gap grows with building size

The starter home is barely oversized: 10.3 GPM against 9.0 GPM, 1.15x. Hunter’s probability math was honest work, and for a handful of fixtures it still lands close. The distortion compounds with scale: by 12 apartments the old method asks for 4.47x the modern figure, and by 100 units it asks for 401.2 GPM where the modern method computes 61.3 GPM, 6.54x. The bigger the building, the worse the 1940 assumptions fit, because they were assumptions about how often fixtures overlap, and overlap is exactly what large-building sizing is about.

2. The modern method barely moves as a home grows; the old one climbs

Add a bath to the 2-bath home and the fixture-unit total climbs from 22 to 27 WSFU; add another and it hits 35.5. The old method’s demand climbs with it: 15.1, 18.1, 22.7 GPM. The modern method computes 11.0 GPM for all three, because the same family lives in the house no matter how many bathrooms it has, and the measured probability of one more rarely-used fixture running at the peak moment is tiny. A fourth bath with its own tub and shower moves it only to 11.2 GPM.

3. Both methods are published code; the difference is what they assume

This is not a vendor method against a code method. Both sides of the table are published in model plumbing codes today. The difference is that one extrapolates 1940 hotel-rush measurements and 1940 flow rates, and the other uses probabilities of use measured in real modern households at today’s water-conserving flow rates. Oversized pipe is not a free safety margin, either: it costs more to install, it makes the hot water take longer to arrive, and it holds more stagnant water.

Methods, in full

The corpus. Eight fixed configurations, committed as scripts/oversize-corpus.json in this site’s repository: five single-family homes from one bath to four (built by adding rooms to the published 6-fixture example house), and three multi-family cases built from the Peak Water Demand Study’s own 2.5-bath apartment archetype at 12, 40, and 100 units. Every fixture count and every flow rate is pinned explicitly in the file, so the study does not depend on any tool’s defaults. Four configurations are deliberately identical to IAPMO’s published worked examples, so the modern side of those rows is a published number, not our output: the starter home (Appendix B, Example 1: 9.0 GPM), the 2.5-bath home (Table 15, service line: 11.0 GPM), the 12-apartment case (Table 16, h = 12: 20.1 GPM), and the 40-unit building (Table 16, h = 40: 35.8 GPM). The comparison point everywhere is whole-building service demand.

The modern side. The peak demand method of UPC Appendix M (2018–2024 Uniform Plumbing Code), per the Peak Water Demand Study (Buchberger, Omaghomi, Wolfe, Hewitt, and Cole; IAPMO Study 4-2024, first published 2017). We used the same engine that runs our free pipe sizing tool; it is validated against twelve published IAPMO worked examples by a script that gates every deploy of this site, so the math that produced this page is the math that is tested.

The legacy side. Two published inputs, cited exactly, accessed August 20, 2026. First, the fixture weights: water supply fixture units from the 2024 Uniform Plumbing Code, Table 610.3, private column, read verbatim in IAPMO’s read-only 2024 UPC reader (Appendix A repeats the same values as Table A 103.1; the private values for our eleven fixture types are unchanged since at least the 2016 printing). Second, the conversion from fixture units to demand: the UPC publishes it only as a curve, Chart A 103.1(1) and A 103.1(2), the flush-tank curve No. 2, so the numeric values come from the two published tabulations of that same curve. Below 100 fixture units we used Hunter’s own published curve readings, from National Bureau of Standards Report BMS 79 (1941), Table 10, one of the two reports UPC Appendix A itself cites as its basis. At and above 100 fixture units we used the 2021 International Plumbing Code’s Appendix E, Table E103.3(3), flush-tank column, which coincides with the curve there. Between published points we interpolated linearly, and we state that plainly.

Where the published sources disagree, we took the lower one. Below roughly 50 fixture units the IPC’s tabulation runs a few GPM above the curve the UPC actually prints; at 25 fixture units the IPC table says 21.5 GPM where Hunter’s own reading of his curve says 17. Using Hunter’s values there makes the single-family legacy demands, and therefore the single-family oversize ratios, smaller. We would rather understate our own headline than let a table quirk inflate it; under the IPC tabulation the starter home’s ratio would be 1.92x instead of 1.15x.

Worked example, end to end (the starter home). One lavatory (1.0) + one bath/shower (4.0) + one water closet, gravity tank (2.5) + kitchen sink (1.5) + dishwasher (1.5) + clothes washer (4.0):

WSFU = 1.0 + 4.0 + 2.5 + 1.5 + 1.5 + 4.0 = 14.5
demand: between the published points (11 FU, 8 GPM) and (17 FU, 12 GPM)
= 8 + (14.5 - 11) / (17 - 11) × 4 = 10.3 GPM
modern method (published, Example 1) = 9.0 GPM
oversize = 10.3 / 9.0 = 1.15x

Reproducing every figure. One command, node scripts/generate-oversize-study.mjs, reads the corpus, runs the live engine for the modern side, applies the legacy tables, and writes the committed results file this page renders from. A second script, scripts/validate-oversize-study.mjs, regenerates the results and fails the build on any byte of drift, checks the four published-case alignments, and checks that every surface quoting these figures quotes them exactly. No number on this page is hand-typed.

Limitations, honestly stated

• Fixed flow rates. Every configuration pins the published example flows (bath/shower 5.5 GPM, kitchen 2.2, and so on). Different fixtures move both sides of the table; the ratios reported here are for these pinned configurations, not for every possible house.
• Whole-building service demand only. We compared one number per building: the demand at the service. Branch-level sizing, hot and cold splits, hose bibbs, irrigation, and every continuous flow sit outside this comparison.
• Jurisdictions differ, and this is not sizing advice. Fixture-unit sizing is the enforceable method in many places, UPC Appendix M is an optional appendix adopted jurisdiction by jurisdiction, and permit calculations may be required to come from the code-referenced software. Nothing here tells anyone to downsize a pipe.
• Fixture-unit tables vary by code family and edition. We used the 2024 UPC’s private-installation values. The IPC family assigns different weights to some of the same fixtures, so an IPC-table version of this study would land on somewhat different ratios.
• The two published tabulations of Hunter’s curve disagree below about 50 fixture units. We used the lower, as described in Methods; the single-family ratios are conservative for that reason. The multi-family rows are unaffected because the tabulations coincide there.
• The 12-apartment row is a branch case. It reproduces the study’s own Table 16 example, one pipe serving 12 of a 40-unit building’s apartments, so it is a riser or wing rather than a standalone building service.

Common questions

How much bigger are Hunter’s curve demand figures than the modern method’s? Across the eight typical configurations we computed in August 2026, the 1940 fixture-unit method specifies 1.15x to 6.54x the flow the modern UPC Appendix M method computes, median 2.23x. The gap grows with building size: 1.15x for a one-bath starter home, about 2x for a 3-bath home, and 6.54x for a 100-unit apartment building.

Why does the fixture-unit method overshoot in residential buildings? Hunter built the method in 1940 around congested use, one person waiting behind another at hotels and apartment houses during the morning rush, and around fixtures that flowed much harder than today’s. Both assumptions fail in a modern home, and the failure compounds as fixture counts grow, which is why the ratio climbs from 1.15x for a starter home to 6.54x for a 100-unit building. The modern method replaces the assumptions with probabilities of use measured in real households.

Does this mean fixture-unit sizing is wrong to use? No. It is the published, enforceable method in many jurisdictions, and pipe sizing is the local authority’s call, not this study’s. The study measures the difference between the two published methods; it is not advice to downsize anything. UPC Appendix M is an optional appendix, adopted jurisdiction by jurisdiction, and permit submittals may require calculations from the code-referenced software.

Where do the study’s fixture-unit numbers come from? From the published codes only, with each value cited: water supply fixture units from the 2024 Uniform Plumbing Code’s Table 610.3 (private column, verified in IAPMO’s read-only 2024 UPC reader in August 2026), and the flush-tank demand conversion from Hunter’s own published curve readings (NBS Report BMS 79, 1941, Table 10) below 100 fixture units and the 2021 International Plumbing Code’s Table E103.3(3) at and above 100, where the two tabulations coincide. Values between published points are linear interpolations, and where the two published sources disagree the study uses the lower one, so its single-family ratios are understated rather than overstated.

What if bathrooms are counted as combined fixture-unit groups instead of individual fixtures? The table this study uses does not offer that option: the 2024 UPC’s Table 610.3 lists every fixture individually and publishes no combined bathroom-group value. Bathroom groups are a feature of the IPC family’s tables, where Table E103.3(2) assigns a private flush-tank bath group 3.6 fixture units against 4.3 for the same three fixtures summed. We recomputed all eight configurations that way, full baths as groups on the IPC’s own values and its own flush-tank conversion, and the effect runs in both directions: the starter home rises from 1.15x to 1.40x, because below about 50 fixture units the IPC’s conversion runs above Hunter’s own readings and this study deliberately used the lower, while the 100-unit building falls from 6.54x to 4.51x. Under the bathroom-group treatment the range is roughly 1.4x to 4.5x rather than 1.15x to 6.54x; the gap still grows with building size and stays a multiple, not a percentage.

Why we did this (and what BeepBack is)

We build BeepBack: it texts a plumber’s missed callers back in seconds, asks the few questions the plumber would ask, and hands over a sorted, ready-to-call lead. We built the pipe sizing tool for the trade that uses us most, and this study exists because the tool kept producing numbers far below what the fixture-unit method taught us all to expect. Rather than assert the gap, we measured it, published the method, and made every figure reproducible from one command.