Water Cost-of-Service Rate Design Calculator

How does the base-extra capacity method allocate costs?

It splits each facility cost by the peaking it was sized for, then charges each class for the peak it actually contributes.

The arithmetic is simpler than its reputation. If the system's peak day runs 1.63 times the average day, then average use needs 1/1.63 of the plant and the peak needs the rest. So a treatment plant is 61.3% base cost and 38.7% peak-day capacity. Distribution and storage are sized for the peak hour, at 3.60 times average day here, so they split three ways: 1/3.60 base, 0.63/3.60 peak day, and 1.97/3.60 peak hour — 27.8%, 17.5% and 54.7%.

Run the defaults' $23,000,000 revenue requirement through that and it lands as:

Divide each component by its units of service and you get unit costs you can apply to any class: $2.56 per thousand gallons of base cost, $248.52 per thousand gallons of peak-day capacity, $156.77 per thousand gallons of peak-hour capacity, $34.09 per account and $22.33 per equivalent meter.

The peak-capacity unit costs look enormous next to the volumetric one because they are priced per unit of daily capacity, not per unit of annual throughput. That is the whole point: a gallon delivered on the peak day draws on infrastructure that sits idle most of the year.

Which class is subsidising which, and by how much?

Not the one you would guess. On the defaults a flat rate moves $1,644,827 a year between classes — 7.2% of the entire revenue requirement — and the largest single distortion is the class nobody complains about.

Irrigation is the vivid case — a flat rate recovers only 62% of what it costs to serve, because a 5.2× peak day buys capacity it never pays for. Its own bill would have to rise 60% to cover its cost. And irrigation costs 2.34 times as much per gallon to serve as industrial despite being the smaller user.

But the biggest dollar distortion is industrial overpaying by $1,180,229 — 46% above its cost to serve. Its demand is almost flat, so it uses the plant you already built, at the hours nobody else wants it. It is the cheapest water in the system to deliver and a flat rate charges it the system average.

That matters because the instinct in a rate study is to go after the seasonal class. Hold everything else and vary the peaking:

Flattening irrigation's peak barely moves the total, because the class is 4.5% of volume. Flattening industrial's moves it five times as much. Peaking factors matter in proportion to the volume behind them, and the quiet 16% class outweighs the dramatic 4.5% one.

Set every peaking factor and system ratio to 1.0 — a system with no peaks at all — and $452,484 of subsidy remains. That residue is pure account and meter cost, which a volumetric rate cannot reach however flat demand gets.

Why does a successful conservation programme force a rate increase?

Because almost nothing in a water utility's cost falls when volume falls. The pipes, plant, storage and debt service were sized for the peak and were paid for in advance.

On the defaults only $3,200,000 of the $23,000,000 is volume-variable — chemicals, pumping power, purchased water. So a 15% demand reduction saves $480,000, which is 2.09% of the revenue requirement, while volumetric revenue falls by the full 15%. The rate has to rise 15.19% to close a $2,970,000 hole.

Look down the first and last columns. They are the same number. Cut demand by 30% and the average bill falls 4.17% — exactly the percentage the utility's cost fell.

That is not a quirk of these inputs. It is an identity, and the calculator carries it as a reconciliation check that reads zero at every input you can type. Under a volumetric rate set to recover the full requirement, total billings must equal the new requirement, so the average bill falls by precisely the share of cost that was avoidable — and how much water customers save makes no difference to it at all.

The two bounds confirm the mechanism. Make every cost volume-variable and a 15% cut needs no rate increase and gives a 15% bill reduction. Make none of it variable and the rate must rise 17.65% while the average bill does not move a cent.

None of this is an argument against conservation. Deferring a treatment plant is worth far more than the chemicals, and that avoided capital cost is real — it is simply not what an embedded cost-of-service study measures. What the model does say is that a utility planning a conservation programme should plan the rate increase in the same memo, because the two are arithmetically joined, and that a customer who conserves more than average still wins. It is the average customer who cannot.

What peaking factors should I use, and how do I know they are right?

From your own billing data, and the diversity factor tells you whether they hang together.

Class peaking factors are non-coincident — each class's own peak, whenever it happens, not its draw at the system peak. Because classes peak at different times, the class peaks must sum to more than the system peak. That ratio is the diversity factor, part of the base-extra capacity method itself; Philadelphia's rate filings put the typical range at 1.1 to 1.4. The defaults here compute to 1.243.

A diversity factor below 1.0 is not merely unusual, it is arithmetically impossible. Each class's own peak is at least as large as its draw at the moment of the system peak, so the class peaks must sum to at least the system peak. If the figure comes back under 1, your class factors and your system ratios contradict each other.

What makes that worth a dedicated flag is how quietly it fails. Flatten the class factors to 1.01 and 1.02 while leaving the system ratios at 1.63 and 3.60, and the diversity factor drops to 0.620 — but every reconciliation still reads zero, the cross-subsidy reads a perfectly plausible $452,484, and the only visible symptom is a maximum-day capacity cost of $25,484 per thousand gallons where it should be $248.52. The peak-capacity unit costs absorb the contradiction by inflating, because the full peak cost is divided across a unit base that has nearly vanished. Nothing goes unallocated, so nothing reconciles wrong.

Only at factors of exactly 1.0 do the peak units sum to zero, and then $6,904,567 lands on no class at all and the reconciliation does catch it. Everywhere in between, the model is confidently wrong. So the workbook reports a flag that reads 1 whenever the diversity factor falls below 1 — and deliberately stays quiet at a diversity of exactly 1.0 when the system ratios match the class factors, because a system whose classes all peak together genuinely has no diversity and that is a legitimate answer.

For calibration, Philadelphia's published rate-case factors run 2.00 max day and 3.60 max hour for residential, 1.80 and 2.65 for commercial, 1.60 and 2.00 for industrial, with Black & Veatch's recomputed estimates spanning 1.66 to 2.26 on max day. The defaults in this calculator are deliberately wider than that — industrial at 1.20 and irrigation at 5.20 — to make the mechanism visible in one screen. A continuous process load really can sit near 1.2, and a dedicated summer irrigation meter really can exceed 5, but neither is a default you should ship into a rate study. Pull the factors from your own monthly and peak-day records.

Worth knowing how common the alternative is. Ivins, Utah applies a single system-wide peak day factor of 2.08 to every class in its published rate study, in a service area where 55% to 58% of annual use is irrigation. That is not negligence; class-level peak data is genuinely hard to get from bi-monthly billing. But it is the assumption this calculator exists to put a number on, because applying one factor to every class makes the cost-of-service step mathematically incapable of finding the cross-subsidy it is there to find.

What does this calculator not cover?

Five boundaries, and the first is the one people most often mistake.

This is cost of service, not rate design. It tells you what each class costs to serve. It does not set your rates. Choosing the split between a fixed monthly charge and a volumetric charge, the tier breakpoints, whether to run a seasonal rate, and how fast to phase a class toward its cost — all of that is the rate design step, where policy, affordability and revenue stability enter. AWWA keeps the two separate and so should you. A class paying 46% above cost is the start of that conversation, not the end of it.

Embedded cost, not marginal cost. Every figure here allocates costs you already incur. The alternative framing asks what the next unit of demand costs: the Turvey avoided-cost method prices a demand reduction by the capital it defers, and in the Alliance for Water Efficiency's worked example a 1,000 CCF/day reduction postpones a $17.0 million plant by a year, giving a marginal capacity cost of $1.59 per CCF. That can diverge sharply from the embedded result and it is the better basis for a conservation business case. It is not modelled here.

Four classes, one year, water only. Real studies carry more classes, multi-year revenue requirement projections, and reserve and debt-coverage targets. Wastewater cost of service works on a different basis entirely — flow plus BOD and suspended-solids loading — so none of this transfers to a sewer rate.

No separate fire protection assignment. Public fire protection is folded into the distribution and storage bucket here. A full study assigns it separately, usually on equivalent hydrants or on the standby capacity reserved for fire flow, and that assignment can move several points of cost between the general classes and the municipality.

The conservation scenario is deliberately simple. It applies one uniform reduction to every class and assumes full volumetric recovery. In practice reductions land unevenly — irrigation usually falls hardest, which changes the cost allocation as well as the revenue — and most utilities recover part of their requirement through fixed charges, which blunts the revenue loss and makes the bill saving smaller still.