Independent Calculations

Every derived figure in the Title 15 Amendment Analysis traces to a numbered calculation here. These are my computations, not sourced facts. Inputs are cited; method is shown; substitute your own assumptions and see whether the conclusion survives.

This is a working document, published as filed. It is the arithmetic behind the report, not a second draft of the report. Where it disagrees with itself, argues with an earlier revision, or reaches a finding that cuts against the report's framing, that is left visible on purpose. A calculation you can only check after it has been tidied is not one you can check.

How to read the notation. C-nn identifies an individual claim in the public record and P-nn a slide in the applicant's presentation deck; both are internal identifiers used to keep each claim traceable to where it was made. CALC-n numbers the calculations here. I-n numbers the shared inputs below.

Two capacity bases appear in this document, and the difference matters. The governing figure is the permit's 393.25 MW nameplate, 1,210 Bloom units. An earlier revision computed against 250 MW, which is the rack-level compute load rather than the generation the fuel cells must supply. CALC-2 and CALC-13 were corrected to the permitted nameplate. CALC-3, CALC-7, CALC-8 and parts of CALC-9 and CALC-12 still run on the 250 MW basis and are therefore floors, not ceilings: the true figures are higher in proportion. Each is flagged where it appears. Nothing in the report rests on a 250 MW figure where the 393.25 MW figure would change the conclusion.

Built: 14 AUG 2026 · Revised and re-verified 15 AUG 2026 against the three governing primary sources: the DAQ Intent to Approve DAQE-IN163550001-26, the Bloom ES 6.5 datasheet 1016932-20260204, and UMPA's FY2022 Integrated Resource Plan. All shared inputs below are resolved and no parametric placeholders remain in the load-bearing calculations. CALC-11, on water, remains parametric and is labelled as such.

Shared inputs

CALC-1: Method validation against a real Utah permit

Before relying on the CO2 arithmetic below, test the method on a facility whose CO2e a regulator has already published.

Novva West Jordan: 72 Jenbacher gas engines, 235.6 MW, permitted 997,261 tpy CO2e.

Implied intensity, assuming PTE at ~100% capacity factor:

997,261 t ÷ (235.6 MW × 8,760 h) = 997,261 ÷ 2,063,856 MWh = 0.483 t CO2/MWh

Now predict it independently. Gas reciprocating engine at ~42% LHV efficiency:

HR_LHV  = 3,412 / 0.42          = 8,124 Btu/kWh
HR_HHV  = 8,124 / 0.902         = 9,006 Btu/kWh
CO2     = 9,006 × 53.06 / 1e6   = 0.478 t CO2/MWh

0.478 predicted vs. 0.483 permitted, within 1%. The method is sound. Proceeding.

This validation is genuine and worth distinguishing from a weaker one. Novva's 997,261 tpy is a regulator-issued permit figure, derived independently of the arithmetic here, so agreeing with it to 1% is real corroboration. A separate check against Bloom's own datasheet is not. That comparison uses the same stoichiometry and the same efficiency range as the calculation it validates, so close agreement is arithmetically guaranteed and proves nothing. Where this review cites "agreement within 0.2%" with Bloom's datasheet, read it as an internal consistency check, not as independent confirmation.

CALC-2: Facility CO2, the central quantitative finding

BASIS CORRECTED 15 AUG 2026. Earlier revisions computed against 250 MW, which is the rack-level compute load, not the generation the fuel cells must supply. Everything below now uses 400 MW of Bloom Energy Server 6.5 capacity, and every input is printed on the datasheet (1016932-20260204) rather than assumed.

Inputs, all from the datasheet

Bloom publishes the CO2 figure directly. It is not derived here.

Internal consistency check on the datasheet itself

Bloom prints heat rate and CO2. They must reconcile, and they do:

5,811 Btu/kWh x 53.06 kg CO2/MMBtu = 308.3 kg/MWh = 679.8 lb/MWh   (printed: 308 / 679)
7,127 Btu/kWh x 53.06 kg CO2/MMBtu = 378.2 kg/MWh = 833.7 lb/MWh   (printed: 378 / 833)

Agreement to 0.1% on both ends. Two independently printed figures reconcile through EPA's standard natural-gas carbon factor, so the carbon number is not marketing-adjusted. It falls out of the fuel.

Why no fuel cell escapes this. A solid-oxide fuel cell fed methane produces CO2 as its primary carbon product:

CH4 + H2O  ->  CO + 3H2      (reforming)
CO + H2O   ->  CO2 + H2      (water-gas shift)
H2 + O2-   ->  H2O + 2e-     (anode)

Every carbon atom entering as methane leaves as CO2, at a fixed 2.743 kg CO2 per kg CH4. Suppressing combustion eliminates thermal NOx. It does not eliminate carbon. This is why the carbon figure is a property of the fuel rather than of the equipment, and why it is stated here rather than derived.

Result at the permitted 393.25 MW

BASIS RE-CORRECTED 15 AUG 2026. Earlier revisions used a rounded 400 MW. The governing number is the permit's own 393.25 MW, 1,210 units at 325 kW (DAQE-IN163550001-26, S1).

Central estimate: ~886,000 metric tons CO2/yr at 393.25 MW and 75% utilization.

The permit's published 1,434,855 short TPY equals 1,301,679 metric tons, which is the 100% row's end-of-life figure. The reconstruction closes on the agency's own number.

(Prior figure at the 250 MW rack load was ~563,000 t; the 400 MW working figure gave ~901,000 t.)

Testing "three or four diesel trucks" (C-34)

A Class 8 truck at 100,000 mi/yr and 6.46 mpg burns 15,480 gal x 11.08 kg CO2/gal = 171.5 t/yr. Three to four trucks is 515-686 t.

permit basis, 1,301,679 t metric / 686 t = 1,897x
permit basis, 1,301,679 t metric / 515 t = 2,528x
at 75% central, 886,193 t / 686 t        = 1,292x
at 75% central, 886,193 t / 515 t        = 1,721x

Wrong by three orders of magnitude on total CO2. C-34 is FALSE on that reading. It survives only on the criteria-pollutant CO2e reading, which excludes CO2 by design. See CALC-5.

The framing that is fair to Volition

Per unit of compute the intensity is unchanged by the load correction, and it is genuinely good:

~29% cleaner than on-site gas engines and 43% cleaner than the Utah grid, per unit of compute. Both things stay true: the facility adds roughly 886,000 tons of CO2 a year to Utah County that does not exist today, and it is the lowest-carbon way to run that compute among what actually gets built in Utah.

The Utah grid figure is now VERIFIED at 0.601 t/MWh, computed from EIA's own filed data for 2024: total electric power industry CO2 of 21,120,098 metric tons against net generation of 35,133,906 MWh. The earlier 0.635 working figure was 5.3% high and is superseded.

CALC-3: Upstream methane

CO2 at the stack is not the whole greenhouse footprint. Gas consumed at η = 60%:

Heat input = 1,642,500 MWh × 6,304 Btu/kWh × 1,000 = 1.036e13 Btu = 10.36 million MMBtu
Volume     = 1.036e13 ÷ 1,036 Btu/scf ≈ 10.0 Bcf/yr
CH4 mass   = 10.0e9 scf × 0.01924 kg/scf × 0.95 CH4 ≈ 182,800 t CH4 throughput

Total CO2e range: ~625,000 t (low leak, 100-yr) to ~900,000 t (high leak, 20-yr).

Note the same upstream penalty applies to the gas-engine and (partly) grid comparators in CALC-2, so the relative ranking is largely unchanged. It matters for the absolute number.

CALC-4: The gas load in context

10.0 Bcf/yr ÷ 365 ≈ 27.4 MMscf/day

For scale: this single facility's annual gas draw is on the order of several percent of Utah's total statewide natural gas consumption. That figure needs EIA verification before it appears in the report, flagged as outstanding, but if it holds it is a materially more informative way to express C-17 ("zero impact on existing gas customers") than anything in the record. The claim may well be true for the distribution system, given a dedicated lateral, while still being a significant new call on regional supply. Those are different questions and the record conflates them.

CALC-5: The "4–5 semi trucks" test

Baseline note. The 04 AUG Council deck and the 05 AUG Planning Commission deck state different figures for the same existing truck fleet. The Council deck gives ≈2 metric tons NOx and ≈50 kg PM2.5; the PC deck gives 101 kg and 4.2 kg. The Council figures govern, they are internally consistent (4.76 g NOx/mi, 0.119 g PM2.5/mi across 420,000 truck-miles), while the PC figures imply 0.24 g NOx/mi, cleaner per mile than many passenger cars. The 04 AUG Council deck reverts to the Council figures.

Tests C-31 ("Equal to 4–5 Semi Trucks per year", P-07).

Inputs

Facility (P-08):     5.17 short tons NOx/yr  = 4,690 kg
                     0.12 short tons PM2.5/yr =   109 kg
Deck fleet figure (CC-11): 420,000 truck-mi/yr, 65,000 gal diesel  [basis unstated]
                     2,000 kg NOx/yr  ·  50 kg PM2.5/yr
Implied fleet rates:  4.76 g NOx/mi  ·  0.119 g PM2.5/mi   [both plausible for Class 8]

Test A, facility against the fleet figure the deck itself states

NOx:    4,690 ÷ 2,000 =  2.3×
PM2.5:    109 ÷    50 =  2.2×

Test B, facility against 5 individual trucks, at 100,000 mi/truck-year and the fleet's own per-mile rates:

NOx per truck-year:   4.76 g/mi × 100,000 mi = 476 kg
                      5 trucks = 2,380 kg    →  4,690 ÷ 2,380 = 1.97×
PM2.5 per truck-year: 0.119 g/mi × 100,000 mi = 11.9 kg
                      5 trucks =   59.5 kg   →    109 ÷ 59.5 = 1.83×

The two pollutants agree independently. Both tests land at roughly 2×.

⚠ But Test B rests on an assumption that appears in no document. The 100,000 mi/truck-year figure is an assumption of this analysis, not Volition's. On the deck's own stated basis, 3,500 trips/month over a 10-mile round trip, a truck at this site drives ~120 miles/year, and 4–5 of them emit a few hundred grams of NOx against the facility's 4,690 kg. On that reading the claim fails by two orders of magnitude.

The two readings differ by ~40× and the slide states no basis, so neither can be called the intended one. The defensible finding is the original one: the claim is unfalsifiable as written. Test A, at 2.2–2.3×, uses only the deck's own numbers. Do not quote "off by about 2×" as if it were settled, and do not read Test A as a site comparison; see the basis warning below.

Verdict on C-31: UNFALSIFIABLE AS WRITTEN. The slide states no basis, and the two available readings differ by roughly 40×.

⚠ What Test A is not. Test A measures the slide against the deck's own stated fleet figure. It is an internal-consistency test of the presentation, not a comparison between the data center and this site. The deck never says what its 420,000 truck-mile figure represents, and it cannot be assumed to describe the fully tenanted warehouse. The building is mostly vacant today and its present truck traffic is very low, so no baseline drawn from current operations is relevant to the decision in front of the Council.

The comparison that matters is the proposed data center against the fully tenanted distribution warehouse the building was built for, and that one is computed from ITE rates against the verified 1,054,000 sq ft rather than taken from the applicant: 6,396 to 12,849 diesel trip ends a month for the warehouse, none for the data center. That calculation, not this one, is the truck finding the report relies on.

What survives as a finding: the PC deck remains internally inconsistent, its own 101 kg fleet figure cannot be reconciled with its own "4–5 semi trucks" claim two slides later. But that is a data error in the PC deck's baseline, not evidence the truck claim is wildly false. Note the direction: understating the fleet's NOx makes each truck look cleaner, which makes "4–5 trucks" a smaller quantity, which makes the equivalence harder to satisfy. The error works against Volition's own argument, the signature of a hasty overnight revision, not a manipulation, and it should be reported that way.

Separately, and to Volition's credit, a dedicated retrieval pass found the absolute NOx figure is well supported. The Delaware Red Lion Bloom permit implies ~0.0071 tpy NOx/MW → ~1.8 tpy at 250 MW. Volition claims 5.17, roughly 2.9× more conservative than a regulator-issued permit for the same technology.

Caveat that cuts the other way, and it is the one that matters for permitting. The 5.17 tpy figure is a stated estimate, not a permit limit. Bloom's CARB-certified NOx ceiling of 0.07 lb/MWh at 250 MW and an 85% capacity factor yields ~57.5 tons/yr, above Utah's 50 tpy R307-421 Utah County offset trigger. §15.3.24 contains no potential-to-emit cap, §7 requires only that DAQ approvals be obtained and maintained. The gap between 5.17 estimated and 57.5 permitted-worst-case is the whole reason a PTE cap in the code would matter.

CALC-6: The emissions comparison table audit (C-32)

P-08 sets Volition's 5.17 tpy NOx against gas turbine ~5,400, recip engine ~59,700, diesel generator 10,200–29,500, and Utah grid average ~2,600, with no capacity, capacity factor, or basis of equivalence stated anywhere on the slide.

Test the recip-engine row against a real permit. Novva: 235.6 MW of gas recip engines, permitted 51.28 tpy NOx. Per MW: 0.218 tpy/MW. Scaled to 250 MW: 54 tpy.

The slide says ~59,700 tpy for a natural gas reciprocating engine. That is ~1,100× higher than what Utah actually permits for 250 MW of exactly that equipment.

Reverse-engineering what would produce 59,700 tpy: uncontrolled lean-burn gas engines emit roughly 2–3 g NOx/bhp-hr versus Novva's SCR-controlled 0.0152 g/bhp-hr, a factor of ~130–200. Applying that to 54 tpy gives ~7,000–11,000 tpy, still far short of 59,700. To reach 59,700 requires either a much larger assumed capacity, an uncontrolled engine at high load hours, or an error.

Verdict on C-32: the table is not a like-for-like comparison and at least one row appears wrong by roughly three orders of magnitude. The "Utah grid average" row is worse than wrong , it has no stated basis at all, so it cannot mean anything. A reader is invited to conclude Volition is ~11,500× cleaner than a gas engine when the defensible figure, against Utah's own permitted comparable, is roughly 10×.

This matters more than it might appear. The true comparison, ~10× cleaner than the facility actually built in West Jordan, ~45× cleaner than the one permitted in Millard County , is a genuinely strong argument. Inflating it to 11,500× replaces a strong verifiable claim with an indefensible one, and hands critics the easiest attack in the record.

CALC-7: Thermal balance and the heat flux claim

Total heat rejection

At steady state, essentially all fuel energy entering the site leaves as heat. The only meaningful energy export is data transmission, which is negligible. A public commenter's argument (C-69, C-144) is thermodynamically correct.

At η = 60%, 250 MW electrical:

Fuel input        = 250 / 0.60      = 417 MW-thermal
Electrical output = 250 MW → becomes heat in the compute halls
Fuel cell waste   = 417 − 250       = 167 MW-thermal
TOTAL REJECTED    ≈ 417 MW-thermal

Heat recovery (C-66) does not change this total. Recovered heat driving absorption chillers still rejects to atmosphere; it displaces electrical parasitic load, improving efficiency, but the site energy balance is unchanged. C-66's "a large portion of the heat will never even meet the atmosphere" is thermodynamically incorrect as stated. The efficiency benefit is real; the heat does not disappear.

The ~13 W/m² claim (C-63)

Note on the denominator. The Utah County Assessor record gives 69.478 acres = 281,167 m², not the ~25 acres stated in testimony. The parcel is the basis used below.

Over the parcel:

417e6 W ÷ 281,167 m² = 1,483 W/m²

That is 114× the deck's stated ~13 W/m². (Over the ~92,900 m² building footprint alone it would be ~4,490 W/m², or 345×; the parcel is the fairer denominator for a site-wide flux and is used here.) So what area produces 13?

417e6 ÷ 13 = 3.2e7 m² = 32 km² ≈ 12.4 sq mi

Spanish Fork's municipal area is ~40.4 km², which yields:

417e6 ÷ 40.4e6 = 10.3 W/m² , the same order as the stated 13

Conclusion: the ~13 W/m² figure is the facility's total heat divided by the entire city's area, then compared against the standard urban anthropogenic heat flux literature range of 2–30 W/m².

Verdict on C-63: MISLEADING BY JUXTAPOSITION, not false. As a statement about city-wide heat flux the comparison is coherent and even fair. But it is printed directly beside "Max of +5°F on-site," inviting the reader to treat the two as the same basis. They differ by a factor of ~317. A slide that said "spread across Spanish Fork's full area this is comparable to the heat the city already produces" would be honest and would make the same point.

The Lake Side comparison (C-64), this one checks out

Lake Side Power Plant, Vineyard: ~1,200 MW combined cycle, deck-cited heat rate 7,146 Btu/kWh → η = 3,412/7,146 = 47.7%.

Lake Side waste heat = 1,200 × (1/0.477 − 1) = 1,315 MW-thermal
Volition ÷ Lake Side = 417 / 1,315 = 31.7%

"Less than half of what an existing neighboring power plant already produces" is CORRECT. C-64 VERIFIED. Credit where due, and note this is the one slide in the deck that discloses its own methodological limitation.

On-site temperature rise (C-62, C-65)

Air-mass energy balance, 417 MW into a plume crossing a 1,000 m width:

ΔT = Q / (ρ · V · A · cp),  ρ ≈ 1.0 kg/m³ at ~4,600 ft, cp = 1,005 J/kg·K

At near-calm the uniform-mixing model breaks down: a 417 MW buoyant source generates strong vertical plume rise, so heat goes up rather than accumulating laterally.

Verdict: C-62 (+5°F max on-site) is plausible and in the right range. C-65 (<1°F beyond a quarter mile) is also plausible given plume rise and dilution.

At approximately one mile the facility's thermal contribution will be far below 1°F, and not perceptible. The "heat plume following prevailing wind" argument is directionally correct physics but the magnitude at this distance is negligible. Heat is not the impact pathway to worry about at the receptor.

CALC-8: Noise at the receptor

Tests C-54, C-59, C-60.

Source spec, confirmed. Module rating and noise both come from the Bloom Energy Server 6.5 datasheet: 325 kW net AC per unit, and "<65 dBA at 10 ft per unit", verbatim, measurement distance included. Volition did not invent, round, or shade it.

Module count

250.00 MW ÷ 0.325 MW =   769 units   (rack-level compute load)
393.25 MW ÷ 0.325 MW = 1,210 units   (the permitted nameplate; matches S1 exactly)
500.00 MW ÷ 0.325 MW = 1,538 units   (ordinance cap; 1,539 would be 500.175 MW, over it)

The structural problem with the deck's number is unchanged. P-12 states "<65 dBA, noise at 10 ft per unit." A per-unit figure says nothing about an array of 770 sources, and the property-line study was still "underway" at hearing time (C-55). The criticism is not that the number is wrong, it is that a source spec is being presented where a receptor prediction belongs.

Per-unit sound power, hemispherical radiation at r = 3.05 m:

Lw       = 65 + 10·log10(2π · 3.05²) = 65 + 17.7  =  82.7 dBA re 1 pW
Lw_array = 82.7 + 10·log10(770)      = 82.7 + 28.9 = 111.6 dBA

At the receptor, R = 1,609 m (approximately 1 mile), measured from the Utah County parcel polygon, geometric spreading only:

Lp = 111.6 − 10·log10(2π · 1,609²) = 111.6 − 72.1 = 39.5 dBA

What §6's residential-boundary standard requires as a setback

§6 sets 55 dBA at a residential zone boundary, at any hour, a stricter figure than the 65 dBA property-line limit, and the one that would govern wherever the nearest residential zoning line falls.

Inverting the model for the setback that standard requires, spreading only, no mitigation credit:

55 dBA = 111.6 − 10·log10(2π · R²)
10·log10(2π · R²) = 56.6   →   R ≈ 300 m  (≈ 985 ft)

So roughly 300 m of separation between the fuel-cell array and the residential zone boundary is the floor for compliance before any attenuation credit. The parcel is 69.5 acres, about 530 m on a side if square, so this is achievable, but it materially constrains where the array and the cooling plant can sit, and no site plan showing that separation exists in the public record.

⚠ The source term is the weak point of this whole calculation, and it belongs up front. Inverting "<65 dBA at 10 ft" via hemispherical radiation treats the unit as a point source , but the Bloom ES 6.5 cabinet is 29'5" long, so at a 10 ft measurement distance the source is three times longer than the distance. Finite-line integration gives 84.5 dBA sound power and an ISO 3744 measurement surface gives 90.6, against 82.7 for the point-source treatment. All three are carried below, because the spread between them is the honest answer.

Array sound power, at 770 units:

Point-source inversion (as used below):  82.7 + 28.9 = 111.6 dBA   [understates]
Finite-line integration:                 84.5 + 28.9 = 113.4 dBA
ISO 3744 measurement surface:            90.6 + 28.9 = 119.5 dBA

Excess attenuation over 1,609 m:

Honest neutral-condition range at the receptor, spanning both the source-term and attenuation uncertainty:

Lowest:  113.4 − 72.1 − 28 = 13.3 dBA
Highest: 119.5 − 72.1 −  9 = 38.4 dBA

≈13–38 dBA.

That range is too wide to be decision-relevant, and saying so is the finding. A 25 dB spread covers everything from inaudible to clearly present. This model cannot discriminate, and no version of it should be quoted at a hearing. Its only legitimate use is to establish that the question is open, which is the argument for demanding the property-line study, not a substitute for it.

Adding the two site-specific enhancers present at this receptor, winter inversion (+5 to +10 dB) and nocturnal down-canyon flow (+3 to +8 dB), pushes the upper end to roughly 50 dBA under coincident worst-case conditions, against a typical rural night ambient of 30–40 dBA.

The 55 dBA setback, recomputed

Inverting for §6's residential-zone-boundary standard, spreading only, no mitigation credit:

Lw = 111.6 (point-source, as originally used) →  R = 270 m   [not the 300 m printed earlier]
Lw = 113.4 (finite line)                      →  R = 332 m
Lw = 119.5 (ISO 3744)                         →  R = 670 m

The parcel is 69.5 acres, roughly 530 m on a side if square. The upper case does not fit. Barriers, enclosure, and ground effect can buy back 5–15 dB and would shrink it substantially, but no site plan showing any separation exists in the public record. That is the point worth making: not that compliance is impossible, but that nobody has shown it is achievable on this parcel, and the ordinance requires no post-construction check either way.

Sensitivity to authorized scale. At the ordinance's 500 MW cap (1,540 units) every figure rises +3.0 dB, the smallest real-world audible step. Scale is a second-order variable; bearing, atmospheric condition, and the source term dominate it.

Model validation against a measured campus, and it does not pass. KSL measured ~70 dB at ~100 m from Aligned's West Jordan campus. The point-source model gives 63.6 dBA at 100 m, 6.4 dB low, against a source (diesel generators and blowers) that should read louder per unit than fuel cells, not quieter. That is evidence against the source term, not for it, and it is consistent with the finite-line and ISO 3744 corrections above. The answer at the receptor, stated at the confidence the evidence supports. Typical rural-subdivision night ambient runs ~30–40 dBA and US-6 already raises the local floor.

• Neutral daytime conditions: most likely at or below background.
• Winter inversion with nocturnal down-canyon flow: possibly audible as a steady low-level hum.
• The honest bottom line is that the uncertainty band is wider than the effect being measured. Anyone, including this review, claiming to predict the receptor level from public information is overreaching.
• Winter inversion with nocturnal down-canyon flow: plausibly at or slightly above night ambient, a steady low-level hum rather than an intrusion, and most noticeable if the cooling equipment carries tonal content. Fan blade-pass tones are the usual culprit and are far more noticeable than broadband noise at the same dBA. The revised module count moves the estimate down ~2 dB from the original; it does not change the character of the answer.

Residual uncertainties, in descending order of consequence:

1. The cooling plant, not the fuel cells, is the likely dominant source. No spec for it exists in any public document. This model covers only the Bloom array and is therefore a floor, not a ceiling.
2. Whether 250 MW is delivered IT load or gross fuel-cell output. If parasitic load sits on top, the true unit count exceeds 770.
3. Real terrain, versus the flat-earth spreading assumed here.

The property-line acoustic study is the binding document; this is a check on it, not a substitute. The right ask is not "is Volition lying about noise", the source number is verbatim from the manufacturer. It is: publish the study, make the limit receptor-located rather than per-unit, and require the post-construction verification Volition's own Justification Statement already says exists (it does not).

CALC-12: Potential to emit at the authorized nameplate

Potential to emit is assessed at what the permit allows, not what the applicant intends to run. The delivered figure is 250 MW, but the ordinance authorizes 500 MW per site and the applicant's own request referenced 400–500 MW nameplate.

Inputs

NOx ceiling:      0.07 lb/MWh   [Bloom CARB DG certification EO DG-058, the certified
                                 ceiling, not the 5.17 tpy the deck estimates]
Capacity factor:  0.85          [conservative for PTE; the applicant stated 75% expected operation]
Hours:            8,760/yr

Result

250.00 MW × 8,760 × 0.85 × 0.07 ÷ 2,000 =  65.1 tons/yr
393.25 MW × 8,760 × 0.85 × 0.07 ÷ 2,000 = 102.5 tons/yr   [the permitted nameplate]
500.00 MW × 8,760 × 0.85 × 0.07 ÷ 2,000 = 130.3 tons/yr   [the ordinance cap]

(At the 75% capacity factor the applicant stated, 250 MW gives 57.5 tpy, which still exceeds the 50 tpy county offset trigger. The 85% case above is the appropriate basis for a potential-to-emit test.)

Thresholds crossed

Reading this honestly, because it cuts both ways.

• This is a ceiling, not a prediction. Volition's stated estimate is 5.17 tpy, roughly 11× below even the 250 MW ceiling case, and that estimate is conservative against Delaware's issued Bloom permits (~1.8 tpy at 250 MW). Nothing here suggests the facility will actually emit 130 tons.
• But PTE is the legal test, and a zoning cap is the only durable control. A facility permitted at 500 MW under §15.3.24 could, on certified-ceiling equipment at high utilization, cross the federal major-source threshold, which would trigger NNSR, offsets, and LAER. The ordinance contains no potential-to-emit cap, and §7 requires only that DAQ approvals be obtained and maintained.
• No such cap appears anywhere in Exhibit A. A ceiling written into the code is the only instrument that would foreclose this permanently.

CALC-13: What the amendment authorizes, at its own stated maximum

This is now the report's primary frame. §15.3.24 caps generation at "not more than 500 megawatts of nameplate generating capacity, measured per Onsite Power Generation Facility" (yellow-highlighted, so a Planning Commission addition), with a floor of 50 MW. §13 caps Data Center sites at two citywide.

Method: the same one Utah DAQ used. DAQE-IN163550001-26 computes every limit as the manufacturer's published rate × full nameplate × 8,760 hours. Applying that method at 500 MW is not a worst case invented here; it is the agency's own arithmetic at the ordinance's own number.

500 MW x 8,760 h = 4,380,000 MWh/yr per facility

Inputs and their sources. Bloom ES 6.5 datasheet 1016932-20260204 for 325 kW/unit, 65–53% LHV efficiency, 5,811–7,127 Btu/kWh HHV, 0.003 lb NOx/MWh, 0.013 lb CO/MWh, 0.01 lb VOC/MWh, 378 kg CO₂/MWh at end of life. PM, SO₂ and HAPs are scaled linearly from DAQE-IN163550001-26 (×500/393.25) because Bloom publishes no PM figure and calls SO₂ negligible. Gas at 1,037 Btu/cf. Water and land scaled from the permitted design.

The quota does not cap what people think it caps

§13, verbatim:

"Not more than two Data Center with Onsite Power Generation Facility sites… shall be approved and active within the corporate limits of Spanish Fork City at any time… This limitation does not apply to a standalone Onsite Power Generation Facility established without an associated Data Center."

So the citywide ceiling is not 1,000 MW. It is 1,000 MW of data-center-paired generation, plus an uncapped number of standalone Onsite Power Generation Facilities, each permitted up to 500 MW.

The one real constraint on those is §2's requirement that generation be "established only in conjunction with, and scaled to serve, one or more principal uses lawfully established or concurrently approved on the same site." A standalone facility still needs a principal use to serve, so this is not a merchant-plant loophole. But the principal use can be any lawful I-1 use, and nothing in the text limits how many such pairings the city may approve.

A second textual point. The 500 MW cap is "measured per Onsite Power Generation Facility," while the quota is measured per site, defined as "a single parcel or unified development." The two are measured in different units, and nothing states that one site may host only one facility.

Both points are questions for the City Attorney, not findings. They are exactly the kind of ambiguity that gets resolved cheaply before adoption and expensively afterward.

CALC-9: Scale against the city (C-06)

250 MW ÷ 90 MW city peak = 2.78×
400 MW ÷ 90 MW           = 4.44×
500 MW ÷ 90 MW           = 5.56×
1000 MW ÷ 90 MW          = 11.11×   (two sites at the cap)

Against UMPA's primary-sourced figures (I-11a, I-11b):
500 MW  ÷ 74.10 MW FY2021 actual   = 6.75×
500 MW  ÷ 115.76 MW FY2027 forecast = 4.32×
1000 MW ÷ 115.76 MW FY2027 forecast = 8.64×

C-06 VERIFIED. At 250 MW the facility generates ~2.8× the entire city's peak consumption; at the requested 500 MW cap, ~5.6×.

This is the single most legible fact in the entire proposal and it is not in the deck. It also reframes C-11: the "zero draw on the municipal grid" claim is accurate and valuable precisely because the load is this large, Spanish Fork Power could not serve it under any scenario. Staff's own testimony put a grid-connected 50 MW load at "two new substations... 30 to 40 million dollars" (CC 2:32:52). A 250 MW grid-tied load is not a thing Spanish Fork could do at all.

CALC-10: Fiscal

Sources. Utah County Assessor parcel 67:228:0003, Tax Area 150; Utah Admin Code R884-24P-33 percent-good schedules.

Confirmed inputs

Parcel:              67:228:0003, 69.478641 acres (NOT ~25 as testified)
Owner of record:     WDC Industrial Utah Owner LP
2026 total assessed: $50,898,400   [confirms the "~$50 million" cited on the record, 04 AUG min. p. 69]
Tax Area:            150  (point-in-polygon confirmed)
Combined rate:       0.010979
Distribution:        Nebo School District 79.7% | County+districts ~11.3% | Spanish Fork City 9.0%

Year one

$4.0B × 0.010979 = $43.9M gross
  → Nebo SD          ~$35.0M
  → County/districts  ~$5.0M
  → Spanish Fork City ~$3.95M

C-83's "approximately $40mm per year" is arithmetically sound in year one and is, if anything, slightly conservative at the real certified rate. The figure is not invented. But it is gross, and 79.7% of it is Nebo School District's, not the city's. A Spanish Fork Council chamber heard a number of which the city keeps roughly 9%.

The decay curve, on Utah's actual schedules

R884-24P-33 assigns two very different classes to the two halves of this facility:

(Each column models the entire $4B in that single class, at 0.010979. Reality is a blend.)

Class 12 reaches its 7% percent-good floor by year five. Class 27 retains ~76% at year ten.

⚠ Two limits on this table. (1) R884-24P-33 is a personal property schedule. A building retrofit is real property and does not depreciate on it, so the portion of the $4B that is construction is modeled wrongly here. (2) Volition has separately described the $4B as excluding tenant compute hardware. If that is right, the Class 12 collapse applies to a much smaller base than modeled, and the decay is far shallower than shown.

The direction of the finding survives both, generation equipment holds value and compute does not, but the magnitudes are illustrative, not predictive. This is precisely why the pro forma is the thing to demand rather than something to substitute a model for.

So the durability of the whole claim reduces to one undisclosed number: the compute/generation split.

All-Class-12 case:  $27.2M → $3.1M/yr forever   ( ~$279k/yr to the city )
All-Class-27 case:  $42.6M → $32.5M at year 10  ( ~$2.9M/yr to the city )

To sustain a flat $40M/yr requires $3.64B in assessed value every single year, continuous compute replacement, each refresh assessed at full value.

C-86 (replacement sustains the base) is plausible for the generation half: The Volition representative told Council the Bloom equipment gets "a phased upgrade approximately seven or eight years after installation" (04 AUG min. p. 68), which matches Class 27 behavior. He gave no comparable figure for the compute side, which is the half that collapses. Replacement is a business decision, not a covenant. A development agreement could require a minimum assessed value; none is proposed.

The document to demand is the tax pro forma, Class 12/27 split, refresh assumption, assessed value at years 1, 5, 10. A councilmember asked for exactly this on 04 AUG: "the underlying figures needed to be clearly documented."

Also note C-91: the downside scenario of 75 MW at $1.4B implies $18.7M/MW, against $16M/MW for the 250 MW/$4B case. Roughly consistent, the ratios do reconcile.

Two errors to avoid, both easy to make:

1. Do not net the sales-tax exemption against property tax. Utah Code 59-12-104(84) is a sales and use tax exemption. It does not touch property tax. (And 59-12-104(80) already exempts fuel cells outright, independent of any data-center status.)
2. Do not cite $81,000,000 as a sale price. That figure in the parcel abstract is a trust deed. Utah is a non-disclosure state; no sale price exists in the public record.

CALC-11: Water plausibility [PARAMETRIC]

C-42: 250,000 gal/yr for a 250 MW closed-loop facility = 685 gal/day. For humidification top-up on a sealed loop this is plausible; closed-loop systems genuinely do consume almost nothing.

But the accounting is incomplete. At Novva West Jordan, public utility records show ~3M gal/yr with roughly two-thirds going to landscaping, about 2× the operational use. Volition's 250,000 gal figure is explicitly humidification-only and appears to exclude site irrigation entirely.

If Spanish Fork's landscaping requirements for a 25-acre industrial site drive irrigation demand comparable to Novva's, total site water could plausibly be several times the headline figure. The unnamed commissioner who raised this (C-52) was asking the right question and never got an answer.

C-41: the 500,000 gal initial fill against the deck's own cited basis (155 gal/person/day × 3.39 persons = 525 gal/household/day = 191,700 gal/yr):

500,000 ÷ 191,700 = 2.6 households, deck says "2 Utah households." ✓ close enough
250,000 ÷ 191,700 = 1.3 households, deck says "one household." ✓

The household arithmetic checks out. The water claims are internally consistent and the cited sources are real. The gap is scope, not honesty.

Running summary of verdicts from calculation