Built 2026-08-30. Source: redteam-codex.md.

Codex red-team review

The report is not ready to drive a purchase or a freezer modification. It correctly catches the bad NH4Cl recipe ratio and the weak recharge margin in the owner's existing freezers. It then recommends designs that its own rate analysis says cannot deliver most of their nominal PCM capacity during a chill. The owner-proposed dedicated recharge freezer and swap bank can work, but the report partly rejects it for the wrong reason. The electrical design also has two blockers: unapproved electrical equipment inside a hydrocarbon-refrigerant cabinet, and an undersized reed switch driving a relay coil that can exceed the switch rating.

1. BLOCKER: the U-tek swap bank is dismissed for the wrong reason

"A -23 C PCM ... is not viable in this box under any scheme."

That is false for the scheme the owner proposed. A -23 C pack kept in a separate recharge freezer until a hot batch is ready does not need to remain frozen indefinitely in the -15 to -18 C working freezer. Its melting there is the intended discharge. The report briefly admits this when it says the pack must go "straight into the batch," then contradicts itself by saying the working box must stay at or below the PCM melt point.

Fisher and ThermoSafe confirm that 035257G is six 48 oz packs, $88, with a -23 C phase point. The bank weighs:

6 x 48 oz x 0.0283495 kg/oz = 8.165 kg

ThermoSafe does not publish latent heat on the retail page. At an illustrative 200, 250, or 300 kJ/kg, the bank would hold 1.63, 2.04, or 2.45 MJ. A 10 kg batch from 60 C to 5 C needs 2.19 MJ before door openings. Six packs might cover most of one 10 kg batch, but the report cannot claim that without the actual enthalpy data. Their six-pack exposed area is about 0.80 m2. At h = 15 W/m2K and an 18 K air-to-pack difference, the initial absorption rate is about 0.80 x 15 x 18 = 216 W. At h = 25, it is about 359 W. That is materially better than calling the packs useless, although it still may not absorb a 400 to 700 W food load by itself.

The recharge freezer must remain below -23 C. A target near -28 C is prudent, but the report presents that target as a product requirement without a ThermoSafe conditioning-time curve. Verify the actual recharge freezer at its warmest location and validate a full charge time with the packs spaced apart.

Fix: Evaluate the owner's scheme as a separate design. Keep two U-tek sets in a dedicated non-food recharge freezer, move one set directly into the working box only when a batch starts, and return it immediately after use. Obtain latent heat, cycling, conditioning, and load-limit data from ThermoSafe before deciding whether six packs or twelve are required.

2. BLOCKER: the PCM purchase rule ignores how little capacity is usable during the chill

"Rule: 1 kg of -15 C PCM per 1 kg of hot food at 60 to 70 C, margin included."

The mass arithmetic is correct only as an energy-capacity calculation:

10 kg x 3.8 kJ/kgK x 55 K x 1.05 = 2,194.5 kJ

2,200 x 1.30 / 290 = 9.86 kg corrected NH4Cl PCM

The report's own rate calculation says thirteen bottles absorb only 25 to 105 W while the food releases 400 to 700 W. Over a 90-minute chill, that is only 0.14 to 0.57 MJ, or 6 to 26 percent of the 2.2 MJ load. A bank with 2 to 3 MJ of nominal capacity cannot protect inventory by 2 to 3 MJ if its surface can transfer only a fraction of that before the food reaches 5 C. This invalidates the staff rule, the bottle count, the claim that five Engel packs are sized for 10 kg, and much of the equipment-life argument.

The report even says the bank "takes 5 to 20 percent" of a chill, then continues to size and price it as if all latent capacity participates. Buying more mass without adding exposed area makes recharge slower and does little for the peak.

Fix: Size both energy and rate. State the required PCM surface area, air velocity, spacing, and maximum absorption power for each layout. Base any purchase on a full 10 kg test that measures pack heat uptake or melt fraction, not on kilograms of PCM alone.

3. BLOCKER: the 800 mL test cannot be scaled by an invented multiplier

"multiply run C's time by about 1.5 for a real 10 kg liquid batch ... and by 3 for solids"

No derivation supports those multipliers. The small test removes about:

0.8 x 4.19 x 50 + 1.0 x 0.50 x 50 = 193 kJ

The proposed 10 kg batch removes about 2,195 kJ. That is 11.4 times as much energy. Two pans have more surface area than the pot, but surface area, product depth, internal mixing, pan metal, airflow bypass, and compressor response do not scale linearly. A 1.5 multiplier could underpredict the real time badly. Solids cannot share one multiplier because rice, braise, beans, and thick sauce have different thermal diffusivity and may not convect internally.

The report's go or no-go decision and the instruction to buy an SF-5 after a failed pot test are therefore not valid decision rules.

Fix: Use the pot test only to locate and set the fan. Validate each intended product class at the real batch mass, pan count, fill depth, and starting temperature. Do not extrapolate a food-safety time from the pot.

4. BLOCKER: a brushless fan is not an approved ignition-safe retrofit

"On an R290 cabinet buy the Noctua array instead."

True's current T-Series manual says not to use electrical appliances inside the food compartment unless the manufacturer recommends them. A brushless Noctua fan has no brushes, but IP67 is an ingress rating. It is not an explosion-protection rating, it is not approved by the freezer maker, and it does not make the controller, connectors, relay, or wiring ignition-safe during an R290 or R600a leak. The report quotes the manufacturer's prohibition and then recommends violating it.

The modification can also void the freezer's UL or ETL evaluation, sanitation listing, and warranty. A cheap chest freezer commonly uses R290 or R600a too, so moving the same unapproved fan into one does not solve the ignition issue.

Fix: Do not install any add-on electrical device inside an R290 or R600a cabinet without written manufacturer approval and a refrigeration or electrical professional's design. Use a purpose-built blast chiller, an approved OEM fan assembly, or a dedicated non-hydrocarbon test cabinet with a documented safe installation.

5. BLOCKER: the repaired door interlock can still overload the reed switch

"Wire the reed switch to the coil ... of the 12 V 30/40 A automotive relay"

The Adafruit 375 reed switch is rated for 100 mA maximum. A common Bosch-style 12 V automotive relay coil is about 1.8 W and 80 ohms. Its steady coil current is:

12 V / 80 ohm = 0.150 A

That is 150 mA, already 50 percent over the reed rating before inrush and contact arcing. The earlier review caught the much worse error of putting the fan through the reed switch, but the new relay does not necessarily fix it. The exact True MODS kit coil current is not documented in the report. There is also no flyback suppression specified, so opening the coil circuit can arc the reed contacts.

Fix: Use a door switch rated for the measured relay coil current with margin, or use the reed switch as a low-current input to a listed solid-state driver. Add flyback suppression, put the driver and relay outside the wet compartment in an enclosure, and prove fail-safe door-open behavior before use.

6. MAJOR: the 12 V circuit is not automatically NEC Class 2, and the fuse may never open

"The 12 V side is below the 60 V DC Class 2 ceiling"

Voltage alone does not make a circuit Class 2. The cited PULS Class 2 note says the power supply must be listed and marked as a Class 2 power source. It gives 12 V, 4.85 A, 58 W as one Class 2 output example. The proposed 12 V, 10 A supply is 120 W and is not shown as a listed Class 2 source.

The 15 A blade fuse is also a poor match. A current-limited 10 A supply may never deliver enough current to open a 15 A fuse. The report says the fuse protects 12 AWG wire, but the flat cable under the gasket has no confirmed conductor gauge or ampacity. The fuse must protect the smallest conductor and connector in the entire run, not the fan lead alone.

Fix: Have an electrician specify a listed supply and output protection. If the circuit is meant to be Class 2, use a supply marked Class 2 within its output limit. Otherwise treat it as a power circuit, enclose it, and fuse each branch to the weakest wire and connector with motor-start behavior accounted for.

"Fan first, PCM later ... Buy the GESEXI 16 in"

The GESEXI listing supplies no cold-temperature rating, IP rating, sanitation certification, or listing for use inside commercial food equipment. The report acknowledges all four gaps but still makes this fan the first purchase. Frost, condensation, cleaning chemicals, grease, and repeated thermal cycling are normal in a commissary freezer. A shelf is not a complete finger guard, and an automotive shroud and motor are not designed for washability or food-zone sanitation.

The SPAL rating is also weakly sourced. The report says the model-specific IP and current data were not confirmed. The open bench supply has no UL or ETL mark, yet the report calls it acceptable for a trial in a wet commercial kitchen.

Fix: Replace the fan BOM with a cold-rated, moisture-rated, cleanable, guarded assembly approved for the cabinet and installed by a qualified technician. Require a listed power supply from the first trial. Do not treat a commercial kitchen as a laboratory exemption.

8. MAJOR: -15 to -18 C does not prove that every freezer is sick

"A commercial reach-in that holds only -15 to -18 C empty is sick"

That conclusion comes from one True T-23F-HC specification. The exact owner models, controller setpoints, loading patterns, probe locations, and rated temperature ranges are unknown. True's T-23F-HC is designed to maintain -23.3 C, so -15 to -18 C would justify investigation on that model. Other commercial freezers operate around 0 F, have adjustable ranges, or may have been intentionally set warmer. A stable cycling band can also reflect the controller differential rather than a fault.

Calling the unit sick before reading its model-specific manual may trigger unnecessary service or an unsafe setpoint change. The report does correctly warn not to force a unit to -23 C if it cannot reach it.

Fix: Read each nameplate and model manual first. Log a calibrated product-simulant probe and air probe through several cycles. Compare the result with that model's specified range before diagnosing a dirty condenser, low charge, iced coil, or gasket failure.

9. MAJOR: hot pull-down is not shown to be a designed daily event

"a hot load is a designed-for event on a healthy unit"

The manufacturer evidence says the opposite. Hoshizaki instructs users not to put warm or hot foods in the cabinet because they can overload the appliance. FDA's cooling guidance also distinguishes holding refrigeration from rapid-cooling equipment. An initial empty-cabinet pull-down does not prove suitability for daily 60 C product loads.

The report's energy comparison is not calculated. Empty-box pull-down energy depends on cabinet steel, liner, shelves, evaporator mass, and insulation. Even if one startup has comparable energy, it happens rarely and under controlled ambient conditions. A 10 kg food load is about 2.2 MJ every day, plus moisture and possibly another 3.2 MJ if the food freezes.

Fix: Describe hot-food loading as outside the normal holding use unless the model manual expressly permits it. Ask a refrigeration technician to check compressor envelope, condensing temperature, superheat, current, and discharge temperature during a representative batch.

10. MAJOR: compressor hours are calculated from generic horsepower, not the owner's equipment

"96 min at 381 W ... 67 min at 547 W ... 37 min at 1,000 W"

The division is correct:

2,200 kJ / 0.381 kJ/s = 5,774 s = 96.2 min

The inputs are not tied to the freezer. Compressor horsepower does not uniquely determine refrigeration capacity. Capacity changes with refrigerant, evaporating temperature, condensing temperature, return-gas temperature, compressor model, and cabinet airflow.

The chest-freezer discussion is internally inconsistent. It identifies an Avantco CF8 as 1/7 HP and 101 W input, then discusses 1/3 HP chest-freezer duty using 381 W of refrigeration. If the small box delivered only 150 W of cooling, a 30 kg load would require:

6,600 kJ / 0.150 kJ/s = 44,000 s = 12.2 h

That is not the table's 3.4 to 4.8 hours. Baseline cabinet load would add to it.

Fix: Remove generic horsepower run times from the decision. Use exact compressor performance data for each owner model or measure recovery energy and continuous run time during a bounded load test.

11. MAJOR: the SOP adds a much larger freezing load that the engineering tables omit

"Refrigerate, or leave to freeze."

The daily-load tables stop at 5 C. Freezing watery food from 5 C to -18 C adds about 320 kJ/kg. For 10 kg, that is another 3.2 MJ on top of the 2.2 MJ chill. Total load becomes about 5.4 MJ before fan heat and moisture. For 30 kg, the added freezing load is about 9.6 MJ and the total becomes about 16.2 MJ.

This is larger than the report's chill load and can overlap the PCM recharge period. It invalidates the stated daily duty additions, overnight recovery claims, and "one batch per set per day" guidance if staff leave food in the same freezer to freeze.

Fix: Make the operating mode explicit. If the goal ends at 5 C, move food to a refrigerator. If the food must freeze, calculate and validate the full load separately and do not recharge PCM in the same box until the food-freezing load is complete.

12. MAJOR: Engel capacity is invented from another product's datasheet

"about 683 kJ each at 301 J/g assumed"

Engel publishes the phase temperature, mass, dimensions, and non-toxic claim. It does not publish latent heat or composition on the cited page. Similar shell wording is not evidence that the fill is PureTemp -15, and "likely PureTemp class" cannot support a $150 purchase or the claim that five packs provide 3.4 MJ.

The Engel page also says 6 to 8 hours in a household freezer, while a current owner review on the same page reports that a pack remained slushy at -5 F. The report wisely uses a colder, longer recharge rule elsewhere, but its capacity number remains unsupported.

Fix: Ask Engel for the enthalpy curve, conditioning temperature and time, and usable-cycle data for ENGICE-FX. If Engel will not provide them, test one pack calorimetrically before buying five or thirteen.

13. MAJOR: the KCl tier conflicts with the report's own inventory limit

"Tier B with KCl ... is the only DIY brew that recharges in a -15 to -18 C box."

The report also predicts that KCl will let box air rise to about -8 to -5 C. Its go or no-go rule rejects any run in which the frozen-inventory sentinel rises above -12 C. Recommending KCl inside a storage freezer while predicting a warmer air cap is internally inconsistent. KCl may work in a dedicated chilling box with no valuable frozen inventory, but it is a poor inventory-protection battery.

The composition and procurement language also overreaches. "USP food grade" and "FCC grade" are not interchangeable claims. An Amazon listing is not a lot-specific certificate of analysis. GRAS status for KCl as a food ingredient does not make an unknown amount of coolant, xanthan, bentonite, and container residue acceptable on food.

Fix: Limit KCl to a dedicated chilling cabinet after a real sentinel test, or drop the tier. Require an identified manufacturer, lot certificate, SDS, and FCC compliance if the chemical remains in scope. Food touched by any leaked coolant still gets discarded.

14. MAJOR: the chemical-storage position is not established, and the production mixer should not be used

"The defensible position is that a sealed, labeled coolant pack is operating equipment"

That is an argument, not an approval. FDA Food Code 7-201.11 requires toxic materials to be separated and not stored above food. FDA 7-202.11 also restricts their presence and use to materials needed for the establishment and used consistently with labeling. The report itself quotes a local rule that may prohibit chemicals in an area where food and equipment are stored. A home-mixed coolant in a reagent bottle has no manufacturer's use label for installation inside a food freezer.

The recipe instructs staff to mix the coolant in a 20 qt kitchen mixer and return the bowl to food use unless carbon fiber was present. That is a needless cross-contamination and employee-error route. Unknown-grade NH4Cl, bentonite dust, borax, and labels make this chemical work, not food preparation.

Fix: Get written approval from the local regulator or use a dedicated non-food chilling cabinet. Use dedicated chemical mixing tools and a separate controlled area for every coolant recipe, regardless of whether carbon fiber is omitted.

15. MAJOR: borax is introduced without NH4Cl-specific evidence or a food-area safety case

"0.5 to 1 wt% borax is the literature's standard choice"

The cited literature discusses nucleators across other hydrated-salt systems. The report does not show that borax is effective in NH4Cl-water at -15 C, compatible with xanthan and the bottle, or suitable for this commissary. It then tells staff to use it "if a source is to hand," with no grade, SDS, mass, or label change.

Adding 50 to 100 g per 10 kg batch also changes the phase system and safety profile. This is not a harmless recipe tweak.

Fix: Remove borax from the build. If supercooling prevents recharge, stop and use a characterized commercial PCM or commission a small controlled formulation study outside the food operation.

16. MAJOR: staff should not rinse a contaminated evaporator and restart the freezer

"If brine reached the evaporator: unplug, thaw, rinse the coil ... restart"

NH4Cl brine can enter a dense aluminum-fin and copper-tube coil, wiring connectors, drain components, fan hardware, and insulation. A surface rinse does not prove chloride removal. Spraying water into an equipment compartment can spread contamination and create a later electrical or corrosion failure. The report's own compatibility section rates several freezer metals poorly.

The instruction also exposes staff to sharp fins and puts an equipment-repair decision into a spill SOP. The SDS says not to discharge the chemical into drains, yet the procedure does not say how the rinse water is captured.

Fix: If coolant reaches the evaporator or electrical parts, take the cabinet out of service, quarantine affected food, and call a refrigeration technician. Capture and dispose of cleanup liquid according to the SDS and local requirements. Restart only after inspection and documented cleanup.

17. MAJOR: placing hot pans directly on cold packs is unvalidated and can damage food or packaging

"set the hotel pans directly on flat frozen packs"

Direct contact raises heat-transfer rate, but the report does not check pack load rating, puncture resistance under pan edges, maximum contact temperature, cleanability, or thermal cycling. A -15 or -23 C surface against a metal pan can freeze the bottom layer while the center remains warm. That creates uneven texture and makes a center-only probe miss the cold boundary condition. Flexible U-tek pouches are shipping refrigerants, not pan supports.

Fix: Do not put hotel pans on packs unless the pack maker approves the temperature and static load. If conduction is tested, use a cleanable rigid cold plate designed for foodservice, control the contact pressure, and map top, center, and bottom food temperatures.

18. MAJOR: both proposed cable routes create predictable cabinet problems

"flat two-conductor under the magnetic gasket ... silicone over the gap"

"Cable out through the defrost drain plug ... you lose the drain"

A gasket must conform to the cabinet face. A cable plus silicone can make a permanent air leak, interfere with door closure, wick condensation, and wear through as the door cycles. The report identifies whole-coil icing as a risk and then accepts the leak route.

Blocking a chest freezer's defrost drain is worse. Manual defrost produces water exactly where the energized cable exits. Losing the drain increases spill, cleaning, and electrical risk. Silicone around a removable drain is not a serviceable strain relief.

Fix: Do not route wiring through a gasket or sacrifice a drain. Use only a manufacturer-approved accessory port or a technician-installed bulkhead at a location verified free of refrigerant tubing and heater wiring. If no approved route exists, do not install the fan.

19. MAJOR: "rock hard" and squeeze deductions are not a state-of-charge measurement

"Cut the batch by 1 kg per soft jug"

A soft jug can be 5 percent melted or 95 percent melted. One kilogram per jug is based on nominal capacity, not the remaining frozen fraction or real transfer rate. Xanthan can also make a mostly liquid bottle feel resistant. Commercial pouches and rigid Engel packs do not share the same tactile behavior.

The sentinel scheme is underspecified. A probe inserted into NH4Cl needs a sealed, compatible thermowell. A probe taped to the outside reads surface temperature. A single sentinel does not represent 12 to 45 containers placed across a variable-airflow shelf, and normal probe accuracy may be comparable to the 1 to 2 K margin being used.

Fix: Validate a charge protocol for each pack type. Use fixed rack positions, a minimum conditioning time demonstrated by repeated tests, multiple instrumented sentinels, and a fail-closed rule. Do not convert squeeze feel into kilograms of allowed food.

20. MAJOR: the defrost response is unsafe operationally

"If 'dEF' appears mid-chill stop the booster and wait 15 minutes."

Defrost duration is model-specific and may be temperature-terminated. Fifteen minutes is not a verified endpoint. Hot food continues on its cooling clock while the compressor is off and the coil heater runs. Restarting the booster after a fixed delay may circulate warm, wet air before the evaporator fan-delay condition is satisfied.

Fix: Record actual defrost schedules and start only after a completed cycle. If an unexpected defrost begins during a critical chill, keep logging product temperature and transfer the pans to a validated alternate cooling method if the process is at risk. Do not use a universal 15-minute wait.

21. MAJOR: the validation test lacks controls and cannot support the claimed thresholds

"Let the freezer return to setpoint between runs (30 to 60 minutes)."

Air returning to setpoint does not mean the frozen inventory, cabinet liner, coil, or PCM has returned to the same energy state. Run order A, then B, then C biases later runs. One test per condition cannot separate the fan effect from defrost timing, compressor cycling, door-open time, probe placement, or a different starting inventory temperature.

The 65 and 80 percent pass thresholds come from estimates, not a manufacturer requirement or confidence interval. A failed threshold may mean the baseline reach-in already has good product airflow, not that the freezer "cannot take the load."

Fix: Repeat A and B at least three times in alternating order after full overnight equilibration. Use the same mass, pan, position, starting temperature, door-open time, and defrost phase. Log food, air, inventory simulant, compressor watts, and fan watts continuously. Treat improvement as measured data, not a universal pass percentage.

22. MAJOR: the Atosa purchase data conflict with the current manufacturer page

"Atosa EBF-05 does 55 lb / 33 lb ... 3 yr parts and labor"

Atosa's current US product page lists 55 lb chilling capacity, 22 lb freezing capacity, 115 V, 12 A, a 2-year parts-and-labor warranty, and a 5-year compressor warranty. The reseller page used by the report lists 33 lb and 3 years, so the sources disagree. The current CKitchen page also shows $5,707, not the report's $5,187.60 sale price.

The Avantco checks are stronger. Current Webstaurant pages confirm the SF-3 at $2,089 with 15 lb chill and 9 lb freeze, and the SF-5 at $3,789 with 31 lb chill and 26 lb freeze. Availability and delivery still need confirmation at checkout.

Fix: Correct Atosa to the manufacturer's current figures or obtain a written quote and warranty document that identifies the exact US model. Do not compare or buy from conflicting reseller data.

23. MINOR: the BOM totals hide real differences between cabinet types

"Tier C buys Engel -15 C packs and a sealed fan ($406)"

The line-item arithmetic is:

The $38 Tier A placeholder count is correct for carbon fiber, fuse, and guard, but the $36 cable is also only a range midpoint and the shipping costs are missing.

Fix: Show separate totals for hydrocarbon and non-hydrocarbon cabinets, add the omitted bentonite, and label shipping, tax, and unconfirmed lines outside the firm subtotal.

24. MINOR: the "10 L" recipes are actually about 10 kg

"Recipe ... at 10 L"

The corrected recipe contains 7.9 kg water and 2.1 kg NH4Cl before additives, so it is 10 kg, not 10 L. At about 1.06 kg/L, its liquid volume is roughly:

10 kg / 1.06 kg/L = 9.43 L

A 900 mL fill weighs about 954 g before additives, not 900 g. This shifts bottle counts and the "salt per 10 L" cost by about 6 percent. The 21 wt% arithmetic itself is correct:

265 / (1,000 + 265) = 20.95 percent

Fix: Label recipes by final mass. If staff fill by volume, measure actual density and state both grams and millilitres per bottle.

25. DEFENSIBLE: the NH4Cl correction is substantially right but too precise

"the NH4Cl-water eutectic is 19.7 wt% ... effective latent heat is about 250 kJ/kg"

The phase-diagram correction is sound. For 705 g water, the eutectic liquid contains about 173 g salt at the eutectic, leaving 295 - 173 = 122 g primary salt outside the eutectic plateau. The full recipe contains about 1.036 to 1.041 kg after additives.

Published modeling values for NH4Cl-water latent heat span about 281 to 314 kJ/kg. Applying those to the 878 g eutectic fraction gives:

878 x 281 / 1,038 = 238 kJ/kg

878 x 313.8 / 1,038 = 265 kJ/kg

The report's 250 kJ/kg is a reasonable midpoint, not a measured recipe property. Its 290 to 320 kJ/kg corrected-eutectic range is also plausible but should not be tied to commercial E-15 merely because density and phase temperature look similar.

Fix: Present 240 to 265 kJ/kg for the video recipe and label it as an estimate. Use differential scanning calorimetry or a controlled melt test before sizing a large bank.

26. DEFENSIBLE: the recharge table arithmetic is reasonable, but it is not a guarantee

"-16.5 C ... about 55 h" and "-23 C ... about 8 h"

For one bottle with 200 kJ latent capacity, 0.065 m2 area, and h = 15 W/m2K:

Adding sensible cooldown brings these close to the report's 55, 22, and 8 hour figures. The weak points are the assumed heat-transfer coefficient, guaranteed nucleation, and the unsupported claim that a packed tub is always two to three times slower.

Fix: Keep the table as a screening estimate. Require measured recharge time in the actual rack and freezer before setting daily throughput.

27. DEFENSIBLE: the fan heat arithmetic is correct only if the assumed power is measured

"The 16 in fan at 9 V dumps about 50 W into the box"

The energy conversions are correct:

50 W x 90 min x 60 s/min = 270 kJ

15 W x 12 h x 3,600 s/h = 648 kJ

The problem is using the cube fan law from a 12 V nameplate as if voltage, motor speed, airflow, and shaft power scale ideally for this brushed motor. At the report's own 9 V current limit of 6 A, electrical input would be 54 W, which is close. At 6 V, current and startup behavior remain unknown.

Fix: Measure cold-start current and steady watts at each operating voltage. Put those measured values into the heat and fuse calculations.

28. MINOR: the moisture and defrost numbers are not sourced well enough for an SOP

"Uncovered 60 C stock evaporates 1.5 to 3 percent of its mass"

The energy balance after assuming 150 to 300 g evaporation is coherent, but the evaporation fraction and the claim that uncovering is 20 to 30 percent faster have no product-specific evidence. Viscosity, exposed area, steam rate, fan velocity, relative humidity, and surface skin formation all matter. Staff may uncover pans based on a number that does not apply to rice, sauce, or braise.

The defrost example also compares all daily defrost-heater energy with the food load, when the retrofit decision needs the incremental defrost energy caused by the hot pans.

Fix: Weigh representative pans before and after cooling, log defrost duration, and compare loosely covered with uncovered runs. Use the measured incremental moisture loss and defrost time.

29. MINOR: "cooling paddles cut the load in half" is an unsupported promise

"use the cooling paddles for the first 30 minutes ... That cuts the freezer's load in half"

A frozen 4 L paddle can hold substantial cold, but the result depends on fill mass, initial temperature, how much ice melts, stirring, product viscosity, and whether the paddle itself was recharged in the same freezer. The report supplies no measurement or calculation for the 50 percent figure. Recharging the paddle in the same freezer shifts load just as PCM does.

Fix: Treat a paddle as a separate pre-cooling method. Weigh and temperature-log a representative batch to determine its actual heat removal, and recharge it in a unit whose load budget includes that energy.

30. DEFENSIBLE: rejecting PET beverage bottles is prudent

"Beverage containers belong only in a dedicated non-food chest freezer."

This is the right direction. PET does not suddenly fail at -25 C, and the report overstates the case when it implies that PET beverage bottles are inherently unsuitable merely because PET is below its glass-transition temperature. The real problems are unknown freeze-thaw cycle life, drop impact, cap compatibility, beverage-container confusion, and lack of a rated chemical closure. Ten percent headspace is plausible but has not been pressure-cycle tested with this gel.

Fix: Keep beverage bottles out of the food freezer. If DIY PCM remains in scope, use a chemically compatible HDPE container with a documented freezer rating and closure, secondary containment, unique shape and color, permanent labels, and a cycle retirement limit established by testing.

What the report gets right

Questions for the owner