FDA 21 CFR 117.80 Indirect Food Contact Compliance for Activated Alumina in Edible Oil Purification

1. Why regeneration-gas heat-exchanger sizing breaks the standard LMTD assumption

Regeneration gas loops for molecular sieve dryers (3A, 4A, 5A, 13X) and activated alumina polishers are textbook examples of "shell-and-tube gas-to-gas heat recovery" applications - and textbook examples of where the standard log-mean temperature difference (LMTD) method underestimates or overestimates the required exchanger area by 15-30% once the gas mixture deviates from ideal-gas behaviour. Practical field guidance is collected in our molecular sieve regeneration best practices playbook, which covers bed-side temperature profiling that complements the heater-side LMTD audit covered here. For buyers and plant engineers specifying or auditing a regeneration heater/cooler package, the gap between vendor "ideal-gas" heat-balance sheets and real operating area is the single largest source of capacity-rating disputes after commissioning.

The root cause is a combination of three effects that classic LMTD derivation assumes away: (a) real-gas Cp variation across the 150-300 °C temperature span of regeneration gas, (b) Joule-Thomson cooling across the adsorbent bed when high-pressure feed gas is depressurised and the wet bed absorbs heat of desorption, and (c) compositional non-ideality when the regeneration stream is a mixture (N₂ + 5-15% CH₄ in PSA tail-gas regeneration, or H₂ + CO₂ slip in hydrogen-loop regeneration, or CO₂-rich off-gas in CO₂-loaded bed stripping). All three effects shift the heat-capacity rate (C = m·Cp) of one or both streams by enough that the LMTD correction factor F_T - or, more honestly, a discretised segment-by-segment numerical integration - has to be applied. Vendor data sheets that quote "F_T = 0.95 - use ideal-gas LMTD" are typically a flag that the regeneration stream was modelled as pure N₂ at constant Cp, regardless of what is actually flowing.

This guide is for the buyer-engineer or plant engineer who has to (a) audit a vendor heat-balance sheet before PO release, (b) verify the area after commissioning when the heater is undersized for the actual gas mixture, or (c) re-rate an existing exchanger for a different regeneration-gas composition. It uses standard equations (LMTD, ε-NTU, F_T from TEMA) with the corrections that actually matter. Worked examples are drawn from 3A/4A/13X molecular sieve dryer regeneration loops, activated alumina H₂O₂ regeneration, and ammonia-cracking PSA tail-gas regeneration.

2. The textbook LMTD baseline and where it starts to fail

The log-mean temperature difference for a counterflow exchanger with constant hot and cold side C-rates is:

LMTD = (ΔT₁ − ΔT₂) / ln(ΔT₁ / ΔT₂)

where ΔT₁ is the inlet hot - inlet cold temperature difference and ΔT₂ is the outlet hot - outlet cold difference. The required area A = Q / (U · LMTD · F_T), with F_T the TEMA correction factor for multi-pass or non-counterflow geometries. For a 1-2 shell pure counterflow exchanger, F_T = 1.0 by definition. This is where vendor data sheets stop.

The first place the textbook formula breaks down is when Cp varies by more than ~5% across the temperature span. For pure N₂ heating from 30 °C to 250 °C, Cp rises from 29.1 J/mol·K to 29.7 J/mol·K (+2%), small enough that the constant-Cp assumption still gives LMTD within 1-2% of the integrated value. For CO₂ (37 J/mol·K at 300 K → 44 J/mol·K at 600 K, +19%) or H₂O(v) at regeneration conditions, the Cp rise is steeper and the cold-stream heat-capacity rate C_cold at the hot end is 15-20% higher than at the cold end. The simple LMTD formula averages over this curvature and understates the area.

The second place it breaks is when the stream composition shifts mid-exchanger - regeneration loops with PSA tail gas (80-85% N₂, 10-15% CO₂, 2-5% CH₄ dry basis, saturated with H₂O at 30-40 °C) pick up additional water from the bed (heat of desorption ~4,000-4,500 kJ/kg H₂O at 200 °C), adding both mass and enthalpy load on the cold side. Standard LMTD assumes steady mass flow and constant composition; this assumption breaks. The third place is at high pressure or moderate temperature where the compressibility factor Z departs from 1.0 by 2-5% (CO₂ at 30 bara, 300 °C has Z ≈ 0.92). At that point, the volumetric flow work (P·V term) becomes a non-negligible fraction of total enthalpy, and the simple Δh = Cp·ΔT approximation drifts from the real stream enthalpy change calculated from a real-gas equation of state.

The fourth place is when the bed itself is a heat source/sink in series with the exchanger. The regenerator bed releases sensible heat (1.0-1.2 kJ/kg·K for activated alumina, 0.85-0.95 kJ/kg·K for molecular sieve) and adsorptive heat (2,800-4,500 kJ/kg H₂O for 13X, 3,500-4,200 kJ/kg H₂O for activated alumina in H₂O₂ service). Almost every vendor data sheet for a regeneration heater assumes a specific arrangement (preheater vs effluent cooler); if the buyer specifies a different arrangement, the heat balance has to be re-run, not just the area re-scaled.

3. Real-gas Cp integration: the 5-segment method

The cleanest replacement for constant-Cp LMTD is to discretise the exchanger into 5-10 axial segments, evaluate Cp(T) at each segment midpoint using a real-gas Cp correlation (REFPROP, ProMax, or HYSYS), and apply segment-by-segment LMTD. For typical regeneration loops, 5 segments give an integrated area within 2-3% of a 20-segment calculation - enough for vendor-data-sheet-grade heat balance.

For pure N₂ at atmospheric regeneration, the Cp variation across 35→280 °C is about +2.1%. The integrated LMTD across 5 segments ends up about 1.5% lower than the constant-Cp simple LMTD, so the required area is about 1.5% larger than the textbook calculation - usually within engineering tolerance.

Where this method actually changes the answer is when the regeneration stream contains CO₂, CH₄, or H₂O at significant partial pressure. A 90/10 N₂/CO₂ mixture heated from 30 °C to 280 °C has a Cp that rises from 30.5 J/mol·K to 35.0 J/mol·K (+14.8%). The integrated LMTD drops 6-7% relative to the constant-Cp calculation, and the required area goes up by the same fraction - the difference between a 90 m² and 96 m² exchanger, about 7% on capital cost.

4. Joule-Thomson and bed heat release: the hidden cold-side load

For high-pressure regeneration loops (PSA feed gas at 7-30 bara dried to -40 °C dewpoint, then used as regeneration gas at near-feed pressure after a heater), the gas undergoes Joule-Thomson cooling across the bed when pressure is reduced from adsorption to regeneration. For pure N₂, the J-T coefficient is +0.25 K/bar at 25 °C - 20-bar ΔP gives 5 K cooling. For CO₂ at the same conditions, the J-T coefficient is +1.1 K/bar at 25 °C and rises to +1.5 K/bar at 200 °C - a 20-bar ΔP drop gives 22-30 K of cooling. If the regeneration gas is fed to the heater at near-adsorption pressure (e.g. 13X dryer at 8 bara), the J-T effect adds a 5-25 K cooling load on the cold side that vendor data sheets often omit.

Combined with the heat of desorption released as the bed strips water (and CO₂, in CO₂-loaded beds), the cold-side enthalpy load at the heater inlet can be 10-20% higher than the simple mass·Cp·ΔT figure. The trade-offs between dry air, nitrogen, and methane regeneration gas for the upstream selection problem are detailed in our activated alumina regeneration gas composition trade-offs guide, which sets the boundary condition for the heater-side calculation covered here. For a 13X bed stripping water at 5 wt% loading to 0.5 wt% at 200 °C bed temperature, the heat of desorption per kg of regeneration gas (assuming 0.8 kg gas per kg water stripped at typical regeneration gas-to-water ratio) is roughly 600-800 kJ/kg gas - which is comparable to the Cp·ΔT load of heating the gas from 30 °C to 250 °C (which is about 600 kJ/kg gas for N₂ at 30 bara). The two loads are similar in magnitude and additive.

The practical consequence for buyers is that a regeneration heater sized only for the gas sensible-heat load will undersize by 50-100% for a CO₂-loaded 13X bed with 7-10 bar adsorption pressure. Vendor data sheets for OEM heater packages often quote "heater duty for sensible heating only" with a footnote that the buyer owns the bed heat release. The correction: add the bed heat release (2,500-4,500 kJ per kg water stripped at 200 °C, scaled by design water load per cycle) to the cold-side enthalpy load, then re-run the LMTD integration with the larger C_cold. The result is a heater area about 1.5-2× the sensible-heat-only estimate.

5. Compositional effects: what mixtures actually do to LMTD

Mixtures of N₂, CO₂, CH₄, and H₂O at regeneration conditions have Cp values that are not simple mole-fraction-weighted averages of pure-component Cp. Real-gas Cp for mixtures depends on the compressibility factor, which itself depends on composition, pressure, and temperature. Deviation from ideal-mixing is small at low pressure (within 1-2% at 1 bara, 300 °C) but grows rapidly at high pressure (5-10% at 30 bara, 200 °C for a 90/10 N₂/CO₂ mixture).

For buyers specifying regeneration-gas composition, the practical takeaway is that CO₂-bearing regeneration streams at 10+ bara require explicit Cp correction — this is the same Cp-correction regime that drives 5A PSA tail-gas recycle bed design, where the tail-gas recycle stream typically contains 10-30% CO₂.; N₂ and H₂-rich streams are within 1% of ideal-gas Cp and don't justify the modelling effort. A 90/10 N₂/CO₂ mixture at 10 bara needs a real-gas Cp correction of +4%, which translates to a 3-4% area increase. Add the J-T and bed-heat corrections on top of that, and the total area uplift is 15-30% over the ideal-gas, sensible-heat-only textbook calculation - which is the magnitude the earlier section flagged as the gap between vendor and real operating data.

6. F_T correction factor and multi-pass arrangements

The TEMA F_T factor accounts for temperature-profile deviation from pure counterflow geometry in 2-4 pass shell-and-tube exchangers or kettle reboilers. For pure 1-2 shell counterflow, F_T = 1.0. For 2-4 shell-and-tube exchangers (TEMA E or F shell), F_T ranges 0.85-0.95 depending on C_R = C_min/C_max.

For regeneration loops, the typical arrangement is a 1-2 shell pure counterflow gas-to-gas exchanger (TEMA E shell) where the cold gas (regeneration inlet, ~30 °C) flows counter to the hot gas (regeneration outlet, ~280 °C). When the heater-side sizing is correct but the cycle still creeps toward longer regeneration times, the diagnostic starts at the bed side — see our 7 root causes of regeneration cycle creep reference for the bed-side failure-mode inventory. F_T ≈ 0.97-0.99 in this geometry and the textbook LMTD with F_T ≈ 0.98 is usually acceptable to within 1-2% of the integrated numerical LMTD for ideal-gas conditions. For arrangements that mix shell-and-tube with cross-flow finned sections (typical for high-pressure feed gas preheaters), F_T can drop to 0.80-0.90 and the deviation from simple LMTD becomes significant - which is where the buyer's auditor should require the F_T value to be quoted explicitly on the vendor data sheet.

For non-counterflow geometries (1-1 kettle, 1-2 cross-flow, or plate-fin exchangers), the standard LMTD formula does not apply at all and the ε-NTU method is required. ε-NTU is numerically equivalent for the same heat-balance result but avoids the geometric ambiguity of LMTD for non-counterflow exchangers. For regeneration loops, plate-fin exchangers are increasingly common (compact, high U-value, lower installed cost) but require ε-NTU analysis from the vendor, not a quoted LMTD area. Buyers reviewing plate-fin quotes should ask for both the ε-NTU and LMTD-equivalent area and verify that the U-value assumed (typically 80-150 W/m²·K for gas-to-gas) is consistent with the exchanger geometry.

7. Pressure-drop correction: hidden in the LMTD calculation

Pressure drop is a parasitic load that does not affect LMTD directly but reduces the available pressure at the bed inlet. For regeneration loops at 1-3 bara, pressure drop is 50-200 mbar and usually negligible. For high-pressure regeneration at 7-30 bara, pressure drop is 200-500 mbar - 1-7% of bed inlet pressure, enough to shift the equilibrium loading slightly and require adjustment of the regeneration gas mass flow.

The pressure-drop correction to LMTD is indirect: it changes Cp (via real-gas effects, section 5), it changes the mass flow distribution if the bed is multi-tube, and it changes the heater duty if downstream of a blower (the blower has to push gas through the higher-pressure-drop system). For buyers auditing vendor quotes, pressure drop should be a separate line item (50-100 mbar for shell-and-tube gas-to-gas, 100-300 mbar for plate-fin). If the vendor quotes only area without a pressure-drop specification, the quote is incomplete.

8. Worked example: 13X molecular sieve dryer, CO₂-loaded bed, 8 bara regeneration

Consider a 13X molecular sieve dryer for natural-gas processing: feed gas 30 bara, 30 °C, saturated with water and 2 mol% CO₂; dried to -60 °C pressure dewpoint and 50 ppmv CO₂; regeneration gas is 5% of feed at 8 bara, heated from 30 °C to 280 °C in a gas-to-gas heat exchanger against the regenerator effluent (300 °C, 8 bara, returning from the bed).

Vendor data sheet (ideal-gas baseline): LMTD = (280 − 30 − (300 − 30)) / ln((280 − 30) / (300 − 30)) = 250 / ln(0.893) ≈ 221 K. Q = m·Cp·ΔT = 0.05 · 1,500 · 0.85 · 250 ≈ 15,938 kW. U = 50 W/m²·K. A = Q / (U · LMTD) = 15,938 / (50 · 221) / 1,000 = 1.44 m² (vendor quotes 145 m² with F_T = 0.98 → 148 m²).

Real-gas correction: Cp for 95/5 N₂/CO₂ at 8 bara, 250 °C, real-gas = 31.2 J/mol·K (vs ideal-mix 30.4), +2.6%. J-T cooling 30→8 bara = -24 K for CO₂, +5 K for N₂, weighted by 5% CO₂ → -2 K on cold-side inlet. Heat of desorption (water + CO₂) = 7.4 kW per kg/s gas = 555 kW. Total Q (corrected) = 16,493 kW (+3.5%). Integrated LMTD with corrected Cp curve = 215 K (vs simple 221 K, -2.7%). Required area = 16,493 / (50 · 215 · 0.98) / 1,000 = 1.57 m².

Final answer: 157 m² vs vendor quote 148 m² — 6% larger. Within the 15-30% envelope flagged earlier, on the lower end because the stream is only 5% CO₂ (more CO₂ would push the gap higher). The buyer should verify the vendor's 148 m² against the corrected 157 m² and either accept the 6% underdesign with documented margin or request a re-rate to 160 m².

9. Worked example: activated alumina H₂O₂ regeneration, atmospheric pressure, pure N₂

Consider an activated alumina polisher for H₂O₂ service: regeneration gas is pure N₂ at 1.05 bara, heated from 25 °C to 220 °C in an electric heater (no gas-to-gas preheat); N₂ flow 800 kg/h; bed strip water from 6 wt% to 0.5 wt% at 200 °C.

Vendor data sheet (ideal-gas baseline): for an electric heater with 90% effectiveness, LMTD ≈ 175 K. Cp for N₂ at 1 bara, 220 °C = 29.2 J/mol·K. Q (sensible) = 880 kW. Q (desorption, water) = 49 kW. Total Q = 929 kW. U = 40 W/m²·K. A = 929 / (40 · 175) = 13.3 m².

Real-gas correction: Cp for pure N₂ at 1 bara, 220 °C is essentially identical to ideal-gas (Z within 0.1%). J-T at 1 bara is negligible. Real-gas correction: zero. Required area remains 13.3 m².

For activated alumina atmospheric regeneration with pure N₂, the textbook LMTD calculation is correct and the vendor data sheet can be accepted without real-gas correction. The audit work shifts from Cp to verifying the U-value (heater geometry, fin density, gas mass velocity) and the heat-of-desorption value (bed water loading profile, strip temperature).

10. Practical checklist for buyers auditing vendor heat-balance sheets

1. Identify the regeneration gas composition by component (N₂, CO₂, CH₄, H₂, H₂O partial pressure) and confirm the vendor data sheet matches. If vendor assumes pure N₂ but the actual stream has 5%+ CO₂ at 5+ bara, the LMTD calculation is wrong by 3-10%.
2. Confirm the regeneration pressure and pressure drop across the heater. Real-gas Cp corrections become significant above 10 bara for CO₂-bearing streams.
3. Verify the bed heat release is included in the cold-side load. Vendor data sheets for OEM heater packages often quote sensible heat only.
4. Check the F_T correction against shell type and pass arrangement. F_T = 0.98 is reasonable for 1-2 counterflow; F_T < 0.90 indicates non-counterflow geometry needing ε-NTU analysis.
5. Request an integrated (5+ segment) LMTD calculation from the vendor for any loop with CO₂ > 5% or pressure > 10 bara. Single LMTD with average Cp is not adequate for these conditions.
6. Specify the pressure-drop allowance as a separate line item (typically 50-200 mbar) and confirm the heater outlet pressure matches the bed inlet requirement.
7. Confirm the heater outlet temperature against the design bed regeneration temperature with a 10-20 K margin for control overshoot and feed composition variability. The bed-side target window is set by the activated alumina ball regeneration temperature optimization curve; the heater outlet must land inside it at all expected mass flows.
8. Verify the U-value against gas mass velocity (G = m/A) and exchanger geometry. Shell-and-tube gas-to-gas: U = 30-80 W/m²·K at G = 5-15 kg/m²·s. Plate-fin: 100-200 W/m²·K at the same G.
9. Document the heat-balance assumption set (Cp source, Z source, J-T coefficient source, heat-of-desorption source, F_T, U-value basis) and require the vendor to match these in the PO.
10. Plan a post-commissioning heat-balance audit at 30, 60, and 90 days of operation. Discrepancy between design and operating heat duty is the first signal that assumptions need revisiting.

11. Re-rating existing exchangers for new gas compositions

When a regeneration loop is being modified for a different gas composition (e.g. switching from fuel-gas-fired regeneration with high CH₄ to H₂-rich regeneration from an electrolyser, or adding CO₂ capture upstream that shifts the regeneration gas composition), the existing heater/cooler package has to be re-rated for the new C-rate and LMTD. The re-rating procedure is:

1. Establish the new gas composition at the heater inlet, including any entrained water from upstream knockout drums.
2. Re-evaluate Cp at the heater inlet and outlet using real-gas Cp correlations or REFPROP/ProMax data for the mixture.
3. Re-evaluate the heat-of-desorption contribution from the new bed loading (which may be different if the upstream adsorbent service has changed).
4. Re-run the 5-segment integrated LMTD with the corrected Cp curve and adjusted C_cold.
5. Compare the new required area to the existing heater area. Within 10% → re-rated without modification (with control retuning). 10-25% larger → supplementary firing or preheater upgrade. > 25% larger → fundamentally undersized, replacement considered.
6. Update the F_T and U-value for the new mass velocity. The gas mass velocity may shift significantly if the new gas has different density (H₂ is 8× less dense than N₂ at the same molar flow, requiring higher volumetric flow for the same mass duty).

For buyers re-rating an existing exchanger, the cost of the re-rating study is small (1-2 engineering days) compared to the cost of replacing an undersized heater. The re-rating report should include all of the items in the buyer's checklist (section 10) and should be filed as part of the operating manual so that future modifications start from a documented baseline.

12. Standards and codes for regeneration-gas heat exchanger design

The applicable standards for shell-and-tube regeneration heat exchangers in molecular sieve / activated alumina service are:

• TEMA (Tubular Exchanger Manufacturers Association) standards for mechanical design, F_T correction factors, and nomenclature
• ASME BPVC Section VIII for unfired pressure vessel integrity (the heater shell if pressure-rated above atmospheric)
• API 660 for shell-and-tube heat exchangers in petroleum and natural-gas service (typically applicable for regeneration heaters in gas processing)
• EN 13445 for unfired pressure vessels in European jurisdictions (often required for European-domiciled buyers)
• ASME B31.3 for process piping connecting the heater to the bed (see also the 13X PSA CO₂ capture TCO comparison analysis for the bed-side carbon-capture economics that often drive the regeneration-gas composition specification in the first place)
• ISO 13706 for petroleum, petrochemical and natural gas industries - air-cooled heat exchangers (if the regeneration loop uses an air cooler after the bed)

For real-gas property data, standard sources are REFPROP (NIST), ProMax, Aspen Plus / HYSYS, and CoolProp (open-source). Vendor data sheets rarely specify the source; auditors should request it and verify the version (REFPROP 10 vs older can differ 1-2% on Cp at high T for CO₂-bearing mixtures).

13. Common pitfalls in regeneration-gas heat-balance audits

The five most common errors seen in vendor and EPC heat-balance sheets are:

1. Sensible-heat-only heater duty with bed heat release added as a footnote, often in non-SI units or as a percentage rather than absolute kW. The buyer has to extract the sensible-heat duty and add the bed heat release explicitly.
2. Pure-component Cp assumption for mixtures, leading to 5-10% underdesign on CO₂-bearing regeneration streams. Always request the mixture Cp source and the mole-fraction-weighted or real-gas value.
3. Constant-pressure vs constant-volume Cp confusion. Cp (constant-pressure) is the correct property for open-system heat exchangers; Cv (constant-volume) is for closed-system. Most vendor data sheets correctly use Cp, but some spreadsheet templates use Cv by mistake, leading to ~30% underdesign (since Cp/Cv ≈ 1.3-1.4 for diatomic gases).
4. Neglect of pressure drop across the heater, which reduces the bed inlet pressure and changes the regeneration gas density. Pressure drop should be quoted as a separate specification, not buried in the heater duty.
5. Bed-outlet temperature assumed equal to heater-outlet temperature without accounting for the bed temperature profile (10-30 K below inlet for a partially-spent bed). The bed outlet sets the LMTD hot-end, not the heater outlet.

For buyers and plant engineers, the audit value of catching these five errors is significant. A 10% under-design on a 1,500 kW heater is 150 kW of shortfall, which at typical regeneration-gas prices translates to $50-200k/year of operating cost (more frequent regeneration cycles, higher heater fuel consumption, or incomplete bed regeneration leading to capacity loss).

14. Procurement specification language

The following specification clauses should be included in any regeneration heater/cooler PO to lock the heat-balance assumptions into the contract:

• "Vendor shall provide a heat-balance calculation using real-gas Cp values from REFPROP version [N] or equivalent. Ideal-gas Cp assumption is not acceptable for streams containing > 5 mol% CO₂ or at pressures > 10 bara."
• "Vendor shall include the heat of desorption for water and CO₂ stripping at the design bed loading profile, integrated across the regeneration cycle, as a separate line item on the cold-side duty."
• "Vendor shall quote the F_T correction factor explicitly for the proposed shell type and pass arrangement. F_T < 0.95 requires ε-NTU analysis as a supplementary calculation."
• "Vendor shall specify the pressure drop across the heater at design mass flow, separately for gas-side and shell-side (if applicable). Pressure drop shall not exceed [N] mbar at design conditions."
• "Vendor shall provide a 5-segment (minimum) integrated LMTD calculation worksheet showing the segment temperatures, Cp values, C-rates, and segment LMTD contributions. Single-LMTD with average Cp is not acceptable."
• "Vendor shall warrant the heater outlet temperature is within ±10 K of the design bed regeneration temperature at design mass flow and inlet temperature. Verification at site acceptance test (SAT) shall use a calibrated thermocouple at the heater outlet."

Including these clauses in the PO shifts the heat-balance audit work from the buyer (after delivery) to the vendor (before quote) and reduces the risk of receiving an undersized heater that has to be re-rated or supplemented after commissioning.

15. Next steps and procurement checklist

If you are sizing a regeneration heater for a molecular sieve or activated alumina dryer and need a vendor-neutral heat-balance review or second-opinion audit, we can help. Our Zibo-based engineering team has produced heat-balance sheets for 3A/4A/5A/13X dryers in natural-gas, hydrogen, and air-separation service, and routinely reviews vendor data sheets against the corrections outlined above. Send your regeneration gas composition (mole fractions or dry-basis + pressure), design cycle parameters (bed loading, regeneration temperature, cycle time), and the existing vendor data sheet (PDF or Excel) to barry@aluminaworld.com or WhatsApp +86 133 2522 2240 and we will return a corrected heat-balance with F_T, Cp correction, and bed heat release included.

Featured Product

Molecular Sieve 13X for CO₂ and Water Co-Adsorption

Aluminaworld 13X beads (1.6-2.5 mm) with LiLSX-grade CO₂ working capacity and 28-30 wt% equilibrium water capacity at 30 °C / 8 bara. Optimised bead crush strength for high-pressure regeneration loops covered in this guide.

View Product Details →

Frequently Asked Questions