Heat Exchanger LMTD Calculator: Counterflow vs Parallel Flow
Calculate the Log Mean Temperature Difference (LMTD) for parallel and counterflow heat exchangers. Compute end temperature differences, apply correction factors, and solve Q = U × A × ΔT relationships.
How do you average a temperature difference that varies along the length of a heat exchanger? Not the arithmetic mean. The log mean. ΔT_lm = (ΔT₁ − ΔT₂) / ln(ΔT₁/ΔT₂), where ΔT₁ and ΔT₂ are the temperature differences at the two ends. The log mean falls out of integrating ΔT(x) along the length when both fluid streams obey simple energy balances, and it's the right number to plug into Q = U·A·ΔT_lm for sizing. Counterflow and parallel flow share the same LMTD formula but use different end-pair definitions, and counterflow always gives a higher ΔT_lm for the same inlet/outlet temperatures. That's why almost every modern shell-and-tube exchanger is wired counterflow, not parallel.
Counterflow consistently delivers higher LMTD because temperature differences stay more uniform along the exchanger length. In parallel flow, both fluids enter hot-vs-cold at one end, creating a large initial ΔT that rapidly collapses; you end up with one huge temperature difference and one tiny one, which averages down logarithmically. Counterflow spreads the driving force more evenly, letting you transfer the same heat with less area (or more heat with the same area). If your design target is a tight approach temperature (cold outlet approaching hot inlet), counterflow is mandatory; parallel flow physically can't get T_c,out above T_h,out.
The Four Modes: Conduction, Convection, Radiation, Phase Change
An LMTD calculation rolls four physical mechanisms into one number U. Strip U apart and the modes show up explicitly. Inside-tube convection (forced, often turbulent, h on the order of 1000 to 5000 W/m²K for water; 50 to 300 W/m²K for viscous oil). Conduction through the tube wall (k = 16.2 W/mK for 304 stainless, 60 W/mK for carbon steel, 400 W/mK for copper). Outside-tube convection (shell side, complicated by baffles and crossflow). Phase change at the surface if either fluid is condensing or boiling, which can multiply h by 5 to 50.
Conduction (tube wall)
Cylindrical wall: R_wall = ln(r_o/r_i) / (2πkL). For thin tubes the wall resistance is small versus film resistance, but in stainless or titanium tubing it can carry 10 to 20% of the total thermal resistance and shouldn't be ignored.
Forced convection (both sides)
Inside-tube h from Dittus-Boelter or Gnielinski. Shell side from Bell-Delaware or Kern method. Water at 1 m/s in 20 mm tubes gives h ≈ 4500 W/m²K; the same geometry with oil at 1 m/s gives 200 to 400.
Radiation (high-temp only)
Negligible below ~200°C surface temperature, but in fired heaters, cracking furnaces, and concentrated-solar receivers, radiation outpaces convection. εσ(T⁴ - T_surr⁴) with ε = 0.85 for oxidized steel.
Phase change
Condensing steam: h = 5,000 to 15,000 W/m²K (Nusselt). Nucleate pool boiling of water: h up to 50,000. These coefficients dwarf single-phase film coefficients but only along the section where phase change actually occurs.
The overall coefficient combines all four through 1/U = 1/h_i + R_wall + 1/h_o (plus fouling). Whichever resistance is largest controls the answer. Doubling h on the water side rarely helps if the oil side is the bottleneck.
Thermal Resistance Networks: Stacking Materials Correctly
The overall heat-transfer coefficient is just a series resistance network. Per unit area (referenced to the outside surface):
1/U_o = 1/h_o + R_f,o + (r_o · ln(r_o/r_i))/k_wall + (r_o/r_i) · R_f,i + (r_o/r_i) / h_i
Each term is a resistance in m²K/W referenced to the outside area A_o.
Clean-surface U values from correlations are optimistic. In real exchangers, deposits accumulate: scale from hard water, biofilm in cooling towers, coke in refinery heaters, corrosion products everywhere. TEMA publishes fouling resistances (R_f in m²K/W) that you add to the overall resistance:
1/U_dirty = 1/U_clean + R_f,hot + R_f,cold
Typical fouling resistances: clean water ~0.0001, treated cooling water ~0.0002, river water ~0.0003 to 0.0005, heavy fuel oil ~0.0009 m²K/W.
Correction factor F for shell-and-tube
Real shell-and-tube exchangers aren't pure counterflow. Baffles create cross-flow zones, and multi-pass arrangements mean some fluid paths run parallel. The correction factor F (0 < F ≤ 1) handles this deviation: ΔT_lm,eff = F × LMTD_counterflow. F depends on two dimensionless ratios:
P (thermal effectiveness): P = (T_c,out − T_c,in) / (T_h,in − T_c,in)
R (capacity rate ratio): R = (T_h,in − T_h,out) / (T_c,out − T_c,in)
Look up F from TEMA charts for your shell/tube pass configuration (1-2, 2-4, etc.).
F warning: If F drops below ~0.75, your exchanger configuration is thermally inefficient. You're fighting the flow arrangement. Either switch to true counterflow (if mechanically feasible) or use the NTU-effectiveness method, which handles low-F situations more gracefully. The calculator does NOT compute F automatically; supply it from charts or correlations.
Steady-State vs. Transient: When Each Applies
When LMTD applies cleanly
Yes, use LMTD when:
- Steady-state operation (no startup transients)
- Single-phase fluids (no boiling or condensation)
- Constant U along exchanger length
- Negligible heat loss to surroundings
- Pure counterflow or parallel flow (or known F factor)
No, LMTD breaks down when:
- Fluid undergoes phase change (condenser, evaporator)
- U varies sharply along length
- Transient operation (batch heating, startup)
- Temperature cross occurs (parallel-flow limitation)
- F factor drops below 0.75 (use NTU-effectiveness instead)
The LMTD derivation integrates dQ = U·dA·ΔT assuming constant U. If your heat-transfer coefficient changes dramatically (say, near a phase boundary where vapor suddenly condenses) the "log mean" loses physical meaning. For condensers where steam enters superheated and exits as subcooled liquid, you'd zone the exchanger into desuperheating, condensing, and subcooling sections, each with its own LMTD calculation.
Edge case: ΔT₁ = ΔT₂ (balanced flow)
When ΔT₁ exactly equals ΔT₂, the LMTD formula (ΔT₁ - ΔT₂) / ln(ΔT₁/ΔT₂) becomes 0/0: mathematically indeterminate. Physically, this happens in counterflow exchangers with perfectly balanced capacity rates (ṁ_h·c_p,h = ṁ_c·c_p,c). The temperature profiles run parallel and ΔT is constant everywhere.
Limiting case: As ΔT₁ → ΔT₂, LMTD → ΔT₁ (arithmetic mean)
The calculator handles this automatically using L'Hôpital's rule or a small-difference approximation. If you see LMTD ≈ ΔT₁ ≈ ΔT₂ in your results, you're in the balanced-flow regime, which is actually an ideal operating point: maximum heat transfer per unit area because the driving force is uniform. Don't panic when the logarithm misbehaves within 1 to 2%; just take the arithmetic mean.
NTU-effectiveness alternative
LMTD shines when you know all four terminal temperatures. In rating problems (where you know inlet temperatures and want to find outlets) you'd need to iterate. The NTU-effectiveness method sidesteps this: given UA, flow rates, and inlet temperatures, it directly computes heat transfer and outlet temperatures without iteration.
NTU (Number of Transfer Units): NTU = UA / C_min
Capacity rate ratio: C_r = C_min / C_max
Effectiveness (counterflow): ε = [1 - exp(-NTU(1 - C_r))] / [1 - C_r·exp(-NTU(1 - C_r))]
Heat transfer: Q = ε·C_min·(T_h,in - T_c,in)
Use NTU-effectiveness when (1) outlet temperatures are unknown, (2) F factor is too low for reliable LMTD, or (3) you're comparing exchanger performance across a range of operating conditions. Both methods give identical results when applied correctly; they're just different entry points into the same physics.
Material Properties That Drive the Answer (k, α, c, ε)
Tube wall conductivity k decides how much resistance the metal contributes. Specific heat c_p of each fluid sets the capacity rate C = ṁc_p that anchors the energy balance Q = ṁc_p·ΔT. Thermal diffusivity α_diff governs transient response in startup and shutdown. Surface emissivity ε matters only for fired and high-temperature exchangers.
| Tube material | k (W/mK) | Typical use |
|---|---|---|
| Copper (C12200) | 390 to 400 | HVAC condensers, refrigerant lines |
| Admiralty brass | 110 | Power-plant condensers (legacy) |
| Carbon steel | 50 to 60 | General process service |
| 304/316 stainless | 14 to 17 | Corrosive duty, sanitary service |
| Titanium Gr. 2 | 17 | Seawater service, chlorides |
Stainless and titanium have nearly identical k around 16 W/mK, an order of magnitude below copper. For a 16 mm OD × 1.5 mm wall tube, the wall resistance in stainless is about 9.4×10⁻⁵ m²K/W referenced to outside area; in copper it's about 3.8×10⁻⁶. With water-side h around 4500 and air-side h around 50, the wall is barely 1% of total in copper but can creep to 5% in stainless. Worth tracking.
Fluid properties at the design point
Use bulk-mean temperature properties for c_p and at film temperature for k, μ, Pr. Water c_p ≈ 4186 J/kgK and varies less than 1% over normal exchanger ranges. Light hydrocarbons run 2200 to 2500 J/kgK. Heavy oils sit around 1900. Glycol/water mixtures shift c_p with concentration; 50/50 ethylene glycol is roughly 3300 J/kgK. The capacity rate C = ṁc_p is what really drives the temperature change of each stream, so getting c_p right matters more than getting U exactly right.
Real-World Conditions: Cold Bridges, Edge Effects, Surface Resistance
Lab values for U are clean, isothermal, and well-behaved. Field exchangers leak heat to the surroundings, foul over time, run with maldistributed flow on the shell side, and develop dead zones behind baffles. The corrections are routine but they have to be applied.
Cold bridges in exchanger work usually mean external heat losses through unlagged piping, support saddles, or partially insulated bonnets. A bare 6" flange on a hot exchanger can radiate and convect 200 to 400 W to surroundings, which shows up as a Q mismatch between the two streams when you do a thermal balance. If hot-side ΔT × ṁc_p doesn't match cold-side ΔT × ṁc_p within 5%, suspect heat loss before you suspect instrumentation error.
Edge effects on the shell side come from baffle leakage. The Bell-Delaware method partitions shell-side flow into ideal cross-flow plus four leakage/bypass streams (A through F): tube-baffle leakage, baffle-shell leakage, bundle bypass, pass-partition bypass, and the cross-flow stream itself. A poorly clearance-controlled bundle can leak 20 to 40% of the shell flow around the bundle, dropping h_o by a similar fraction. That's a U deficit you only catch by running Bell-Delaware or equivalent simulator output.
Design margin for fouling: A fouling resistance of 0.0003 m²K/W on each side can cut U by 20 to 40% depending on the base coefficient. Size for the fouled condition, not the clean condition; otherwise performance degrades within weeks of startup and operations starts cleaning the bundle on a tighter schedule than the maintenance budget allows.
Limitations and Assumptions
- The calculator computes LMTD and Q = U·A·ΔT_lm relationships for educational purposes. It assumes ideal conditions: constant U, no fouling unless you build it into U_clean, no phase change, steady state.
- Real heat-exchanger sizing requires fouling factors, pressure-drop calculations, tube layout, baffle spacing, vibration analysis, and mechanical design per TEMA and ASME Section VIII.
- The calculator does NOT compute correction factor F. Supply it from TEMA charts or correlations based on your shell/tube pass arrangement (1-2, 2-4, etc.).
- For industrial heat-exchanger specification, consult qualified process or mechanical engineers with access to professional simulation software (HTRI Xchanger Suite, Aspen EDR, or equivalent).
Worked Example: Counterflow Water-Water Exchanger, 100 kW Duty
Scenario: A counterflow water-water exchanger transfers heat from a hot stream entering at 80°C and exiting at 50°C to a cold stream entering at 20°C and exiting at 40°C. Estimated overall coefficient U = 1500 W/m²K (typical for plate-and-frame water-water service). Heat duty Q = 100 kW. Find LMTD and required heat-transfer area.
Step 1: ΔT₁ and ΔT₂ for counterflow
ΔT₁ = T_h,in - T_c,out = 80 - 40 = 40°C
ΔT₂ = T_h,out - T_c,in = 50 - 20 = 30°C
Step 2: LMTD
LMTD = (40 - 30) / ln(40/30)
LMTD = 10 / ln(1.333)
LMTD = 10 / 0.2877 = 34.76°C ≈ 34.4°C (rounded)
Step 3: Required area (assume F = 1 for pure counterflow)
A = Q / (U × LMTD) = 100,000 / (1500 × 34.4)
A = 100,000 / 51,600 = 1.94 m²
Step 4: Sanity check via energy balance
Hot side ΔT = 30°C → ṁ_h·c_p,h = 100,000/30 = 3333 W/K
Cold side ΔT = 20°C → ṁ_c·c_p,c = 100,000/20 = 5000 W/K
Capacity rate ratio C_r = 3333/5000 = 0.667
Hot side has lower c_p·ṁ, so hot side ΔT is larger, consistent with the inputs.
Result: Required clean-surface area is roughly 1.94 m². Add 15% margin for fouling and tolerance buildup, giving a design target of about 2.2 m². For a plate-and-frame unit with 0.3 m² per plate, that's 8 plates plus a couple of extras for thermal margin.
Reality check on U: Plate-and-frame water-water service typically runs U = 3000 to 7000 W/m²K thanks to high-turbulence corrugated plates. The 1500 W/m²K used here is conservative; if your real U sits at 4000, the same duty needs only 0.73 m². Conversely, shell-and-tube water-water without enhancement runs U = 800 to 1500. Pick U from the actual exchanger type, not generic textbook values.
References
- TEMA (Tubular Exchanger Manufacturers Association). Standards of the Tubular Exchanger Manufacturers Association (10th ed., 2019). Defines shell-and-tube classes (R = refinery, C = chemical, B = general), nomenclature (BEM, AES, BFM), mechanical rules, fouling-resistance tables, and F-factor charts for correction factors.
- ASME Boiler and Pressure Vessel Code, Section VIII Division 1. Pressure-vessel code governing shell and head design, tube-to-tubesheet joints, and material selection for pressure-containing components in heat exchangers.
- HEI (Heat Exchange Institute). Standards for surface condensers, closed feedwater heaters, and other power-industry exchangers.
- Incropera, F. P. & DeWitt, D. P. Fundamentals of Heat and Mass Transfer (8th ed., 2017). Wiley. Standard textbook derivation of LMTD, NTU-effectiveness, and Bell-Delaware methods (Chapter 11).
- Holman, J. P. Heat Transfer (10th ed., 2010). McGraw-Hill. Compact engineering reference with worked examples on multi-pass exchangers, condensers, and reboilers.
- Bird, R. B., Stewart, W. E. & Lightfoot, E. N. Transport Phenomena (2nd ed., 2007). Wiley. Rigorous derivation of the convective and conductive transport equations behind every h and U.
- Kern, D. Q. Process Heat Transfer (1950, still in print). McGraw-Hill. Classic reference for shell-and-tube design, the original Kern shell-side method, and pre-computer-era correction factors.
- Kakaç, S., Liu, H., & Pramuanjaroenkij, A. (2020). Heat Exchangers: Selection, Rating, and Thermal Design (4th ed.). CRC Press. The practitioner reference for ε-NTU, LMTD with F correction, and the trade-off between counterflow vs. parallel-flow vs. shell-and-tube configurations during selection.
- Bell, K. J. Bell-Delaware Method for shell-side heat-transfer and pressure-drop calculations. The modern industry method behind HTRI and Aspen EDR shell-side modules.
- HTRI Design Manual. Heat Transfer Research, Inc. The proprietary correlations and methods used by HTRI Xchanger Suite for industrial exchanger design.
Troubleshooting LMTD and Heat Exchanger Sizing Errors
Real questions from engineers stuck on ΔT definitions, correction factors, fouling, and why calculated duty doesn't match reality.
What is the log mean temperature difference (LMTD)?
LMTD is the average temperature difference between two streams in a heat exchanger, weighted to account for the fact that ΔT changes along the exchanger length. The formula is LMTD = (ΔT₁ − ΔT₂) / ln(ΔT₁/ΔT₂), where ΔT₁ and ΔT₂ are the temperature differences at the two ends. For counterflow, ΔT₁ = T_h,in − T_c,out and ΔT₂ = T_h,out − T_c,in. For parallel flow (both streams entering the same end), ΔT₁ = T_h,in − T_c,in and ΔT₂ = T_h,out − T_c,out. Different configurations give different LMTDs even at the same inlet and outlet temperatures. Example: hot water cools 90°C → 60°C while cold water heats 25°C → 50°C in counterflow. ΔT₁ = 40°C and ΔT₂ = 35°C. LMTD = (40 − 35)/ln(40/35) = 37.4°C. Parallel flow with the same temperatures gives ΔT₁ = 65, ΔT₂ = 10, LMTD = 24.0°C. Counterflow is always at least as efficient as parallel flow for the same surface area. Required heat transfer area follows Q = U · A · LMTD · F, where U is the overall heat transfer coefficient and F is a correction factor for cross-flow and shell-and-tube geometries (F = 1 for pure counter or parallel flow). LMTD assumes constant U and constant fluid properties along the exchanger. For phase changes or large temperature swings where U varies, divide the exchanger into zones.
My LMTD came out negative—did I enter the temperatures backward or is something else wrong?
Negative LMTD means either ΔT₁ or ΔT₂ is negative, which indicates a temperature cross: the cold fluid outlet is hotter than the hot fluid at that location. This is physically impossible in parallel flow (T_c,out can't exceed T_h,out) and indicates an error. Check that T_h,in > T_h,out (hot fluid cools) and T_c,out > T_c,in (cold fluid warms). Also verify you're using the right ΔT definitions for your flow arrangement.
I got LMTD = 35°C for counterflow but my colleague got 28°C for the same temperatures—turns out we used different ΔT definitions. How do I know which is correct?
The ΔT definitions depend on flow arrangement. For counterflow: ΔT₁ = T_h,in − T_c,out and ΔT₂ = T_h,out − T_c,in. For parallel flow: ΔT₁ = T_h,in − T_c,in and ΔT₂ = T_h,out − T_c,out. Using counterflow definitions on a parallel flow exchanger (or vice versa) gives the wrong LMTD. Check your P&ID to confirm which end the streams meet—that determines the correct formula.
The formula blows up when I try ΔT₁ = ΔT₂ = 30°C—is this a bug or am I supposed to handle it differently?
When ΔT₁ = ΔT₂, the formula (ΔT₁ − ΔT₂)/ln(ΔT₁/ΔT₂) becomes 0/0, which is indeterminate. Physically this means balanced capacity rates: both fluids change temperature by the same amount and the driving force is constant. The limiting value is LMTD = ΔT₁ = ΔT₂ (arithmetic mean). Most calculators handle this automatically using L'Hôpital's rule or a small-difference approximation.
I'm getting F = 0.65 from the TEMA charts for my 1-2 shell-and-tube—should I redesign or just use it?
F < 0.75 is a red flag. It means your exchanger configuration is fighting itself—some fluid paths run nearly parallel flow while others run counterflow, wasting heat transfer area. Consider: (1) adding more shell passes to approach true counterflow, (2) increasing tube passes to even out the flow pattern, or (3) using a different exchanger type entirely. Using F = 0.65 means you need 35% more area than pure counterflow to transfer the same heat.
My U value is supposed to be 500 W/m²·K but after six months the exchanger barely hits 60% of design duty—what went wrong?
Fouling. Deposits accumulate on heat transfer surfaces: scale from hard water, biofilm, corrosion products, or process residue. Each layer adds thermal resistance: 1/U_dirty = 1/U_clean + R_f,hot + R_f,cold. A fouling resistance of 0.0003 m²·K/W on each side can cut U by 25–40%. TEMA publishes fouling factors by fluid type. Design for the fouled condition, not clean, and schedule regular cleaning.
I have inlet temperatures but not outlets—can I still use LMTD, or do I need a different method?
LMTD requires all four terminal temperatures. If you only know inlets, you'd need to iterate: guess outlets, calculate LMTD and Q, check energy balance, adjust, repeat. The NTU-effectiveness method is better for this 'rating' problem: given UA, flow rates, and inlet temperatures, it directly computes outlets without iteration. Both methods give identical results when applied correctly—NTU-effectiveness is just more convenient when outlets are unknown.
I sized my exchanger for Q = 200 kW but it's only delivering 150 kW in operation—besides fouling, what else could cause this?
Several possibilities: (1) actual flow rates differ from design (measure them), (2) fluid properties changed (temperature, concentration, viscosity), (3) air or non-condensables trapped on shell side (common in condensers), (4) bypass flow around baffles (poor manufacturing or erosion), (5) partial tube plugging. Check that inlet temperatures match design—even 5°C difference changes duty significantly because it affects LMTD.
The textbook says counterflow is always better, but my plant runs a parallel flow cooler—is it actually wrong?
Counterflow gives higher LMTD (more heat transfer per area) and allows closer approach temperatures, but parallel flow has legitimate uses: (1) when you need to limit cold fluid temperature rise (parallel flow outlet ΔT is always positive), (2) when tube wall temperature limits matter (parallel flow keeps metal cooler at the hot end), (3) when flow distribution is more important than thermal efficiency. It's application-dependent, not universally wrong.
My condenser has steam entering at 150°C saturated and leaving as 100°C subcooled—can I use one LMTD for the whole thing?
Not accurately. Phase change creates discontinuities in the heat transfer coefficient and temperature profile. Zone the exchanger: (1) desuperheating zone (if inlet is superheated), (2) condensing zone (isothermal at saturation), (3) subcooling zone (liquid cooling). Calculate LMTD and area for each zone separately, then sum. Using a single LMTD averages over these regimes and typically undersizes the subcooling section.
I'm comparing two exchangers with the same LMTD but different U values—which one transfers more heat?
The one with higher U transfers more heat per unit area (Q = U·A·LMTD). If you're comparing two units, look at UA product: higher UA = more heat transfer regardless of how it's achieved (bigger area, better coefficients, or both). If buying equipment, compare $/kW transferred including operating costs (pumping power for pressure drop, maintenance for fouling). LMTD alone doesn't tell you which exchanger is 'better.'