Estimating Heat Exchanger Area from Duty, U, and Temperature Difference

Every heat-exchanger sizing exercise reduces to one compact relationship: heat transferred equals an overall coefficient, times the transfer area, times the mean temperature difference. Rearranged, the required area is A = Q / (U · ΔT_lm) — the heat duty Q divided by the overall heat-transfer coefficient U and the log mean temperature difference ΔT_lm. This article defines each of those three inputs, works a complete example drawn from a published design case, and sets out when the shortcut can be trusted and when it needs correction.

The sizing equation and its three inputs

Heat crosses an exchanger’s wall wherever one fluid is hotter than the other, and the rate at each point is proportional to the local temperature difference. Because that difference changes along the exchanger — wide where the hot fluid enters and narrow where it leaves, or the reverse — sizing the whole unit means integrating duty against driving force. The Heat Exchanger Design Handbook gives the general result as the integral of dQ / [U(T_h − T_c)] taken along the exchanger, where T_h and T_c are the hot- and cold-side temperatures at each position. When U and the fluids’ heat capacities are approximately constant and neither fluid changes phase, that integral collapses to the familiar form:

A = Q / (U · ΔT_lm)

The units must agree: a U quoted in kW/(m²·K) pairs with a duty in kW, while a U in W/(m²·K) requires the duty in W. Mixing the two is one of the most frequent computational slips.

Heat duty: Q

The duty normally comes from an energy balance on the process stream. For sensible heating or cooling of a single-phase fluid, Q = ṁ · C_p · (T_in − T_out), where ṁ is the mass flow rate and C_p the specific heat. For condensation or boiling, the latent heat replaces the sensible term. The duty computed on the hot side should match the duty on the cold side; a mismatch signals an error in flows, temperatures, or assumptions before any sizing begins.

Overall heat-transfer coefficient: U

U condenses every resistance between the two streams — the convective films on each side, conduction through the wall, and any fouling deposits — into a single coefficient. Designers usually assume a value on the first pass, drawn from experience with similar fluid pairs and exchanger types, and then verify it: the check uses heat-transfer correlations that require the Reynolds and Nusselt numbers (dimensionless measures of flow regime and convective transfer) of both streams, iterating until the assumed and calculated U agree, as described in a University of Delaware design case study. In that study, a first-iteration U of 1.5 kW/(m²·K) was used for cooling a water-like process stream with chilled water in a double-pipe exchanger. Because fouling, low velocities, or viscous fluids can reduce U dramatically, an optimistic assumption here becomes an area error later.

Mean temperature difference: ΔT_lm

Because the driving force differs from one end of the exchanger to the other, a single representative value is needed, and it is not the simple average of the two end differences but a logarithmic one. The log mean temperature difference (LMTD) is:

ΔT_lm = (ΔT₁ − ΔT₂) / ln(ΔT₁ / ΔT₂)

Here ΔT₁ and ΔT₂ are the temperature differences at the two ends of the exchanger. As the KEVOS series “Heat Exchangers by the Numbers” explains, the log mean is the single value that, multiplied by U and the area, reproduces the true duty, and the choice of mean matters most exactly where exchangers are hardest: close approaches and lopsided ends.

Which ends those are depends on the flow arrangement. In counterflow (countercurrent) operation the two fluids run in opposite directions, so each end difference pairs a hot-stream temperature with the cold-stream temperature at its opposite extreme; this keeps the driving force spread across the whole exchanger, and counterflow is almost always preferred. In parallel (co-current) flow both fluids enter at the same end and the driving force collapses toward the shared outlet, giving a smaller mean for the same stream temperatures. Counterflow is also the only arrangement that permits a temperature cross, in which the cold fluid leaves hotter than the hot fluid leaves. The approach — the smallest gap the two streams reach, usually at an outlet end — is the number that chiefly drives area: shrinking it demands disproportionately more surface.

When the two end differences happen to be equal, the log mean and the arithmetic mean coincide; the further the ratio ΔT₁/ΔT₂ departs from one, the further the log mean falls below the plain average.

Worked example: area for a 100 L/min cooling duty

The following example, from the Delaware design case study, sizes a cooler for 100 L/min of a water-like process stream from 80 °C to 50 °C, using chilled water available at 5 °C. Both streams are treated as water-like (density ≈ 1 kg/L, C_p ≈ 4.18 kJ/(kg·K)), and the first-iteration design assumes U = 1.5 kW/(m²·K) in a counterflow double-pipe exchanger — a smaller pipe inside a larger one, one fluid in each:

1. Duty. The process flow is 100 kg/min ≈ 1.67 kg/s, so Q = 1.67 × 4.18 × (80 − 50) ≈ 209 kW.
2. Utility temperatures. The balanced design uses 46 L/min of chilled water entering at 5 °C and leaving at 70 °C. Energy-balance check on the utility side: 0.77 kg/s × 4.18 × 65 ≈ 209 kW — the two sides of the exchanger agree.
3. End differences (counterflow). At the hot end, 80 − 70 = 10 K; at the cold end, 50 − 5 = 45 K.
4. LMTD. ΔT_lm = (45 − 10) / ln(45/10) = 35 / 1.504 ≈ 23.3 K.
5. Area. A = 209 / (1.5 × 23.3) ≈ 6.0 m².

The technically feasible design proposed on that basis was a double-pipe exchanger with 120 ft of 2-in. pipe inside a 3-in. outer pipe, consistent with the 6.0 m² estimate. Note how lopsided the end differences are (10 K versus 45 K): the arithmetic mean, 27.5 K, would imply only about 5.1 m² — roughly 15% below the area actually required. The KEVOS series likewise documents a heat-recovery case in which an arithmetic mean left the duty 12% short. Errors of this scale are common whenever one end difference is much smaller than the other.

When the estimate can be trusted

The shortcut A = Q / (U · ΔT_lm) rests on assumptions worth checking before relying on the number:

• Constant U and constant heat capacities. The Design Handbook notes that these can usually be assumed in single-phase duty with no phase change, which is what reduces the general integral to the LMTD form.
• One side changes phase. In a steam-heated exchanger, the condensing-steam temperature is almost constant throughout, so the hot side holds essentially one temperature and the mean difference is easy to evaluate for the other fluid.
• Multipass shell-and-tube geometries. Real shell-and-tube exchangers route fluids through repeated passes, so the true mean difference is below the pure counterflow value; a correction factor below one is applied to the LMTD, and established quick manual design procedures for shell-and-tube exchangers estimate the approximate size, shell arrangement, and dimensions needed before detailed rating calculations.
• Strongly varying properties. Where viscosities, heat capacities, or U vary appreciably along the exchanger, the integral must be evaluated piecewise. Today such exchangers are “generally designed with the aid of sophisticated computer programs,” as the Handbook puts it — but its authors add that it remains worth considering the fundamentals, and the manual estimate is still how designers frame and sanity-check the problem.

Common mistakes

• Using the arithmetic mean or the inlet difference alone. Both overstate the driving force and undersize the area, especially when the end differences are lopsided.
• Inconsistent units. Mixing kW with W, or volume flow with mass flow, distorts the result immediately; convert L/min to kg/s using the density first.
• An optimistic U. Assuming a clean, high coefficient with no fouling allowance or velocity check shifts the error into service, where the built exchanger underperforms.
• A one-sided energy balance. The area estimate is only as good as the temperatures feeding it; both streams must close on the same duty, as the worked example demonstrates.
• Applying the LMTD uncorrected. Multipass shell-and-tube units need correction factors; using the pure counterflow value directly overstates the effective mean difference.

From estimate to workable design

The area from A = Q / (U · ΔT_lm) is a first iteration, not a finished design. From it, a designer proposes a concrete geometry — for the example above, a length and diameter of concentric pipe — and then closes the loop: compute Reynolds and Nusselt numbers for both streams, recalculate U from correlations, and revise the assumed value until assumption and calculation agree, yielding what the Delaware case calls a technically feasible design. For shell-and-tube work, the manual design procedures expanded from Bell’s Heat Exchanger Design Handbook method take the estimate further, to approximate dimensions and multiple-shell arrangements suitable for entering detailed rating calculations. The practical sequence, then, is: fix the duty with a balanced energy balance, favor a counterflow arrangement with a realistic approach, assume a defensible U, size the surface with the log mean temperature difference — and then verify every one of those choices.