Heat exchangers: NTU method
NTU (Number of Transfer Units) method for sizing/rating heat exchangers using effectiveness–NTU relations for parallel, counterflow, and crossflow configurations.
Drafted with Aria, reviewed by the AiCanCode.org team · Updated . How we write and check our notes. Spotted an error? Use Give Feedback at the bottom of the page.
Why it matters
Heat exchangers are everywhere: condensers in power plants, radiators in vehicles, economizers in boilers, air coolers in process plants. The NTU method lets you size a heat exchanger or find outlet temperatures when one or both outlet temperatures are unknown — a frequent design and troubleshooting need where the LMTD method is cumbersome.
Key ideas
- A heat exchanger transfers energy between two fluid streams through a wall. We analyze it assuming steady state, negligible heat loss to surroundings, no internal heat generation, constant properties, and a uniform overall heat transfer coefficient U over the area A.
- Define capacity (heat capacity) rates: Capacity rate C of a fluid is
C = ṁ · cp(W/K), whereṁis mass flow rate andcpis specific heat at constant pressure. - The smaller of the two capacity rates is Cmin and the larger is Cmax. The capacity ratio is
Cr = Cmin / Cmax(0 ≤ Cr ≤ 1). Many limits and formulas use Cr. - The largest possible heat transfer between the two streams, if one stream could be cooled/heated all the way to the inlet temperature of the other, is
qmax = Cmin · (Th,in − Tc,in). - Effectiveness ε is the fraction of the maximum possible heat transfer actually achieved:
ε = q / qmax. By definition,0 ≤ ε ≤ 1. - Number of Transfer Units (NTU) measures heat transfer area relative to the minimum capacity rate:
NTU = U · A / Cmin. It is dimensionless. Larger NTU usually means larger area or larger U and thus higher effectiveness. - Why NTU method: In design/sizing problems, outlet temperatures are unknown, but U and A (or U and target ε) are specified/estimated. The NTU–ε relations provide q and outlet temperatures directly from NTU and Cr without solving coupled energy differential equations for each configuration.
Assumptions (unless stated otherwise):
- Steady state, no heat loss to surroundings (adiabatic external surface).
- No axial (streamwise) conduction in fluids or wall; negligible kinetic/potential energy changes.
- Constant
cp, constantUover the area, uniform flow distribution. - No phase change unless using the special Cr → 0 model.
- Sign convention: Take
qas positive from hot to cold fluid; temperature drops on the hot side and rises on the cold side are positive magnitudes.
Energy balance and outlet temperatures:
- For any exchanger under the above assumptions:
q = Ch (Th,in − Th,out) = Cc (Tc,out − Tc,in). - Combine with ε to get
q = ε · qmax = ε · Cmin (Th,in − Tc,in); then compute outlets from the energy balance.
Configuration-dependent ε–NTU relations (standard results)
- Parallel flow (co-current):
- Exact relation:
ε = [1 − exp(−NTU · (1 + Cr))] / (1 + Cr). - Limiting behavior: as
NTU → ∞,ε → 1 / (1 + Cr)(parallel flow cannot reach ε = 1 unlessCr → 0). ForCr → 0,ε = 1 − exp(−NTU).
- Exact relation:
- Counterflow:
- Exact relation:
ε = [1 − exp(−NTU · (1 − Cr))] / [1 − Cr · exp(−NTU · (1 − Cr))]. - Limiting behavior: as
NTU → ∞,ε → 1for anyCr < 1; forCr = 1,ε = NTU / (1 + NTU). - Counterflow allows temperature cross (Tc,out can exceed Th,out) when
Cr < 1— physically permissible.
- Exact relation:
- Crossflow:
- “Both fluids unmixed” (e.g., finned passages that prevent mixing in both streams):
- Common correlation:
ε = 1 − exp{ − (1/Cr) · [1 − exp(−Cr · NTU)] }. - Checks: for
Cr → 0,ε → 1 − exp(−NTU). AsNTU → ∞,ε → 1 − exp(−1/Cr)(less than unity for finite Cr).
- Common correlation:
- “One fluid mixed, other unmixed”:
- Common correlation:
ε = (1/Cr) · [1 − exp{ −Cr · (1 − exp(−NTU)) }]. - Checks: for
Cr → 0,ε → 1 − exp(−NTU); asNTU → ∞,ε → (1/Cr) · [1 − exp(−Cr)](lower than both-unmixed case).
- Common correlation:
- “Both fluids unmixed” (e.g., finned passages that prevent mixing in both streams):
Special/standard cases:
- Phase change on one side at essentially constant temperature (e.g., condensing or boiling) corresponds to
C → ∞on that side, henceCr = 0. In both parallel and counterflow, the effectiveness reduces toε = 1 − exp(−NTU). - Upper bounds:
- Parallel:
εmax = 1 / (1 + Cr) < 1. - Counterflow:
εmax → 1forCr ≤ 1asNTU → ∞. - Crossflow:
εmaxbelow counterflow for givenCrandNTU; both-unmixed performs better than one-mixed.
- Parallel:
Outline of where ε–NTU relations come from (no derivation details):
- Start from differential energy balances along the flow direction(s):
d(Th − Tc)/dxproportional to local U, perimeter, and inverse of capacity rates. - Non-dimensionalize temperatures by
(Th,in − Tc,in)and length by NTU; solve the resulting first-order linear ODEs with appropriate boundary conditions for each flow arrangement. - Integrate to obtain outlet temperature ratios and hence ε as functions of
NTUandCr.
How real devices differ from ideal models:
- Property variation with temperature (especially cp, μ, k) and non-uniform U across the area cause deviations.
- Flow maldistribution, bypassing, leakage in shell-and-tube exchangers reduce effective area (lower ε than predicted).
- Longitudinal heat conduction in walls (notably in compact exchangers) slightly alters temperature fields.
- Fouling adds thermal resistance over time, reducing U and thus ε for a fixed A. Allowances (fouling factors) are covered under Heat Exchangers fundamentals.
When to use NTU vs LMTD:
- NTU is preferred when at least one outlet temperature is unknown and U and A (or U and target ε) are given/assumed (sizing, rating).
- LMTD is efficient when both outlet temperatures are known or constrained; a correction factor may be needed for mixed configurations. See “Heat exchangers: LMTD method”.
Crossflow “mixed” vs “unmixed” meaning:
- “Unmixed” means fluid passages prevent mixing normal to the main flow direction (e.g., fins), leading to temperature profiles closer to plug flow.
- “Mixed” means near-complete mixing normal to heat transfer surface in that stream (e.g., shell-side with few baffles), approximated as uniform bulk temperature in the cross-stream direction.
Formulas
C = ṁ · cp
- C: capacity rate of a stream (W/K). ṁ: mass flow rate (kg/s). cp: specific heat at constant pressure (J/kg·K). Applies for single-phase fluids with nearly constant cp over the temperature range.
Cmin = min(Ch, Cc), Cmax = max(Ch, Cc), Cr = Cmin / Cmax
- Ch, Cc: hot- and cold-side capacity rates (W/K). Cr is dimensionless. Applies to all configurations.
qmax = Cmin · (Th,in − Tc,in)
- qmax: maximum possible heat transfer rate (W). Th,in, Tc,in: inlet temperatures of hot and cold streams (K or °C; use differences in K). Applies generally.
ε = q / qmax
- ε: effectiveness (dimensionless). q: actual heat transfer rate (W). Applies to all configurations.
NTU = U · A / Cmin
- NTU: number of transfer units (dimensionless). U: overall heat transfer coefficient (W/m²·K). A: heat transfer area (m²). Applies when U is approximately uniform.
q = ε · Cmin · (Th,in − Tc,in)
- q: heat transfer rate (W). Use with configuration-specific ε(NTU, Cr).
Th,out = Th,in − q / Ch, Tc,out = Tc,in + q / Cc
- Outlet temperatures (K or °C). From energy balance assuming no heat loss.
Parallel flow effectiveness:
ε = [1 − exp(−NTU · (1 + Cr))] / (1 + Cr)
- For co-current flow. ε dimensionless; NTU, Cr dimensionless.
Counterflow effectiveness:
ε = [1 − exp(−NTU · (1 − Cr))] / [1 − Cr · exp(−NTU · (1 − Cr))]
- For counter-current flow. ε dimensionless; NTU, Cr dimensionless.
Crossflow, both fluids unmixed:
ε = 1 − exp{ − (1/Cr) · [1 − exp(��Cr · NTU)] }
- For compact finned crossflow where both streams are unmixed. ε dimensionless; NTU, Cr dimensionless.
Crossflow, one fluid mixed and the other unmixed:
ε = (1/Cr) · [1 − exp{ −Cr · (1 − exp(−NTU)) }]
- For crossflow with one stream well mixed transversely. ε dimensionless; NTU, Cr dimensionless.
Special case (Cr → 0, e.g., phase change on one side):
ε = 1 − exp(−NTU)
- Applies to both parallel and counterflow limits when one side has very large capacity rate (or effectively constant temperature).
Invertible forms (use natural log ln):
- Crossflow, both unmixed: from
ε = 1 − exp{ − (1/Cr) [1 − exp(−Cr NTU)] }, definey = exp(−Cr NTU). Theny = 1 + Cr · ln(1 − ε)andNTU = −(1/Cr) · ln[y](requiresy > 0). - Crossflow, one mixed: from
ε = (1/Cr) [1 − exp{ −Cr (1 − exp(−NTU)) }], definez = exp(−NTU). Thenz = 1 + (1/Cr) · ln(1 − Cr ε)andNTU = −ln[z](requires0 < 1 − Cr ε ≤ 1).
Worked examples
Example 1 — Direct rating with given UA (counterflow) Given: Counterflow exchanger. Hot water: ṁh = 1.0 kg/s, cph = 4180 J/kg·K, Th,in = 80 °C. Cold water: ṁc = 2.0 kg/s, cpc = 4180 J/kg·K, Tc,in = 20 °C. Overall UA = 8.0 kW/K. Assume: Steady state, no heat loss, constant properties, U uniform.
- Capacity rates:
Ch = ṁh · cph = 1.0 · 4180 = 4180 W/KCc = ṁc · cpc = 2.0 · 4180 = 8360 W/KCmin = 4180 W/K,Cmax = 8360 W/K,Cr = 4180/8360 = 0.50
- NTU:
NTU = UA / Cmin = (8000 W/K) / (4180 W/K) = 1.913
- Effectiveness for counterflow:
ε = [1 − exp(−NTU (1 − Cr))] / [1 − Cr · exp(−NTU (1 − Cr))]- Compute:
(1 − Cr) = 0.5,NTU (1 − Cr) = 0.9565,exp(−0.9565) ≈ 0.384 - Numerator:
1 − 0.384 = 0.616; Denominator:1 − 0.5 · 0.384 = 0.808 ε ≈ 0.616 / 0.808 = 0.762
- Maximum and actual heat transfer:
qmax = Cmin (Th,in − Tc,in) = 4180 · (80 − 20) = 4180 · 60 = 250800 Wq = ε · qmax = 0.762 · 250800 ≈ 1.912 × 10^5 W = 191 kW
- Outlet temperatures:
Tc,out = Tc,in + q / Cc = 20 °C + (1.912 × 10^5 W)/(8360 W/K) ≈ 42.9 °CTh,out = Th,in − q / Ch = 80 °C − (1.912 × 10^5 W)/(4180 W/K) ≈ 34.3 °C
Answer: q ≈ 191 kW, Tc,out ≈ 42.9 °C, Th,out ≈ 34.3 °C (temperature cross is permissible in counterflow since Cr < 1).
Example 2 — Sizing area A for a crossflow exchanger (both fluids unmixed)
Given: Crossflow (both fluids unmixed) air-to-air exchanger. Hot air: ṁh = 1.2 kg/s, cph = 1007 J/kg·K, Th,in = 120 °C. Cold air: ṁc = 1.5 kg/s, cpc = 1007 J/kg·K, Tc,in = 30 °C. Target cold outlet Tc,out = 70 °C. Overall U = 50 W/m²·K. Find A.
Assume: Steady, adiabatic to surroundings, constant properties, U uniform.
- Capacity rates and Cr:
Ch = 1.2 · 1007 = 1208.4 W/KCc = 1.5 · 1007 = 1510.5 W/KCmin = 1208.4 W/K,Cmax = 1510.5 W/K,Cr = 1208.4 / 1510.5 ≈ 0.80
- Heat duty from cold-side target:
q = Cc (Tc,out − Tc,in) = 1510.5 · (70 − 30) = 1510.5 · 40 = 60420 W
- Maximum possible heat transfer and required effectiveness:
qmax = Cmin (Th,in − Tc,in) = 1208.4 · (120 − 30) = 1208.4 · 90 = 108756 Wεrequired = q / qmax = 60420 / 108756 ≈ 0.556
- Find NTU from ε and Cr (crossflow, both unmixed):
- Use invertible form:
ε = 1 − exp{ − (1/Cr) [1 − exp(−Cr · NTU)] } - Let
y = exp(−Cr · NTU); theny = 1 + Cr · ln(1 − ε) - Compute:
ln(1 − ε) = ln(1 − 0.556) = ln(0.444) ≈ −0.811 y = 1 + 0.80 · (−0.811) ≈ 1 − 0.649 ≈ 0.351NTU = −(1/Cr) · ln(y) = −(1/0.80) · ln(0.351) ≈ 1.31
- Area A from NTU:
A = NTU · Cmin / U = 1.31 · (1208.4 W/K) / (50 W/m²·K)A ≈ 1.31 · 24.168 m² ≈ 31.6 m²
- Hot outlet check (consistency):
Th,out = Th,in − q / Ch = 120 °C − (60420 W)/(1208.4 W/K) ≈ 70 °C(consistent with energy balance; both outlets at ~70 °C is acceptable for these rates).
Answer: Required heat transfer area A ≈ 31.6 m² to achieve Tc,out ≈ 70 °C.
Common mistakes
- Using the wrong ε–NTU formula for the configuration (parallel vs counterflow vs crossflow, and “one fluid mixed” vs “both unmixed”).
- Forgetting that for counterflow temperature cross is allowed; rejecting correct answers where Tc,out > Th,out.
- Using degrees Celsius directly in exponential relations is fine because only temperature differences matter — but keep all ΔT in K for consistency and clarity.
- Using base-10 log instead of natural log when inverting exponential formulas (use ln, not log10).
- Assuming U is the same for gas–gas and liquid–liquid without checking; U depends strongly on convection coefficients and fouling.
- Ignoring fouling and property variation in design margins; NTU method gives idealized performance unless fouling resistances are included in U.
For GATE ME
- Compute effectiveness, heat duty, and outlet temperatures for given ṁ, cp, U, A, and configuration.
- Given target outlet temperature(s) and U, find required area A (i.e., compute NTU from ε and Cr; for crossflow use the invertible relations or iterate).
- Compare parallel vs counterflow performance; identify which can achieve a given target with smaller area.
- Limiting cases: Cr → 0 (phase change), NTU → 0 or large NTU behavior; identify temperature cross conditions.
- Conceptual: when to use NTU vs LMTD; definitions of ε, NTU, Cr; sign conventions and energy balance checks.
- Practise:
- Quick computation of Ch, Cc, Cmin, Cmax, Cr, NTU.
- Using the exact ε formulas for parallel and counterflow; using crossflow correlations and, when needed, inversion with ln.
- Checking answers with both-side energy balances for consistency.
Quick check
- A counterflow exchanger has Cr = 1 and NTU = 2. What is ε?
- In a parallel flow exchanger with Cr = 0.5, what is ε as NTU → ∞?
- A condenser (phase change on hot side) has NTU = 1.5. What is ε approximately?
Answers: 1) ε = NTU/(1 + NTU) = 2/3 ≈ 0.667. 2) ε → 1/(1 + Cr) = 1/1.5 ≈ 0.667. 3) For Cr → 0, ε = 1 − exp(−NTU) ≈ 1 − e^(−1.5) ≈ 0.777.
Finished this topic? Mark it so your progress, study plan and readiness keep up.
Stuck on something here?