Heat Exchanger Sizing: LMTD & U-Values

In chemical process plants, petroleum refineries, power stations, and cryogenic refrigeration facilities, shell-and-tube heat exchangers (STHE) are the workhorses of thermal energy management. Accounting for more than 65% of all heat exchange equipment deployed across heavy process industries, shell-and-tube units operate reliably across extreme pressures (vacuum up to $300+\text{ bar}$), severe temperatures ($-200^\circ\text{C}$ to $1000^\circ\text{C}$), and corrosive, fouling multi-fluid environments.

However, designing and rating a shell-and-tube heat exchanger requires balancing conflicting thermal and hydraulic objectives. Maximizing fluid velocity boosts the convective heat transfer coefficients ($h_i$ and $h_o$), shrinking the required heat transfer surface area ($A$) and reducing capital expenditure (CapEx). Yet, hydraulic pressure drop ($\Delta P$) scales quadratically with velocity ($\Delta P \propto v^2$), escalating pump and compressor operating costs (OpEx). Furthermore, excessive crossflow velocities on the shell side induce severe Flow-Induced Vibration (FIV), leading to tube fretting, fatigue wear against baffle holes, and catastrophic bundle failure within months of commissioning.

This engineering guide presents a rigorous, first-principles foundation for shell-and-tube heat exchanger sizing and rating: thermodynamic energy balances, Logarithmic Mean Temperature Difference (LMTD) derivations, Bowman-Mueller-Nagle $F_t$ correction factors, the resistance-in-series overall heat transfer coefficient ($U_o$), TEMA mechanical standards and fouling factors, the classic Kern empirical method versus the rigorous Bell-Delaware stream analysis method, flow-induced vibration criteria ($\rho v^2$), the $\epsilon$-NTU rating methodology, and a step-by-step industrial kerosene-crude oil preheater design example. You can model and verify your own designs instantly using the free ChemProCal Heat Exchanger Sizer & Rating Tool (HEX-IQ).

1. Heat Exchanger Sizing vs. Rating: Two Distinct Engineering Workflows

Process engineers encounter two fundamentally different thermal design tasks:

2. Thermodynamic Heat Balance & Thermal Duty ($Q$)

The foundation of every heat exchanger calculation is the first law of thermodynamics: steady-state conservation of energy. Assuming negligible heat loss to the ambient environment through insulated shell walls:

When specific heat capacities ($C_p$) remain reasonably constant across the operating temperature range (or evaluated at the caloric/mean bulk fluid temperatures):

Where:

• $Q$ = Total thermal heat duty ($ \text{kW}$ or $ \text{W}$)
• $\dot{m}_h, \dot{m}_c$ = Mass flow rates of the hot and cold streams ($ \text{kg/s}$ or $ \text{kg/h}$)
• $C_{p,h}, C_{p,c}$ = Specific heat capacities of hot and cold fluids ($ \text{kJ/(kg}\cdot \text{K)}$ or $ \text{J/(kg}\cdot \text{K)}$)
• $T_{h,in}, T_{h,out}$ = Hot stream inlet and outlet temperatures ($^\circ \text{C}$ or $ \text{K}$)
• $t_{c,in}, t_{c,out}$ = Cold stream inlet and outlet temperatures ($^\circ \text{C}$ or $ \text{K}$)

Latent Heat Duty (Condensers & Reboilers)

When a fluid undergoes an isothermal phase change (such as pure saturated steam condensing or a pure refrigerant boiling), the heat transfer rate is governed by the latent heat of vaporization ($\lambda$):

For multi-component mixtures boiling or condensing across a temperature glide (such as petroleum fractions or mixed refrigerants), the total duty integrates sensible cooling of the vapor, condensation enthalpy, and subcooling of the condensate:

3. Logarithmic Mean Temperature Difference (LMTD) & Temperature Cross

The local driving force for heat transfer across the exchanger surface is the local temperature difference $\Delta T = T_h - t_c$. Because fluid temperatures change continuously along the tube length, heat transfer rate depends on the integrated mean temperature difference:

Pure Counter-Current Flow

In a pure counter-current arrangement, the hot fluid enters at one end of the exchanger while the cold fluid enters at the opposite end. Integrating the differential heat transfer equation $dQ = U (T_h - t_c) dA$ yields the classical Logarithmic Mean Temperature Difference (LMTD):

Where the terminal approach temperature differences are defined as:

If $\Delta T_1 = \Delta T_2$ (which occurs when the heat capacity rates are perfectly balanced, $C_h = C_c$), the expression evaluates to the indeterminate form $0/0$. Applying L'Hôpital's rule reveals that the logarithmic mean converges exactly to the arithmetic difference: $\Delta T_{lm} = \Delta T_1 = \Delta T_2$.

Pure Co-Current (Parallel) Flow

In parallel flow, both hot and cold streams enter at the same end of the exchanger:

The Temperature Cross Phenomenon

A temperature cross occurs when the outlet temperature of the cold fluid is higher than the outlet temperature of the hot fluid:

While a temperature cross is effortlessly accommodated in pure counter-current flow (such as double-pipe or spiral heat exchangers), it poses severe complications in multi-pass shell-and-tube exchangers.

4. LMTD Correction Factor ($F_t$) & Bowman-Mueller-Nagle Formulations

Industrial shell-and-tube exchangers rarely operate in pure counter-current flow. To maintain high tube-side velocities and compact footprints, exchangers are constructed with multiple tube passes ($N_p = 2, 4, 6, 8$) inside a single shell pass (TEMA E shell). As a result, fluid in half of the tube passes flows co-current with the shell fluid, while the other half flows counter-current.

To account for this departure from ideal counter-current flow, the effective mean temperature difference is calculated using the LMTD Correction Factor ($F_t$):

Where $F_t \le 1.0$. For pure counter-current flow, $F_t = 1.0$. For multi-pass exchangers, $F_t$ is determined analytically from two dimensionless temperature parameters:

1. Temperature Effectiveness ($P$) and Capacity Rate Ratio ($R$)

Here, $P$ ranges between $0$ and $1$, representing thermal effectiveness. $R$ represents the ratio of heat capacity rates between the two streams.

2. Analytical Equations (1-2 Exchanger: 1 Shell Pass, Even Tube Passes)

Derived by Bowman, Mueller, and Nagle (1940), the exact analytical equation for a 1-shell pass, 2-or-more tube pass exchanger is:

When heat capacity rates are equal ($R = 1$):

3. Multiple Shell Passes in Series ($N$ Shells)

When a single shell exhibits an unacceptable $F_t$, splitting the duty across $N$ identical shells in series dramatically increases $F_t$. The overall temperature effectiveness $P$ is converted into the single-shell effectiveness $P_1$:

The single-shell $F_t$ is then evaluated using $P_1$ and $R$. Because $P_1 \ll P$, the resulting $F_t$ is substantially higher.

5. Overall Heat Transfer Coefficient ($U$) & Resistance Network

Heat transfer between two fluids separated by a cylindrical metallic pipe wall encounters five serial thermal resistances:

Where:

• $U_o$ = Overall heat transfer coefficient referenced to outer tube surface ($ \text{W/(m}^2\cdot \text{K)}$)
• $h_i$ = Tube-side convective heat transfer film coefficient ($ \text{W/(m}^2\cdot \text{K)}$)
• $h_o$ = Shell-side convective heat transfer film coefficient ($ \text{W/(m}^2\cdot \text{K)}$)
• $d_o, d_i$ = Outer and inner diameters of the heat exchanger tube ($ \text{m}$)
• $R_{f,i}$ = Inside tube fouling resistance ($( \text{m}^2\cdot \text{K)/W}$)
• $R_{f,o}$ = Outside tube fouling resistance ($( \text{m}^2\cdot \text{K)/W}$)
• $k_{wall}$ = Thermal conductivity of the tube metal material ($ \text{W/(m}\cdot \text{K)}$)

The conductive resistance of the cylindrical tube wall is rigorously expressed by Fourier's law in cylindrical coordinates:

Tube Wall Thermal Conductivities ($k_{wall}$)

TEMA Standard Fouling Resistances ($R_f$)

Fouling layers—scale deposition, particulate sedimentation, wax crystallization, biological growth, and hydrocarbon coking—drastically degrade performance over time. The Tubular Exchanger Manufacturers Association (TEMA) establishes standard design fouling allowances:

Clean vs. Dirty Overall U & Fouling Penalty

Engineers evaluate two distinct overall coefficients:

The fouling penalty percentage quantifies the proportion of thermal resistance caused by dirt layers:

Typical Overall Heat Transfer Coefficient Ranges ($U$)

6. Tube-Side Thermal & Hydraulic Sizing

Fluid mechanics inside tubes follow closed-conduit internal pipe flow principles:

1. Flow Area, Velocity & Reynolds Number

The total number of tubes ($N_t$) is divided equally among the number of tube-side passes ($N_p$):

2. Tube-Side Film Coefficient ($h_i$)

The dimensionless Prandtl number is defined as:

Convective heat transfer correlations depend strictly on the flow regime:

A. Turbulent Flow ($Re_t \ge 10,000$): Calculated using the classical Dittus-Boelter correlation:

Where $n = 0.4$ if the fluid is being heated, and $n = 0.3$ if the fluid is being cooled. For high temperature differences where fluid viscosity varies substantially between the bulk fluid and the tube wall, the Sieder-Tate correlation incorporates a viscosity correction:

B. Transitional Flow ($2,300 < Re_t < 10,000$): Evaluated via the Gnielinski correlation:

C. Laminar Flow ($Re_t \le 2,300$): For fully developed thermal and hydrodynamic flow under uniform wall temperature:

Finally, the tube-side heat transfer coefficient is:

3. Tube-Side Pressure Drop ($\Delta P_t$)

The total tube-side hydraulic loss incorporates two distinct contributions: frictional loss along straight tube lengths and localized dynamic losses in the inlet/outlet headers and $180^\circ$ return bends:

Frictional pressure drop in the straight tubes:

Where the Darcy friction factor ($f_t$) is determined by the Blasius equation for smooth drawn tubes ($f_t = 0.3164 / Re_t^{0.25}$) or the Colebrook-White implicit equation.

Return header and nozzle turning losses (typically 4 velocity heads per tube pass):

7. Shell-Side Mechanical Geometry & TEMA Standards

The shell side houses the tube bundle, segmental baffles, tie rods, sealing strips, and nozzles. Understanding TEMA mechanical designations is essential for practical specification.

TEMA Nomenclature: Three-Letter Code

TEMA exchangers are designated by a three-letter code representing [Front Head Type] - [Shell Type] - [Rear Head Type]:

• Front Stationary Head:
  • Type A: Channel and removable cover (easiest tube access for cleaning without breaking piping connections).
  • Type B: Bonnet (integral cover, lower cost, but piping must be disconnected to expose tube sheet).
  • Type N: Channel integral with tube sheet (minimizes gasketed joints for lethal/toxic service).
• Shell Type:
  • Type E: One-pass shell (most common, lowest cost).
  • Type F: Two-pass shell with longitudinal baffle (pure counter-current capability, but risk of thermal leakage across longitudinal baffle).
  • Type G: Split flow (for horizontal thermosyphon reboilers).
  • Type J: Divided flow (one central inlet, two end outlets, cutting shell pressure drop by up to $80\%$).
  • Type K: Kettle reboiler (enlarged shell vapor disengagement space above tube bundle).
  • Type X: Crossflow shell (no segmental baffles; ultra-low pressure drop for deep vacuum condensers).
• Rear Head:
  • Type L / M / N: Fixed tube sheets (lowest cost, but shell and tubes are welded; cannot pull bundle; requires expansion bellows if thermal differential expansion $> 50^\circ \text{C}$).
  • Type U: U-tube bundle (tubes are bent into U-shapes, free to expand thermally; completely eliminates differential expansion stress; only one tube sheet needed; cannot mechanically clean bends).
  • Type S: Floating head with backing device (removable bundle, thermal expansion accommodated, fully cleanable, but high cost).
  • Type T: Pull-through floating head (entire bundle can be extracted without removing shell cover, but larger shell-to-bundle clearance bypass).

Tube Pitch Layouts & Cleaning Lanes

The tube pitch ($P_t$) is the center-to-center distance between adjacent tubes, typically specified as $1.25 \times d_o$. Four standard layout angles are utilized:

Equivalent Diameter ($D_e$) Formulas

Because flow outside the tubes does not occur in a circular conduit, fluid dynamics correlations use the hydraulic equivalent diameter ($D_e = 4 \times \text{Free Flow Area} / \text{Wetted Perimeter}$):

For Square Pitch ($90^\circ$ and $45^\circ$):

For Triangular Pitch ($30^\circ$ and $60^\circ$):

8. Shell-Side Rating: Kern's Method vs. Bell-Delaware Stream Analysis

The Classical Kern Method (Empirical Shortcut)

Developed by Donald Q. Kern in 1950, this method treats shell-side flow as an idealized, uniform crossflow across the tube bundle.

1. Kern Crossflow Area ($a_s$):

Where $D_s$ is the shell inside diameter, $B$ is the baffle spacing, and $C = P_t - d_o$ is the tube clearance.

2. Kern Shell Mass Velocity ($G_s$) and Reynolds Number ($Re_s$):

3. Kern Heat Transfer Coefficient:

4. Kern Shell Pressure Drop:

Where $N_c = L / B$ is the number of baffle crosses.

The Bell-Delaware Rigorous Stream Analysis Method

Originally developed by Kenneth Bell at the University of Delaware (1963), this method divides shell-side flow into five distinct flow streams identified by Townsend Tinker (1951):

The Bell-Delaware J-Correction Factors

The Bell-Delaware method first computes the heat transfer coefficient for an ideal crossflow tube bank ($h_{ideal}$), then applies five empirical penalty factors ($J$):

1. Baffle Cut Configuration Factor ($J_c$): Accounts for the window zone where fluid flows parallel to the tubes rather than in crossflow. For a standard $25\%$ baffle cut, $J_c \approx 1.0$. For larger cuts ($35\% - 45\%$), $J_c$ drops to $0.70 - 0.85$ due to diminished crossflow velocity.
2. Baffle Leakage Correction Factor ($J_l$): Accounts for the combined bypass of Streams A (tube-hole leakage) and E (shell-baffle leakage). Depending on manufacturing tolerances: $$J_l = 0.95 - 0.04 \, L_{sb} - 0.02 \, L_{tb}$$ Where $L_{sb}$ is shell-baffle clearance ($ \text{mm}$) and $L_{tb}$ is tube-baffle clearance ($ \text{mm}$). Typically, $J_l$ ranges from $0.70$ to $0.85$.
3. Bundle Bypass Factor ($J_b$): Accounts for Stream C bypassing the bundle through the clearance between the shell wall and the Outer Tube Limit (OTL). Sealing strips ($N_{ss}$) physically block this annular lane: $$J_b = \exp\left[-C_{bh} \frac{S_b}{S_m} \left(1 - \sqrt[3]{\frac{2 N_{ss}}{N_c}} \right) \right]$$ Without sealing strips, $J_b$ can collapse to $0.50$ (halving shell performance). Adding 2 to 4 pairs of sealing strips restores $J_b$ to $0.85 - 0.95$.
4. Baffle Spacing Unequal Factor ($J_s$): Accounts for larger inlet and outlet baffle spacing required to accommodate large inlet nozzles ($J_s \approx 0.85 - 1.0$).
5. Laminar Temperature Gradient Factor ($J_r$): Accounts for adverse temperature gradients in laminar flow ($Re_s < 100$). In turbulent flow ($Re_s > 1000$), $J_r = 1.0$.

9. Flow-Induced Vibration (FIV) & Mechanical Integrity

A thermally perfect heat exchanger is utterly useless if it destroys itself mechanically. Flow-induced vibration is the leading cause of premature failure in shell-and-tube heat exchangers. High-velocity shell-side fluids exert periodic alternating hydrodynamic forces against flexible, unsupported tube spans.

The Shell Momentum Flux Criterion ($\rho v^2$)

The simplest and most vital screening metric monitored in the ChemProCal Heat Exchanger Sizer is the crossflow momentum flux ($\rho v^2$):

The Four Fundamental FIV Excitation Mechanisms

1. Vortex Shedding: As fluid flows past a cylindrical tube, alternating von Kármán vortices shed from opposite sides. When the vortex shedding frequency ($f_s = S_t \cdot v / d_o$, where $S_t$ is the Strouhal number) matches the tube's natural mechanical frequency ($f_n$), severe lock-in resonance occurs.
2. Fluidelastic Instability (FEI): Discovered by Connors (1970), FEI is a self-excited aeroelastic runaway condition. Above a critical crossflow velocity ($v_{crit}$), energy absorbed by the tube from fluid motion exceeds the structural damping capacity: $$v_{crit} = \beta \cdot f_n \cdot d_o \sqrt{\frac{m_e \delta_o}{\rho d_o^2}}$$ Where $\beta$ is Connors' threshold coefficient, $m_e$ is effective mass per unit length (tube metal + internal fluid + hydrodynamic added mass), and $\delta_o$ is the logarithmic decrement of structural damping.
3. Acoustic Resonance: Occurs in gas or vapor services when the vortex shedding frequency couples with the acoustic natural frequency of the shell cavity, producing ear-splitting sonic booms ($140+\text{ dBA}$) and fatigue failure of the shell casing.
4. Turbulent Buffeting: Random, broadband turbulent pressure fluctuations buffet the tube bundle, causing long-term low-amplitude fretting wear at baffle supports.

10. The $\epsilon$-NTU Method: Performance Simulation for Existing Exchangers

When sizing a new exchanger where all terminal temperatures are specified, the LMTD method is simple. However, when rating an existing heat exchanger where only the inlet temperatures ($T_{h,in}, t_{c,in}$) and fluid flow rates are known, the outlet temperatures are unknown. Solving for outlet temperatures using LMTD requires tedious, nested trial-and-error iterations because LMTD depends on the very temperatures being calculated.

The $\epsilon$-NTU (Number of Transfer Units) method, developed by Kays and London (1955), solves this problem directly without trial-and-error:

1. Heat Capacity Rates ($C_{min}$ and $C_{max}$)

2. Number of Transfer Units (NTU)

NTU is a dimensionless measure of the physical size and thermal capacity of the heat exchanger:

3. Heat Exchanger Thermal Effectiveness ($\epsilon$)

Effectiveness ($\epsilon$) is the ratio of the actual heat transfer rate to the maximum thermodynamically possible heat transfer rate:

4. Analytical $\epsilon$-NTU Formulas

For Pure Counter-Current Flow:

For TEMA 1-2 Shell-and-Tube Exchanger (1 shell pass, 2 even tube passes):

Once $\epsilon$ is computed from geometry and flow rates, the actual duty is calculated directly as $Q = \epsilon \cdot C_{min}(T_{h,in} - t_{c,in})$, immediately revealing both outlet temperatures without iteration.

11. Shell-Side vs. Tube-Side Fluid Allocation Rules

A critical early design decision is deciding which fluid should travel through the tube side and which through the shell side. Experienced process engineers follow these established industrial heuristics:

12. Comprehensive Worked Industrial Engineering Design Example

To tie every formula and principle together into a concrete engineering artifact, let us work through a complete, rigorous industrial design calculation for a refinery kerosene-to-crude oil preheater.

Step 1: Process Specifications

Step 2: Heat Duty ($Q$) and Cold Fluid Outlet Temperature

1. Calculate hot side thermal duty:

2. Calculate cold crude oil outlet temperature ($t_{c,out}$):

Notice that $t_{c,out} = 121.38^\circ \text{C} > T_{h,out} = 100.0^\circ \text{C}$. A temperature cross of $21.38^\circ \text{C}$ exists!

Step 3: LMTD and Bowman-Mueller-Nagle $F_t$ Correction

1. Terminal approach temperatures for counter-current flow:

2. Dimensionless parameters $P$ and $R$:

Testing for a 1-shell pass ($N=1$): $$P \times R = 0.5086 \times 1.2288 = 0.625 < 1.0$$ However, calculating the single-shell denominator term: $$2 - P\left(R + 1 + \sqrt{R^2 + 1} \right) = 2 - 0.5086\left(1.2288 + 1 + \sqrt{1.2288^2 + 1} \right) = 2 - 0.5086(3.813) = 2 - 1.939 = 0.061$$ This value is dangerously close to zero, yielding an unacceptably low $F_t \approx 0.58 < 0.75$.

Engineering Solution: Select Two Identical Shells in Series ($N = 2$):
Calculating the effectiveness per shell ($P_1$):

Recomputing $F_t$ with $P_1 = 0.3559$ and $R = 1.2288$ yields:

Step 4: Selected Mechanical Geometry (Per Shell)

• Two identical shells in series ($N = 2$). Heat duty per shell: $Q_1 = 1701.39 / 2 = 850.70 \, \text{kW}$.
• Shell Inside Diameter: $D_s = 0.600 \, \text{m} \, (600 \, \text{mm})$.
• Tube Specifications: Carbon steel ($k_{wall} = 50.0 \, \text{W/m}\cdot \text{K}$), $25.4 \, \text{mm} \, (1.0 \, \text{in})$ OD, $2.11 \, \text{mm}$ wall thickness (BWG 14).
  • $d_o = 0.0254 \, \text{m}$, $d_i = 0.0254 - 2(0.00211) = 0.02118 \, \text{m}$.
• Tube Length: $L = 6.0 \, \text{m}$.
• Tube Count: $N_t = 280$ tubes, arranged in $N_p = 2$ tube passes ($140$ tubes per pass).
• Tube Pitch: Square Pitch ($90^\circ$), $P_t = 31.75 \, \text{mm} \, (1.25 \times d_o)$ for mechanical hydroblasting cleanability.
• Baffles: Single segmental, $25\%$ cut, spacing $B = 0.300 \, \text{m}$ ($N_c = L/B = 20$ crosses).
• Clearances: Shell-to-baffle $L_{sb} = 3.0 \, \text{mm}$, tube-to-baffle $L_{tb} = 0.4 \, \text{mm}$, 2 pairs of sealing strips ($N_{ss} = 2$).

Step 5: Tube-Side Hydraulic & Thermal Rating

1. Flow area per pass:

2. Tube fluid velocity and Reynolds number:

3. Tube-side pressure drop ($N_p = 4, v_t = 0.481 \, \text{m/s}$):

Step 6: Shell-Side Rating (Bell-Delaware Stream Analysis)

1. Square pitch equivalent diameter ($d_o = 0.0254 \, \text{m}, P_t = 0.03175 \, \text{m}$):

2. Bell-Delaware crossflow area ($S_m$):

3. Shell mass velocity and crossflow velocity:

4. FIV Momentum Flux Verification ($\rho v^2$):

5. Shell Reynolds and Prandtl numbers:

6. Ideal crossflow film coefficient ($h_{ideal}$):

7. Bell-Delaware J-Correction Factors:

• $J_c$ (Baffle Cut $25\%$): $J_c = 1.00$
• $J_l$ (Leakage: $L_{sb}=3.0 \text{ mm}, L_{tb}=0.4 \text{ mm}$): $J_l = 0.95 - (0.04 \times 3.0) - (0.02 \times 0.4) = 0.822$
• $J_b$ (Bypass with 2 pairs sealing strips): $J_b = 0.900$

8. Shell pressure drop via Kern crossflow:

Step 7: Overall $U$ and Thermal Area Margin

1. Thermal resistance components referenced to outside tube surface ($d_o / d_i = 0.0254 / 0.02118 = 1.199$):

2. Clean and Dirty Overall Heat Transfer Coefficients:

3. Surface Area Required vs. Area Available (Per Shell):

13. Engineering Rule Engine & Automated Diagnostics

When running simulations in the ChemProCal Heat Exchanger Intelligence Engine, your inputs are continuously checked against 12 automated engineering rules to flag hidden design flaws:

14. Frequently Asked Questions (FAQ)