Introduction
The shell-and-tube heat exchanger is the workhorse of process heat transfer — rugged, cleanable, and buildable in any pressure and material combination a facility needs. It is also routinely undersized, not because the arithmetic is hard, but because the two numbers that dominate reliability — the overall heat transfer coefficient (U) and the fouling allowance — are the two most often guessed.
This post walks the full sizing chain: energy balance, log-mean temperature difference, the U calculation, fouling resistance, and the TEMA configuration decisions that turn a required area into a real piece of equipment. Heat exchanger duty rarely stands alone — the inlet conditions come straight out of the equation of state chosen for the process, and a cooler that underperforms lands on the next machine downstream, which is often a compressor tripping on high suction temperature.
The Duty and Energy Balance
Everything starts with the heat duty, Q — how much energy must move from the hot stream to the cold stream per unit time.
Q = m × cp × ΔT (sensible heat)
Q = m × λ (latent heat, phase change)
Where m is mass flow, cp is specific heat capacity, ΔT is the temperature change of the stream, and λ is the latent heat of vaporisation or condensation.
The non-negotiable rule: the duty must balance on both sides. The heat given up by the hot stream equals the heat absorbed by the cold stream (plus a small ambient loss, usually neglected for insulated exchangers).
Q_hot = m_hot × cp_hot × (T_hot,in − T_hot,out)
Q_cold = m_cold × cp_cold × (T_cold,out − T_cold,in)
Q_hot = Q_cold = Q
If you know three of the four terminal temperatures and both flow rates, the fourth temperature is fixed by this balance — it is not a free design choice. A common early error is specifying all four temperatures independently and then wondering why the exchanger will not converge in simulation.
LMTD and the Correction Factor
Heat transfer is driven by the temperature difference between the two streams, but that difference changes along the length of the exchanger. The correct average is the log-mean temperature difference (LMTD):
LMTD = (ΔT1 − ΔT2) / ln(ΔT1 / ΔT2)
Where ΔT1 and ΔT2 are the temperature differences at each end of the exchanger. For counter-current flow (the two streams travelling in opposite directions), this LMTD applies directly and is always larger than the co-current equivalent — which is why counter-current is the default arrangement.
Real shell-and-tube exchangers are rarely pure counter-current. A single shell pass with multiple tube passes has some tubes flowing with and some against the shell stream, which degrades the effective driving force. This is corrected with a factor F:
ΔT_effective = F × LMTD
F is read from TEMA/Kern charts as a function of two dimensionless ratios (R and P) built from the four terminal temperatures.
The F-factor rule of thumb: never design below F = 0.80. Below that, the exchanger is on the steep part of the curve where a small temperature shift collapses the driving force — and it signals a temperature cross (the cold outlet exceeding the hot outlet) that a single shell pass cannot handle. The fix is multiple shells in series, or a genuine counter-current design.
The Sizing Equation
With duty and effective driving force in hand, the required area falls out of the master equation:
Q = U × A × (F × LMTD)
therefore A = Q / (U × F × LMTD)
Where A is the heat transfer area (m²) and U is the overall heat transfer coefficient (W/m²·K). Three of these four are now known; A is the answer. The entire difficulty of the calculation collapses onto one number: U.
Overall Heat Transfer Coefficient and Fouling
U is a series of thermal resistances added in reverse — the film on each side, the tube wall, and critically the fouling layers that build up over the operating run:
1/U = 1/h_i + R_fi + (t_wall/k_wall) + R_fo + 1/h_o
Where h_i and h_o are the inside and outside film coefficients, R_fi and R_fo are the fouling resistances, and t_wall/k_wall is the tube wall conduction term (usually small for metals).
Typical clean U values give a sanity check before any detailed rating:
| Service | U (W/m²·K) |
|---|---|
| Water / water | 800–1,500 |
| Light hydrocarbon / water cooler | 400–800 |
| Gas / gas (both sides low pressure) | 50–200 |
| Condensing steam / water | 1,500–4,000 |
| Viscous oil / water | 100–350 |
The reliability killer is fouling. The fouling resistances are added at the design stage as an allowance for the dirt, scale, wax, and corrosion products that accumulate between cleanings:
| Fluid | Fouling resistance R_f (m²·K/W) |
|---|---|
| Treated cooling water | 0.0002–0.0004 |
| Seawater | 0.0004–0.0009 |
| Crude oil | 0.0004–0.0009 |
| Produced water | 0.0005–0.001 |
| Amine / glycol solutions | 0.0002–0.0006 |
Fouling can easily account for 20–40% of the total resistance. An exchanger rated clean, with no fouling allowance, will meet duty on the commissioning day and slip off-spec within months. The fouling allowance is not conservatism — it is the difference between the day-one and end-of-run performance the equipment must span.
TEMA Types and Configuration
TEMA (Tubular Exchanger Manufacturers Association) classifies exchangers by a three-letter front-head / shell / rear-head code. The practical decisions that follow from the sizing:
- Fixed tubesheet (L, M, N rear head) — cheapest, but the shell side cannot be mechanically cleaned and thermal expansion must be absorbed by a bellows. Use for clean shell-side services.
- U-tube — one tubesheet, free thermal expansion, but the U-bends cannot be mechanically cleaned inside. Good for clean tube-side, high thermal differential.
- Floating head (S, T, P, W rear head) — the tube bundle is removable for full mechanical cleaning on both sides. The default for fouling or dirty services despite the higher cost.
- Tube side vs shell side allocation — put the dirtier, higher-pressure, more corrosive, or lower-flow fluid in the tubes (easier to clean, cheaper to build in exotic metals). Put the fluid needing a low pressure drop on the shell.
Two more constraints size the physical bundle: velocity (keep tube-side liquid roughly 1–2.5 m/s to limit fouling without eroding — cross-check against API 14E erosional velocity) and pressure drop (typically an allowable 0.3–0.7 bar per side; too little velocity fouls, too much wastes pump/compressor head).
A Worked Example
Scenario: Cool a light hydrocarbon condensate stream from 95°C to 45°C using treated cooling water rising from 30°C to 40°C.
Duty (condensate side): m = 25 kg/s, cp = 2.4 kJ/kg·K.
Q = 25 × 2,400 × (95 − 45) = 3,000,000 W = 3.0 MW
Cooling water flow (check the balance): with cp_water = 4.18 kJ/kg·K and a 10°C rise,
m_water = 3,000,000 / (4,180 × 10) = 71.8 kg/s
LMTD (counter-current):
ΔT1 = 95 − 40 = 55°C (hot in vs cold out)
ΔT2 = 45 − 30 = 15°C (hot out vs cold in)
LMTD = (55 − 15) / ln(55/15) = 40 / 1.299 = 30.8°C
Correction factor: for a 1-shell / 2-tube-pass arrangement with these temperatures, F ≈ 0.88 (above the 0.80 floor — a single shell is acceptable).
Overall U: light hydrocarbon / water cooler, take a dirty (fouled) design value of U = 550 W/m²·K.
Area:
A = Q / (U × F × LMTD)
= 3,000,000 / (550 × 0.88 × 30.8)
= 3,000,000 / 14,907
= 201 m²
Result: specify roughly 200 m² of fouled-condition area, 1 shell pass / 2 tube passes, floating-head (condensate can foul). Had the same exchanger been rated at a clean U of 800 W/m²·K, the calculated area would have been only 138 m² — a machine 30% too small the moment the tubes start to dirty. The fouling allowance is the difference between a cooler that holds 45°C for five years and one that drifts to 55°C by the first turnaround.
Common Pitfalls
- Sizing on clean U. The single most common cause of chronic underperformance. Always design to the fouled coefficient; verify the clean coefficient only to confirm start-of-run margin, not to set the area.
- Over-specifying the terminal temperatures. Four independent temperatures plus two flows over-constrains the energy balance. Fix three and let the fourth follow.
- Ignoring the F-factor. A design that pencils out at F = 0.72 has a temperature cross and needs multiple shells — a single shell will simply never reach the target outlet.
- Wrong fluid allocation. Seawater on the shell side of a fixed-tubesheet exchanger is a maintenance trap — it fouls the one surface you cannot mechanically clean.
- Neglecting pressure drop. Chasing a high U with high velocity can blow the allowable ΔP and starve the downstream pump or compressor. U and ΔP are traded, not maximised independently.
- No margin for end-of-run. The exchanger must meet duty on the dirtiest day before cleaning, not the cleanest. Rate it there.
Conclusion
Sizing a shell-and-tube exchanger is a four-link chain — duty, LMTD × F, U, area — and the chain is only as strong as the U value and the fouling allowance behind it. The arithmetic is unforgiving in a quiet way: an exchanger sized on clean coefficients passes every acceptance test and then fails slowly, one fouled tube at a time, until a product runs off-spec or a machine downstream trips on temperature.
Get the fouling allowance right and the exchanger holds its duty across the full run between cleanings. Guess it, and you have bought a machine that works perfectly on the one day nobody is watching and disappoints on every day after.
