Author: Senthil Kumar, Technical Director | Updated: June 2026 | Reading time: 12 min
Jump to a Question
- Why is steam fed to the shell side, not the tube side?
- Not cooling enough — more process fluid or more cooling fluid?
- Standard practice for plumbing a shell and tube heat exchanger
- Can the U value exceed the design value? Is that expected?
- Can the U value stay constant even when tubes are fouled?
- Can fouled tubes reduce flow through the exchanger?
- Prandtl number vs Nusselt number — what is the real difference?
- Heat loss from uninsulated exchangers — should Q hot = Q cold?
- What happens without an impingement plate — and how to prove it?
Why is steam fed to the shell side in heat exchangers rather than the tube side?
This seems to have an obvious answer — until you dig into it and realize it depends on the application. Steam isn't always on the shell side. But when it is, there are good reasons for it.
Condensate Drainage
Steam condenses as it gives up heat. On the shell side, condensate pools at the bottom and drains through a dedicated nozzle with a steam trap. Inside tubes, condensate causes water hammer, slug flow, and damage at tube bends.
Uniform Heat Distribution
Shell-side steam bathes the full tube bundle — every tube gets approximately the same steam temperature. Multi-pass tube-side steam arrangements risk condensate accumulation causing temperature non-uniformity across passes.
Condensing Film Coefficient Is Already High
Steam condensing coefficients are 6,000–15,000 W/m²K. Whether shell or tube side, steam is almost never the controlling resistance. The tube-side process fluid usually limits the overall U — so steam placement has little thermal impact.
Pressure Rating Economics
At moderate steam pressures (below ~20 bar), the shell handles it cost-effectively. At very high steam pressures, this can flip — thick-walled tubes are cheaper than a heavy-wall shell, which is why some high-pressure steam heaters put steam tube side.
The real rule isn't "steam goes shell side" — it's "put the fluid where it fits the design best." Steam goes tube side in U-tube steam heaters, kettle reboilers (where steam is in the bundle), and some fired-tube applications. Condensate management is the dominant factor. If you can drain condensate reliably from either side, steam can go either way.
If a shell and tube heat exchanger isn't cooling enough — would you add more process fluid or more cooling fluid? Other troubleshooting ideas?
Short answer: increase the cooling fluid, not the process fluid. Adding more process fluid increases your heat load — the opposite of what you need. But the real answer is more systematic than that.
The governing equation tells you exactly where to look:
Q must equal: ṁhot × Cphot × (Tin − Tout) = ṁcold × Cpcold × (tout − tin)
If the exchanger isn't hitting the target outlet temperature, one or more of U, A, or LMTD has dropped below design. Work through these in order:
| What to Check | How to Check It | What It Tells You |
|---|---|---|
| Fouling — tube side or shell side | Compare actual ΔP to design ΔP on both sides | Higher ΔP = fouling building. Clearest diagnostic signal — fouling reduces U and raises ΔP simultaneously |
| Cooling water inlet temperature | Measure actual vs design CW supply temperature | If CW is warmer than design (summer conditions), LMTD drops and you lose duty — a very common seasonal cause |
| Cooling water flow rate | Check flow meter vs design value | Reduced CW flow lowers shell-side h and reduces LMTD — the most common operational cause of underperformance |
| Shell-side bypassing | Temperature survey at multiple points across the shell | Damaged baffles let fluid bypass the bundle without exchanging heat — classic cause of sudden performance drop |
| Air or vapour pockets | Open the shell-side vent. Any gas? | Gas pockets insulate sections of the bundle. Venting alone can recover significant duty instantly |
| Process flow rate increase | Check actual flow vs design | Higher process flow raises heat load but U×A×LMTD hasn't changed — the exchanger is now undersized for the duty |
| Tube plugging | Check maintenance records; count plugged tubes | Plugged tubes reduce effective area A. More than 5–10% plugged is a significant concern |
Fastest field check: Measure the cooling water outlet temperature. If it is well below design, you have more than enough cooling capacity — the problem is on the process side (fouling, bypassing, or increased load). If CW is leaving close to the process temperature, your cooling water supply is the constraint.
What is the standard practice for how a shell and tube heat exchanger is plumbed?
A handful of rules that most process engineers follow by habit — not because they are regulatory requirements, but because decades of experience have shown they work.
Fluid Allocation — Which Side Gets Which Fluid
Rules governing which fluid goes tube side vs shell side- Fouling / dirty fluid → tube side — tubes are easier to clean mechanically
- Corrosive fluid → tube side — only need corrosion-resistant tubes, not the whole shell
- High-pressure fluid → tube side — cheaper to make thick-wall tubes than a thick-wall shell
- Viscous fluid → shell side — lower pressure drop; baffle-directed cross-flow breaks up viscous boundary layers better
- Lower flow rate fluid → tube side — multiple passes make tube velocity acceptable
Flow Arrangement and Piping Connections
Rules governing flow direction and piping layout- Counter-current flow is standard — highest LMTD, most efficient. Hot inlet faces cold outlet end.
- Shell nozzles on opposite sides — shell inlet at one end, outlet at the other
- Piping must be self-draining — no dead legs at low points
- Install vent and drain nozzles on both shell and tube sides — mandatory for safe maintenance
- Block valves and bypass on both sides for isolation and service access
- Allow for thermal expansion — loops or bellows in connected piping; do not anchor both ends rigidly
Most commonly ignored rule in the field: Cooling water should enter the bottom tube-side nozzle and exit the top — keeping the tubes fully flooded and preventing air pockets. A surprising number of exchangers are plumbed the wrong way during construction and nobody notices until performance is mysteriously poor.
Does it make sense for a heat exchanger to have a U value greater than its design value? Is this expected in some practical cases?
Yes — and it is not just possible, it is expected for a new or recently cleaned exchanger. Once you understand how the design U is calculated, this makes complete sense.
The design U already has fouling factors built in. When a thermal engineer sizes an exchanger in HTRI, they calculate a clean U from the film coefficients, then subtract fouling resistances per TEMA to get the design U — the value the unit must still meet at end-of-run fouled condition:
So a brand-new, perfectly clean exchanger will have U ≈ Uclean, which is higher than Udesign. This is intentional — the extra surface area compensates for the performance loss expected as fouling builds over time. An overdesign factor of 20–30% is typical in refinery service.
New Exchanger, First Run
No fouling has accumulated yet. U ≈ Uclean, which is above Udesign by the full overdesign margin.
After Cleaning
Tubes restored to near-clean condition. U recovers toward Uclean — performance temporarily exceeds design.
Higher Flow Than Design
Higher velocity increases film coefficients on one or both sides. U exceeds design U if the increase is sufficient.
Lower Viscosity Than Specified
Higher Re, higher Nu, higher h — all resulting in a measured U above the design value for the specified (more viscous) fluid.
If your measured U stays significantly above design U for a long period: either the TEMA fouling factors used were overly conservative for your specific service, or your process conditions are consistently milder than the design case. Neither is a problem — it means you have more capacity margin than you thought.
If the tubes become fouled, is it still possible for the U value to remain constant — could an increase in the tube-side convective coefficient cancel out the fouling resistance?
Theoretically yes, practically almost never significant enough to cancel the fouling effect. Here is the mechanism you are thinking of:
Fouling deposits reduce the tube inside diameter. If volumetric flow stays constant, velocity increases as cross-sectional area decreases. Higher velocity → higher Re → higher Nu → higher htube. So the convective coefficient does go up slightly as the tube fouls.
But here is why it does not cancel the fouling resistance:
Fouling Deposits Are Poor Conductors
Typical fouling deposits have thermal conductivity of 0.1–1.5 W/m·K versus ~50 W/m·K for steel. A 0.5 mm deposit has the thermal resistance of 25 mm of steel wall — a massive penalty.
The h Increase Is a Modest Effect
h ∝ V0.8 in turbulent flow. Reducing tube ID by 10% increases velocity by ~23% and h by only ~19%. The fouling resistance penalty is far larger than this recovery.
Flow May Actually Redistribute
Fouling increases tube-side ΔP, which in a parallel system can reduce flow to the fouled exchanger — lowering velocity and removing even the partial benefit you were counting on.
One real-world case where this effect is more pronounced: When a smooth, hard scale (like calcium carbonate) deposits at high tube velocities, the net performance loss is sometimes less than the fouling resistance alone would predict. Hard, smooth deposits are less insulating than soft, porous biological films. This is a partial offset — not a full cancellation. In most fouling scenarios, U drops and keeps dropping.
Can fouled tubes decrease the flow through the exchanger — and does this further reduce heat transfer?
Yes — and you have described the mechanism exactly right. This is one of the most insidious failure modes in cooling water systems because it is a self-reinforcing loop.
Fouling Reduces ID
Deposits build up inside tubes, reducing the internal diameter and increasing tube-side pressure drop significantly (ΔP ∝ 1/D⁵).
Flow Redistributes
In a header system with parallel exchangers, higher ΔP causes flow to shift to lower-resistance paths — the fouled exchanger gets starved of flow.
Velocity Drops
Lower flow in the fouled exchanger means lower velocity → Re drops → may fall from turbulent toward transitional → Nu and htube drop sharply.
U Drops Further
Lower h compounds the fouling resistance already present — double penalty on U. Heat transfer collapses.
More Fouling Deposits
Higher tube wall temperature and lower velocity create ideal conditions for further fouling — the loop completes and repeats.
This is why cooling water systems need per-exchanger flow monitoring, not just header pressure. An exchanger can be starving for flow while header pressure looks fine — because the flow has simply gone elsewhere. Minimum tube-side velocity for most cooling water service is 0.9–1.2 m/s. Below this, fouling accelerates and biological growth becomes a serious risk.
After cleaning, always verify flow recovery. If flow does not return to design value after cleaning the tubes, check for a blocked inlet strainer, a partially closed isolation valve, or damage at the tube sheet face that is restricting tube inlets. Cleaning the tubes alone does not fix a flow distribution problem.
I am having a hard time understanding the Prandtl number. Some sources describe it as the ratio of convective to conductive heat transfer — but isn't that the Nusselt number?
You are right to be confused — the way Prandtl is often described is genuinely misleading. Let us clear this up properly, because mixing these two up causes real errors in heat transfer analysis.
Nusselt Number vs Prandtl Number — What Each One Actually Means
| Fluid | Pr (approx.) | What It Means for Heat Transfer |
|---|---|---|
| Liquid metals (mercury, sodium) | 0.003–0.03 | Heat diffuses much faster than momentum — exceptional heat transfer fluid, used in nuclear reactors |
| Air and gases | 0.7–1.0 | Velocity and thermal boundary layers are roughly equal thickness — well-balanced |
| Water at 20°C | ~7 | Momentum diffuses faster than heat — velocity BL thicker than thermal BL |
| Light oils | 50–300 | Viscous effects heavily dominate thermal diffusion — noticeably harder to heat or cool |
| Heavy oils | 1,000–100,000 | Extremely poor thermal diffusivity — heat barely penetrates, very difficult to heat exchange |
To answer directly: A high Prandtl number does not mean the fluid transfers heat more easily — it means the opposite. High Pr fluids (viscous oils) have poor thermal diffusivity: heat is reluctant to spread through them. This is why the Dittus-Boelter correlation includes Pr0.4 — to capture how much harder high-Pr fluids are to heat. A liquid metal with Pr ≈ 0.01 is an extraordinary heat transfer fluid precisely because its thermal diffusivity is so high relative to its viscosity.
For shell and tube heat exchangers that are not insulated, what percent of heat loss can we expect from radiation? Should we not expect Q (hot side) = Q (cold side)?
Strictly speaking, Qhot ≠ Qcold for any real exchanger — there is always some heat lost to the surroundings. The question is whether the loss is large enough to matter for your calculation.
Natural Convection (Dominant)
For shell surface temperatures below ~300°C, natural convection from the shell OD to surrounding air is almost always the dominant mechanism of external heat loss — not radiation.
Radiation (Secondary)
For a shell at 80–120°C, radiation contributes roughly 20–35% of total external heat loss, with convection making up the rest. Radiation grows in importance above ~200°C where the T⁴ term dominates.
Total External Loss (Typical Range)
For most industrial exchangers, total external heat loss is 1–5% of process duty if uninsulated. For large, high-temperature, low-duty units, losses can reach 8–15% of duty.
When It Actually Matters
Increased shell temperature (surpassing 150°C), extensive shell diameter, reduced process duty, or detailed energy balance investigations (heat integration reviews, efficiency analyses). Always account for it in these cases.
Qloss typically = 1–5% of duty for uninsulated exchangers
Qloss can reach 8–15% for high-temperature, low-duty, large-shell units
Practical rule: Assume Qhot = Qcold for insulated or moderate-temperature exchangers. Apply a 2–5% loss factor to hot-side duty for uninsulated high-temperature units in energy balance calculations. If doing a field performance test, always measure both sides independently — the difference gives you the actual heat loss, which is useful data for assessing shell condition and insulation effectiveness.
What can happen to a shell and tube heat exchanger without an impingement plate — and how do I prove it is the root cause of corrosion opposite the shell-side inlet nozzle?
This is a textbook impingement damage case. The location you described — top tube rows directly opposite the shell-side inlet nozzle, operating with 26% MDEA since 1997 — matches the classical impingement-erosion pattern exactly. Let us go through the mechanism and then build the technical case.
What an impingement plate does — and what happens without one
When the shell-side fluid enters through the inlet nozzle, it arrives as a high-velocity jet. Lacking a plate along its course, this jet collides directly with the tube bundle. The result is erosion-corrosion — the mechanical energy of the impinging stream strips the protective oxide layer from the tube surface faster than it can reform, continuously exposing fresh metal to the corrosive fluid.
In your case, 26% MDEA with dissolved CO₂ and H₂S is corrosive to carbon steel at operating temperature. Combine that chemistry with high-velocity direct impingement and the damage accelerates dramatically — which explains why the top rows (directly in the jet path) are affected while deeper rows are not.
What TEMA says about this
ρV² > 1,490 kg/(m·s²) for non-corrosive, non-abrasive fluids → impingement plate REQUIRED
For corrosive or abrasive fluids → impingement plate REQUIRED regardless of velocity
26% MDEA solution qualifies as a corrosive fluid. Under TEMA, an impingement plate has been mandatory for this exchanger regardless of nozzle velocity — from the original design in 1997 onwards.
How to build a watertight root cause case
Map the Damage Location
Impingement damage follows the jet cone from the nozzle — maximum thinning directly in line with the nozzle centreline, tapering off at the edges. Sketch the nozzle position and overlay the corrosion map. They must match.
UT Thickness Survey
Use ultrasonic thickness gauging on a grid pattern at 50–100 mm centres across the affected bundle. Plot the thickness map. Impingement produces a clear bulls-eye of thinning centred on the nozzle axis — impossible to explain by corrosion chemistry alone.
Calculate Nozzle ρV²
Calculate inlet nozzle ρV² from actual operating flow rate, MDEA density, and nozzle inside diameter. If it exceeds 1,490 kg/(m·s²), you have a clear TEMA velocity violation. Even if it does not, the corrosive fluid classification makes impingement protection mandatory.
Inspect the Undamaged Zones
Examine the tubes and bottom rows far from the nozzle face. If those tubes show significantly less corrosion despite being in the same MDEA stream, the fluid chemistry alone cannot explain the top-row damage. The location-specific pattern points conclusively to impingement.
Examine Tube Surface Appearance
Impingement-erosion produces smooth, scalloped, or pitting on the tube face pointing toward the nozzle. Pure corrosion without impingement is more uniform around the tube circumference. The asymmetric damage pattern is strong physical evidence.
Review the Damage Timeline
If operating conditions have been stable since 1997 but corrosion has progressed steadily, it is consistent with cumulative erosion-corrosion — not a sudden change in chemistry. Time under impingement is the driving variable.
Immediate action recommended: Have the top tube rows UT-measured before the next planned shutdown. If wall thickness is below the minimum calculated thickness per ASME UG-27 minus corrosion allowance, plug those tubes immediately. At the next shutdown, install an impingement plate or distributor baffle at the inlet nozzle. Given 26+ years of operation without protection, also inspect the first baffle and baffle space for erosion damage.
For your root cause analysis report: Reference TEMA 9th/10th Edition, RGP-T-2.4 (impingement protection) and RGP-T-3.5 (corrosive service requirements). Cite the nozzle ρV² calculation alongside the MDEA corrosive classification — both independently require an impingement plate. Attach the UT thickness map with the bulls-eye damage pattern overlaid on the nozzle centreline. Physical evidence + TEMA violation + corrosive service classification = a case that is very difficult to argue against in a root cause investigation or insurance claim.
Have a Technical Question About Your Heat Exchanger?
Our engineering team reviews performance issues, troubleshooting data, and design questions for shell and tube heat exchangers in active service. Share your process conditions and we will respond with a technical assessment within 48 hours.
Talk to Our Engineering Team →Author: Senthil Kumar, Technical Director — United Heat Exchangers Pvt. Ltd. | Last Updated: June 2026