It’s the first question to hit you with every shell-and-tube heat exchanger. It comes up when you are designing a heat exchanger, or just trying to draw one into a process sketch, process flow diagram, or piping and instrumentation diagram. Fluid allocation: which fluid goes into the shell side and which into the tube? Alas, there is no straightforward answer.
Which fluid goes on which side?
Some considerations, or rules-of-thumb, follow. I have gathered these from various books, articles, presentations, and other sources. Please review this list but you must not follow it as gospel. Every situation needs to be evaluated on its own merits and some of these rules will conflict with the others. I offer a tentative "ranking" for when the advice conflicts at the end of the post.
Heat Exchanger Fluid allocation advice:
- The tubeside has less metal than the shellside. Therefore, capital cost pressures favor putting corrosive fluids in the tubeside. Why? Corrosive services tend to call for expensive, exotic metals that can withstand corrosion better than plain old carbon steel. Better to minimize the use of the expensive material by sending corrosive services to the tubeside
- Similarly, extreme pressures and temperatures (high or low) can increase the metal thickness or cost of the materials of construction required. One would rather place extreme P/T services on the tube-side. (It is easier to make the tubes resist high pressures rather than the entire, larger, shell)
- It is easier to keep velocities higher in the tubeside. This is good for services that demand a consistently high velocity, like cooling water
- For maintenance, it is easier to clean the tubeside than the shellside. Sometimes you can just open the “head” of the exchanger and hydroblast each tube, instead of having to remove the entire tube bundle to get at the shell. Also the inside of tubes is an easier surface to deal with then the complex surface of the tube bundle outsides and shell insides. Often only chemical cleaning can be reasonably performed on the shellside. Therefore, the maintenance department would prefer that any fouling/viscous/solid-carrying/dirty streams go into the tubeside
- The shellside tends to be preferred for services with phase changes
- Finned tubes can be used to increase the effective surface area the shell-side fluid sees, letting you increase the heat transfer effectiveness on this side
- Film resistance / laminar flow is more easily overcome on the shellside. The shellside tends to experience vortex shedding, and rapid changes in direction due to tube support baffles. These factors promote mixing between “layers” of fluid. This means that often, from a heat transfer perspective, you would prefer to put viscous fluids in the shell-side
- There are more variables you can “play” with in the shellside, allowing you more options to deal with high pressure drops or low heat transfer co-efficients and to more precisely target certain values. For example, tube baffle support spacing can be changed in small increments. This flexibility tends to favor putting the fluid with poor heat transfer properties on the shellside. With the tubeside usually all you can do is alter the number of tube passes to an even number (2, 4, 6, etc.). Going from 2 to 4 passes will roughly double the velocity and increase the pressure drop by a factor of eight. (ΔP is proportional to Length and to Velocity squared, and doubling the tube passes doubles both L and V values)
- Twisted tubes or tube inserts can be used to overcome laminar flow in the tubeside and provide an extra “variable” for use. However, these tend to be rare/proprietary technologies and more research is required to find one that will work reliably for your service
- “Critical” fluids (i.e. hazardous fluids or high cost fluids) such as corrosive, lethal or expensive fluids should be positively contained to prevent leaks. This means that certain heat exchanger designs or TEMA types, which feature gaskets or floating heads, are not suitable for shellside service. (Conversely, all-welded exchangers could be suitable putting the critical fluid into for shellside service, assuming you remember that the shellside is more difficult to clean)
- For reboilers, Table 1 in the article Reboiler Circuits for Trayed Columns1 tells you which side typically holds the boiling fluid in various reboiler designs
To try to put this in a ranking, normally (not always), put the highest item on this list on the tubeside:
- Cooling water2
- The more fouling, erosive or corrosive fluid
- Less viscous fluid
- The fluid under higher pressure (this could override all other conditions if the fluid is at very high pressure)
- The hotter fluid
- The smaller volumetric flow rate
Typical practices for refinery service include:
- Condensing vapours are normally on shell-side
- Condensing steam is normally on tube-side
- Thermal expansion problems may govern selection of fluid for tube-side and shell-side. (i.e. if fluid temperature change is > 150°C (300°F) then normally you would put it in the shells-side which can better handle large temperature change in certain exchanger designs)
- Very small volumetric flow rates or high viscosity streams could be shell-side services if laminar tube-side flow can be avoided (e.g. first exchanger in a crude preheat train).
In the end, sometimes there is no clear answer and it is worthwhile trying the design both ways. If you have problems with high pressure drop or low transfer co-efficients on either side of the exchanger, sometimes switching the tubeside and shellside fluids is the solution you need. Also, vibration concerns may recommend you switch the fluid allocation.
- The reboiler article mentioned above is: “Reboiler Circuits For Trayed Columns” by Design Practices Committee Fractionation Research Inc. Chemical Engineering January 2011 [↩]
- Cooling water is usually the more fouling service. It is more likely to have solids, or algae, or a risk of minerals plating out if it gets too hot. Therefore the cooling water is usually put on the tubeside, and the velocity kept high, to reduce fouling. About 5 – 7 ft/s or 1.5 – 2.1 m/s. Temperature control will normally put bypasses on the process side rather than the cooling water side, to keep velocity high, flow variations at the local unit rather than the utility, and to avoid trapping cooling water behind a valve where it could boil [↩]