Mar-2016

Improving heat transfer in reboilers and condensers

Guidelines and experience on how to deal with a loss of condensing or reboiling capacity

NORMAN LIEBERMAN
Process Improvement Engineering

Article Summary

Condensing steam against the surface of clean, cold tubes of a water cooled condenser ought to produce an excellent heat transfer coefficient (U value). The author has observed coefficients as high as 300-350 Btu/h/ft2/°F. But, quite often, the U value is only 60-80, or even less.

Wherein lies the problem? And, what can be done about it? Certainly, fouling on the cooling water side due to low tube side velocities, poor water treatment, and tubes plugged with crawfish are quite common. Back-flushing, acid cleaning, and chlorine shocking the water side are the usual steps to correct tube side water fouling. However, the subject of this article is the loss of shell side heat transfer.

Vapour binding
Ordinarily, the steam sides of reboilers and surface condensers are clean. The usual steam side fouling factor of 0.0005 to 0.001 is normally sufficient to allow for the steam side fouling with silicates and carbonates. However, there is another type of temporary fouling that greatly inhibits the rate of steam condensation. This is called vapour binding and is a consequence of non-condensables in the steam. There are four origins of non-condensables in steam:
• Tube leaks in reboilers, such as those that serve deethanisers, depropanisers and debutanisers
• Carbon dioxide due to the thermal decomposition of the carbonates that are not effectively removed in the preparation of boiler feed water (BFW)
• Nitrogen inadvertently introduced in the plant steam supply (from an inert gas purge)
• Air leaks into steam turbines that exhaust to a vacuum due to defective turbine shaft seals, and/or surface condenser external leaks.

Figure 1 illustrates this common problem of vapour binding. The steam contacts the cold metal surface of the tubes, and drips off. The air, CO2, or non-condensable light hydrocarbon molecules do not condense. As vapour velocities in a steam condenser are inherently low – at least towards the latter portion of the exchanger – the non-condensable molecules remain in close proximity to the cold exterior of the tube. Thus, they prevent efficient contact between the tube and the condensing steam. Or, in more conventional terms, the steam side heat transfer film resistance increases.

To the observer, it appears as if the exchanger is fouled. And so it is, by vapour binding.

Depropaniser reboiler tube leak
The author was once in charge of the operation of a 23 000 b/d alkylation unit C3-C4 splitter. The 30 psig steam reboiler had developed a tube leak. The rate of heat transfer, as measured by the reduction in the 30 psig steam flow, had declined by approximately 50%. However, the operators were easily able to restore the heat transfer rate by venting the channel head. Perhaps the most interesting aspect of this was that venting (see Figure 2) from point A at the high point of the channel head was largely ineffective at restoring the rate of reboiling. Only when the vent at point B was opened did the rate of steam condensation in the reboiler increase. Point B was located on the channel head, just below the lower channel head pass partition baffle. Venting at point A only serves to vent fresh steam. Venting at B after the steam has condensed purges the non-condensable butane vapours out of the tubes and helps to strip off the non-condensable vapours binding to the outside of the tubes.

Venting from a vertical thermosyphon reboiler
How can a vertical reboiler, with steam on the shell side of the reboiler, which is the typical configuration, be vented for non-condensables? This particular question is often asked when such reboilers are commonly used (see Figure 3). The answer is to violently and periodically blow out the condensate to the sewer, through valve C. Experience has shown that this is a far from satisfactory answer from the operator’s perspective. But there does not appear to be any other alternative.

Carbon dioxide accumulation
Steam is always contaminated with CO2. In earlier days, when BFW was prepared by hot lime softening, CO2 accumulation was a major problem. But, even with modern ion exchange demineralisation plants, there will be some carbonate contamination of BFW, and hence some CO2 in the plant steam. In one refinery in Australia, heat transfer coefficients on a plant debutaniser would slip down over a period of a week by 30-60%, until the CO2 was blown out of the channel head. While the reboiler was a horizontal exchanger, with the steam on the tube side, the refiner had neglected to connect a valve to the nozzle at point B (see Figure 2) and was thus forced to periodically blow the channel head clear of CO2, by bypassing the condensate steam trap.

In addition to vapour binding and the consequent loss of heat transfer rates, accumulation of CO2 promotes the formation of carbonic acid (H2CO3), which is quite corrosive to carbon steel tubes, even at a moderate pH of 6.1

Air leaks in surface condensers
A common problem observed in vacuum towers is excessive back pressure at the ejector discharge due to low heat transfer coefficients in the downstream surface condensers. The main problems noted with these surface condensers are:
• Shell side condensate back-up
• Tube side water fouling
• Vapour binding on the shell side.

While most of the vapour binding in refinery vacuum towers is due to cracked gas rather than air leaks, the literature describing this problem is based on air leaks in steam turbine exhaust vacuum surface condensers.2 Whether one is dealing with hydrogen, methane, ethane and H2S, or nitrogen and oxygen, the problem and the solution are much the same.

Surface condenser air baffle
A typical design U value for a refinery vacuum tower surface condenser may be 180 Btu/h/°F/ft2 (service), and 300 Btu/h/°F/ft2 (clean). In practice, even for clean surface condensers, observed U values are 40 to 80. Much of the problem appears to be due to leakage of steam around the leaf seals or seal strips (see Figure 4). This allows the condenser inlet steam to bypass most of the tubes and overload the downstream ejector.3 The best way to retard this leakage is:
• Seal strips should extend ¼” to ½” beyond the I.D. of the shell, before the tube bundle is reinserted into the shell, so as to crush the strips against the shell I.D
• Seal strip material should be 316 stainless steel, of a special flexible grade, and never copper based alloys
• Three pairs minimum should be used.

In addition, the design of the tube support baffles should allow the vapour trapped just below the air baffle shown in Figure 4 to flow without hindrance to the ejector. Retrofitting an existing condenser shell (see Figure 5) can ensure that this objective will be accomplished.

Cooling water inlet location
In the now idled former Coastal Aruba Refinery, the performance of the vacuum tower pre-condensers was greatly enhanced by reversing the cooling water flow. The general rule for condensers is to have the cold water inlet on the bottom of the channel head and the warm water outlet on the top of the channel head. For vacuum system condensers, the correct practice is to have the cool water inlet as close as possible to the vapour outlet nozzle that is flowing to the downstream ejector. This will minimise the vapour outlet temperature to the ejector. The lower the vapour outlet temperature, the less the water vapour content of the ejector inlet vapour. This unloads the ejector, which may then develop a greater vacuum.