Apr-2013

Sour water strippers: design and operation

Simpler, more cost-effective sour water stripper designs can outperform more generally accepted designs in the refining industry

NORMAN LIEBERMAN
Process Improvement Engineering

Article Summary

In 1965, while working as a process design engineer with the American Oil Refinery in Whiting, Indiana, I designed a sour water stripper to remove ammonia, hydrogen sulphide and phenols from coker and FCC effluent sour water. Many years later, in 2012, I was working in two refineries in India, where I was troubleshooting sour water strippers in both plants. None of these three strippers had worked properly, in that the residual NH3 in the stripped sour water was excessive. All three strippers suffered from low tray efficiency, as well as reboiler and feed preheater exchanger fouling. While conventional, all three stripper designs were needlessly complex. This excessive complexity contributed to their poor performance in the field. As a result of these deficiencies, the stripper bottoms were diverted to the effluent water treatment plant rather than being reused for their intended purpose as crude unit desalter wash water.

There are three general ways to design refinery sour water strippers, but only one of these three designs reflects correct process engineering principles. The two sour water strippers that I was troubleshooting in India both reflected common but incorrect process engineering design.

Purpose of sour water strippers
Refinery sour water originates largely from delayed cokers, hydrodesulphuriser reactor effluents, fluid catalytic cracking units and visbreaker fractionators. The main contaminants are NH3 and H2S. Sour water stripper bottoms are reused in two places:
Wash water for the crude desalter on a once-through basis, to remove chloride salts that promote HCl evolution in the crude tower overhead condensers
Make-up water for hydrotreater effluent recycle wash water to remove ammonia sulphide salts that plug the downstream condensers.
When sour water stripper bottoms are used in the crude desalter, the NH3 content should be about 10 to 20 ppm. Higher NH3 levels interfere with crude unit corrosion control due to chlorides.1 One hundred ppm of NH3 is excessive for desalter operation.

When sour water stripper bottoms are used as make-up to the hydrotreater effluent wash water, an NH3 content of a few hundred ppm is acceptable. After all, the recirculated wash water has over 10 000 ppm of NH3, so the NH3 content of the make-up water is irrelevant, as long as over 90% of the NH3 in the sour water stripper column is stripped out.

Modern conventional stripper design
Figure 1 shows the current conventional design used at most locations. The operating conditions shown are typical of many refineries. Feed is heated by exchange with the stripper bottoms in E-1. The reboiler duty (E-3) is required for three purposes:
• Heat feed from 180°F (82°C) to the 250°F (120°C) bottoms temperature
• Generate internal reflux at tray 32
• Break the chemical bonds between water, NH3 and H2S.

The reflux is generated in the E-2 circulating cooler. The reboiler duty is adjusted so as to control the moisture content (by temperature) of the ammonia-rich gas flowing to the sulphur plant thermal reactor.

This design, while conventional, is not logical. Why preheat the feed onto tray 32 with  E-1, and then remove this heat in E-2? It would have the same effect on the stripping trays to:
•    Bring in the feed colder onto tray 32 (see Figure 1)
•    Reduce the heat extracted in E-2, the circulating cooler, to offset the colder feed.

What would the effect be of eliminating the E-1 feed preheater on the stripping efficiency for trays 30-32? A single field measurement is worth a thousand computer simulation calculations. I conducted a field test by shutting off the circulating reflux (E-2) and bypassing the feed preheat exchanger (E-1) to maintain a constant reboiler duty and a constant tower top temperature. The effect on the NH3 in the stripper bottoms was quite negligible. And why should it change? Will not the stripping factor (the ratio of vapour to liquid) for the stripping trays remain unaltered?
The conclusion is that the preheat exchanger (E-1), the circulating reflux loop (E-2) and trays 33-40 serve no actual engineering purpose.

Older conventional stripper
Figure 2 is based on older conventional sour water stripper designs. In this version, there are only 16 trays rather than 40. Field observations indicate that 15 or so stripping trays are sufficient to remove over 99% of the NH3 from the sour water and an even greater percentage of the H2S. Note that reducing the reflux drum temperature much below 165-170°F (74-77°C) results in salt plugging. The 190°F (88°C) reflux drum temperature is likely conservative (high) and results in increased water in the sulphur plant feed and thus more condensation in the sulphur plant feed NH3 gas knock-out drum (and thus water that has to be recycled back to the sour water stripper).

The design shown in Figure 2 was common in the 1960s. However, it also suffers from the same heat balance drawbacks and needless complications as seen in Figure 1. In other words, a lot of equipment is added to generate reflux when no fractionation is required between the feed and overhead product. Again, the only purpose of the tower is to strip out the NH3 and the H2S.
   
Correct stripper design
In 1969, while working for the now vanished Amoco International Oil Company in the UK, I designed a sour water stripper that eliminated the unnecessary features of the unit shown in Figure 2.

Figure 3 shows the essentials of a correct sour water stripper design. Feed is brought in at ambient conditions (70-100°F, 21-38°C) from the sour water feed tank. To heat the feed from 90°F (32°C) to 250°F (120°C) requires about 16 wt% steam flow, or about 1.3-1.4lb of steam per gallon of stripper bottoms, which is close to a typical design stripping steam ratio for sour water strippers. The E-1 feed preheater, reflux pump (P-2) and the reflux cooler (E-2) shown in Figure 2 are all eliminated. How, then, does one know that the design shown in Figure 3 will work? Because it was built this way (at the Amoco refinery in Milford Haven, Wales, UK) in 1970, where it worked just as well as the conventional design shown in Figure 2.