logo


Dec-2017

Silencing hammering in a condenser system

Detailed analysis leading to correct diagnosis meant that only minor equipment changes were needed to cure severe hammering in a naphtha splitter.

HENRY KISTER, Fluor
CASEY MUELLER and MATT GUNN, HollyFrontier

Viewed : 7107


Article Summary

Flooded condensers are, and are likely to remain, the prime tower pressure control methods for total condensers generating liquid product only. The principles of these methods had been described in the literature more than 60 years ago, yet these methods continue to rank among the most troublesome distillation controls. Good understanding of the principles and learning from experience is the key to avoiding potential traps and improving operation for the industry. This article describes a recent experience solving a problem that, until now, has been unknown and poorly understood in the industry.

In its first year in operation, a naphtha splitter experienced intermittent hammering in the line from the tower overhead flooded condenser to the reflux drum. The problem occurred at cold condenser outlet temperatures. In some of the episodes, the hammering was severe, opening a flange and causing damage to a valve and a thermocouple.

Fluor, which was not involved in the original design, was requested to join the HollyFrontier task force investigating the hammering incidents. Our task force identified the large open slots at the top of the reflux drum dip pipe as a likely root cause. We also identified total closure of the tower overhead pressure control valve as another potential source of hammering.

Based on this diagnosis, we blanked the slots near the top of the dip pipe and added a DCS clamp on the overhead pressure control valve that prevents the valve from closing to less than 20%. HollyFrontier also emphasised in its operator training the importance of keeping tight control of the condenser outlet temperature at 180°F. The column was returned to service in March 2015, with no further hammering to date, and experiencing smoother operation.

This article describes our investigation and our solution.

Background
The HollyFrontier naphtha splitter had been installed to help the Cheyenne refinery comply with the requirements of the MSATII rule for benzene in gasoline by removing benzene precursors in the feed to the naphtha reformer (a UOP Platformer). The tower was started up in February 2014.

Process description
Figure 1 shows the HollyFrontier naphtha splitter. Process conditions shown are those at 10:30 pm on 10 November 2014, just prior to the incident described below. Feed to the tower is naphtha hydrotreater (NHT) debutaniser bottoms. This feed is preheated to 290°F by exchange with the tower bottoms. The preheated feed then enters the 9ft ID, 70-tray naphtha splitter. The tower separates a light naphtha overhead product from a heavy naphtha bottom product. The tower is reboiled by a fired heater. Tower bottoms are pumped through the feed preheaters, an air cooler and a sulphur sorber to the Platformer. Tower overhead vapour at 217°F is condensed by an air condenser, which is a total condenser (no vapour product), and the condensate flows into the reflux drum. From the drum, most of the condensate is returned to the tower as reflux, and the rest is pumped to storage via an air cooler.

Tower pressure is controlled at about 29 psig using a control valve in the tower overhead line to the air condenser. There is a bypass around the condenser with a control valve that is manipulated by the differential pressure between the tower and the reflux drum, kept at 5-10 psi.

The condensate line leaving the condenser enters at the top of the reflux drum and is extended inside the drum by a dip pipe that takes the condensate to near the bottom of the drum. To raise tower pressure, the pressure control valve in the vapour line to the condenser is throttled. This reduces the pressure in the condenser, which in turn reduces the delta T (temperature difference between the condensing side and the cooling side) and also sucks liquid back from the drum. This liquid floods some of the condenser tubes, which reduces its condensation area, in turn reducing condensation, causing the tower pressure to rise. Conversely, opening the pressure valve lowers the condensate level, exposing more condenser tube area for condensation, and also raises the condenser pressure, which improves its delta T. Both increase the condensation rate and lower tower pressure.

The condenser bypass line helps keep up the pressure in the reflux drum. If the valve in the bypass line is closed, the pressure in the drum will be the vapour pressure of the condensate liquid. When the condensate is near its boiling point (no subcooling), the drum pressure will be much the same pressure as the pressure at the condenser outlet. In contrast, when the condensate is subcooled, its vapour pressure may be much lower, typically by about 0.4 psi/°F of subcooling in naphtha splitters. The hot vapour bypass (HVB) line and valve prevent the drum pressure from falling too low. When condensate subcooling causes the pressure difference between the tower and the drum to rise, the HVB valve opens and diverts additional hot vapour to the drum. The hot vapour warms up the drum liquid surface, making the drum pressure equal to the vapour pressure of this warmed-up liquid surface (which is warmer than the subcooled bulk liquid). Process liquids are good thermal insulators, so, as long as the subcooled liquid does not reach the surface in excessive or fluctuating quantities, the surface temperature will remain steady and warm enough to sustain the differential pressure set by the dP controller. A detailed discussion is included in Reference 1.

Focus on the condenser and reflux drum
Because the hammering occurred between the condenser and the reflux drum, this area was the focus of our investigation. Figure 2 shows the physical layout of the piping from the air condensers to the reflux drum. Both the vapour and condensate lines descend to the drum with no pockets according to good design practices.1

Figure 3 shows the reflux drum. The drum is a 6.5ft ID x 20ft tangent to tangent horizontal drum. Condensate enters the drum via the 12in nozzle on the top right hand side of the drum shown in Figure 3. This nozzle is equipped with a dip pipe, shown in detail in Figure 4. Condensate leaves the drum via the 8in nozzle at the bottom left of the drum in Figure 3. The vapour bypass around the drum enters via a 10in nozzle at the top left of the drum in Figure 3.

Figure 4 shows a close-up of the dip pipe. The pipe terminates 1ft above the bottom of the drum. The bottom of the dip pipe is open. The dip pipe is equipped with four 2in wide x 12in long slots, starting 6in below the top of the drum, and continuing to 18in below the top of the drum.

During the design of the naphtha splitter, a concern was raised that non-condensables can be present in the column feed at times. The upstream debutaniser may occasionally lose reboil, resulting in light ends (butanes and lighter) being present in the feed to the naphtha splitter. During such events it was envisioned that these light ends would need to be vented from the overhead accumulator; accordingly, slots were added to the dip tube to allow vapours to disengage from the liquid at the outlet of the condenser.

Emphasised in Figure 4 are the areas of the slots compared to the area of the opening at the bottom of the dip pipe. The slot area is 
96 sq in, while that of the opening at the bottom of the pipe is only slightly higher, 113 sq in. The middle of the slots is 4.5ft above the bottom of the pipe. When the drum level is about 50%, the additional head acting on the bottom of the pipe is about 2.25ft higher than that acting on the slots. The total liquid head acting on the slots is about 16ft, compared to a head of about 18ft acting on the bottom of the pipe. With the liquid heads being close, the liquid split between the slots and the pipe bottom will roughly follow the opening areas, 46% through the slots, 54% through the bottom of the pipe. That is without considering any vapour pressure effects.


Add your rating:

Current Rating: 4


Your rate: