Simple methods solve exchanger problems

Andrew Sloley
Process Consulting Services Inc.
Houston

Simple field measurements can reveal exchanger-operating problems. Understanding the process, the equipment, and how they interact is important for successful troubleshooting.

Even without level bridles, field personnel can identify levels in exchangers. At proper temperature levels, the naked hand is sufficient. At elevated and depressed temperatures, a temperature gun or surface thermocouple will often work.

Two case histories present simple solutions to condenser problems. The first case reviews an unnoticed operating problem with a naphtha-stabilizer condenser. The second is a problem with the steam side of a reboiler on a depropanizer. Reboiler heat input was not being properly controlled with a condensate-flow valve.

Both case histories show that heat transfer system performance depends upon consideration of the auxiliary equipment as well as the exchanger design.

Revamping a stabilizer

The naphtha stabilizer was checked to verify its maximum capability before a crude-unit revamp. The new design was to be built to meet tightened product quality specifications: Reid vapor pressure (Rvp) was to be 13.0 psig and the C ₅+ overhead content was to be less than 0.5 vol %. A test run verified unit performance. Table 1 [43,398 bytes] shows the performance of the system under various conditions. Although a new reboiler would not meet C ₅+ overhead content specifications, 2.2 vol % was deemed acceptable. Fig. 1 [57,308 bytes] illustrates the changes required.

The following information was gathered by studying the system:

• The existing system with clean reboilers had sufficient reboiler surface, 10% more surface than required.
• The existing system, when fouled at the end of the run, would require additional reboiler surface.
• Internals modifications in the bottom section of the tower and a new top section of the tower were required to increase the tower capacity.
• The overhead condenser and reflux drum had sufficient capacity, 10% more than required.

System analysis

After project details were reviewed with operations, as a last check, a field trip was organized with Project Consulting Services (PCS) to gather data to verify that the proposed plan was the best plan.

Local temperatures, pressures, and control system flows were gathered for the condenser system and drum. Fig. 2 [57,102 bytes]shows a general schematic of the overhead system along with measured readings.

The condenser is a flooded condenser. Tower-overhead pressure is controlled by a hot-vapor bypass. The exchanger takes stabilizer overhead, passes it through the exchanger at grade, then up to an elevated overhead drum. This is a common arrangement.

The system is self-regulating. As long as the exchanger has excess surface, the level in the exchanger will be set by condensation in the exchanger and pressure in the overhead drum. The pressure in the overhead drum is set by the composition (defined by the set reflux rate from the drum to the tower) and the overhead temperature (controlled by the hot-vapor bypass). Thus, the exchanger level varies automatically to expose the required heat transfer surface to meet the duty requirement. If the exchanger has excess capacity, the liquid level should be high.

The liquid temperature was 93° F. The vapor temperature from the top of the tower to the condenser was 140° F.

No special instruments were necessary to identify the liquid level; it was easy to find by running a hand around the exchanger. Fig. 3 [61,764 bytes] shows the liquid level detected by shell temperature. The liquid level was low in the drum that day. Low liquid levels in flooded condensers imply that the exchanger is operating near its maximum capability.

Calculations showed that the exchanger should not have had any problems condensing the naphtha-stabilizer overhead. Because the liquid level was low, the exchanger was straining at near maximum capability; the system, however, was only operating at 59% of its projected load.

A review of historical operating data showed that the stabilizer had once operated at rates as high as 196 gpm. With low-liquid levels, rates were only 109 gpm.

When exchangers are limited, one method to improve performance is to increase the logarithmic-mean temperature difference (LMTD) across the service by increasing the flow of cooling utility. In this plant, cooling was provided by seawater, which varied from 61 to 68° F. The cooling-water supply valves to the naphtha overhead stabilizer had to be wide open to achieve maximum available LMTD. Needing maximum LMTD for a given duty is another clear symptom of a limited exchanger.

Although a major source of exchanger problems is fouling, maintenance records for the exchanger showed that it has had little fouling present during previous turnarounds.

The situation was assessed as follows:

• The exchanger is generally clean.
• Sufficient duty to obtain 196 gpm of liquid had been achieved in the past.
• A total of 109 gpm of liquid is currently made.
• The exchanger-condensate level is low.
• Maximum cooling-water flow is required to achieve desired duty.

Field discovery

Process calculations showed that the exchanger should not have any problems condensing the naphtha stabilizer overhead. Without a trip to the field to investigate the equipment, it is easy to conclude from the symptoms that the exchanger is fouled and would need to be cleaned during the next turnaround.

By observing the system in the field, however, it was apparent that a vent line was missing. No vent line ran from the exchanger shell to the drum to vent noncondensables that accumulate in the vapor space. The noncondensables blanket the surface area in the top of the exchanger shell (Fig. 3). The blanketed area has ineffective heat transfer.

During operation, a straight-run naphtha stabilizer would have problems condensing the overhead stream. Intermittently, the overhead drum must be vented to flare to improve condenser performance. Without the venting, insufficient reflux is available to prevent high C₅ losses overhead.

Discussion with the operators revealed that they purged the reflux drum intermittently. After the cooling-water flow valve was wide open, the hot-vapor bypass would eventually close and then the drum pressure would rise. At that point, the operators purged the drum vapor space for 15-30 min and then put the unit back on control.

Fig. 4 [59,954 bytes] shows the events that occur when the operators vent the reflux drum.

First, the hot-vapor bypass valve is closed. Venting the drum of vapor drops the drum pressure. Next, decreasing drum pressure lifts the liquid from the piping to the drum. The exchanger has no liquid inventory, and the bundle is already fully exposed. Vapor is carried through the exchanger by the sudden withdrawal of liquid from the piping volume. Then, the vapor blow-by in the condenser flushes the noncondensables to the overhead drum. As the overhead drum vents, the noncondensables are flushed from the system. Next, the operators put the system back in manual, and everything works fine until the noncondensables build up again.

Purging to the flare line dumps salable product to flare. While this happens infrequently (one to two times per week) in current operations, the frequency of dumping to the flare would increase to a daily (or more often) event after the revamp. Frequent dumps to the flare contribute to operating problems, product losses, and environmental releases.

Although a bigger exchanger would decrease the frequency of dumps to the flare, it merely lets more time pass between flushes.

The solution is to install a manual vent line from the condenser to the overhead drum. When the exchanger starts to be limited, only a small purge would be necessary and the hot-vapor bypass control automatically compensates to maintain condensate drum pressure without forcing unnecessary losses to the flare.

The vent line was added to the project scope. Assuming an LPG cost of $4/bbl and 3 bbl/hr of venting, this solution saved about $95,000/year in product losses. The cost to install the needed jumpover was small, $10,000, compared to the generated savings. In addition, the amount saved does not include the better operation efficiencies caused by the new line.

Steam blow-by case

A gas plant's depropanizer reboiler was built as a fully drained condenser with an external pot on the steam side. Condensate level was controlled in the condensate pot. The set variable was the temperature of the third tray in the depropanizer. The tray temperature was controlled by the steam-inlet valve into the system.

Effectively, the pressure drop across the steam supply control valve set the reboiler's LMTD, which set the overall heat transfer. Fig. 5 [51,652 bytes] shows this system.

The operating superintendent insisted that this control system did not work. In fact, for most of the operation, the condensate pot ran with steam blow-by into the condensate return line because the control valve under the condensate pot was kept wide open.

Reboiler heat input was controlled by varying tower pressure. Tower pressure variations at a fixed temperature resulted in large variations in bottoms composition.¹ Both the steam blow-by into the condensate line and the bottoms composition variation caused a variety of operating problems.

System review

One quick way to start a field troubleshooting effort is to compare a properly operating system with an improperly operating system. Fig. 6 [49,398 bytes] shows a reboiler system that should work, sketched without looking at the existing configuration.

Comparing Figs. 5 and 6 leads to one interesting observation: the steam vent line from the condensate pot in Fig. 5 goes from the condensate pot to the steam line feeding the reboiler. In contrast, Fig. 6 shows the steam vent line from the condensate pot returning to the steam side of the reboiler.

In normal operation, a steam-vent line from the condensate pot to wherever it goes should have close to zero flow. The main purpose of the vent is to prevent entrained steam from vapor-blanketing the pot.

If there is no flow, there is no pressure drop. Thus, the condensate pot's pressure must become the same pressure as the reservoir to which it is connected, which in this case is the feed line to the exchanger. If the condensate pot's vent line returns to the exchanger, gravity flow will drain the shell side of the exchanger.

Because the pressure upstream of the exchanger in the piping and the pressure downstream in the condensate pot are equal, a liquid head must build up somewhere in the system to overcome the pressure drop. If a pipe run from the exchanger to the condensate pot has height to allow for building up a liquid leg sufficient to overcome the pressure drop, the system works fine.

If insufficient height is available in the piping run, condensate builds up in the exchanger bundle until sufficient height is reached to make condensate flow. Fig. 7 [53,860 bytes] illustrates both cases.

Condensate build-up in the exchanger shell forced by system hydraulics can limit exchanger performance. The condensate level in the bundle can only drop to the minimum level required to make the condensate flow to the condensate pot.

If at minimum condensate level, not enough surface is available to meet the tower duty requirement, the system can never meet service duty requirements. In this case, however, steam blow-by through the exchanger was sufficient to keep the surface useable.

Based on these observations, a field trip verified that the condenser was partially flooded. In a normal operating mode, the condensate will sub-cool by up to 36° F. Because the temperature is too hot for naked hand detection, an infrared heat gun or a thermocouple placed under a swatch of fiber glass insulation quickly identified a liquid layer.

Modifying the condensate pot's vent line to connect to the exchanger rather than the steam supply and allowing the flow valve on the bottom of the condensate pot to operate automatically has real savings. Steam that formerly passed through the bottom control valve can damage the hot condensate system, which was not designed for steam. Also, without controlling the steam, much more steam was required to obtain the required exchanger duty.

In this situation, about 2,000 lb/hr of steam was lost. At $4/1,000 lb of steam, at least $70,000/year of energy was saved.

Reference

1. Sloley, A.W., and Martin, G.R., "Process Modeling for Control System Design and Analysis," presented at the IASTED Conference on Modeling, Simulation, and Control in the Process Industry, May 25-27, 1994, Ottawa, Paper 208-007, pp. 107-115.

The Author