INADEQUATE INSPECTION CAUSE OF FLAWED VAC TOWER REVAMP

Norman P. Lieberman, Elizabeth T. Lieberman
Process Improvement Engineering
Metairie, La.

Thorough inspection during a revamp turnaround and prior to restreaming is a vital factor in the success of revamps.

Three serious problems came to light when a vacuum tower was recommissioned after a revamp, without adequate inspection. This revamp failed because the problems were not found and resolved during revamp turnaround inspections.

The purpose of inspection is to ensure that all new equipment is correctly installed and that all existing equipment is in good working order. Inspection also serves as a final defense against design errors.

An earlier article showed how careful inspection and good communication between all parties avoided some severe operational problems.' This article illustrates how inadequate inspection can destroy most of the expected benefit of a revamp.

LVGO DRAWOFF

The principal objective of this vacuum tower revamp was to increase the production of 650-950 F. boiling range gasoil, or light vacuum gas oil (LVGO). This cut was to be the main component of the hydrocracker feed.

A simplified sketch of the revamped vacuum column is shown in Fig. 1.

The calculated liquid flows from the LVGO collector tray were:

Pumparound, 28,000 b/sd
LVGO product, 16,400 b/sd
Internal reflux, 4,900 b/sd.

The internal reflux was to overflow the collector tray onto an internal trough-type liquid redistributor. After unit start-up, the following rates were observed:

Pumparound, 24,000 b/sd
LVGO product, 5,300 b/sd
Internal reflux, 11,800 b/sd.

The internal reflux rate was calculated on the basis of an overall vacuum tower heat balance. Obviously, the production rate of LVGO was unsatisfactory to the client. Because 17,100 b/sd of liquid was condensing in the LVGO pumparound section, Process Improvement Engineering concluded that the LVGO collector tray was either leaking, overflowing, or both, at a rate of 11,800 b/sd.

Product pump capacity was proved not to be the limiting factor by the observation that the product pump would lose suction pressure at LVGO production rates above 5,800 b/sd.

A 1,000 b/sd reduction in pumparound rate would permit an increase in product rate of only about 250 b/sd. This fact indicated that part of the diminished LVGO production was apparently caused by tray leakage.

The initial perception of the problem was that the drawoff nozzle might be partially plugged. While this is a common enough problem, it was not consistent with the observed symptoms.

If the nozzle had been plugged, it would be possible to increase the product rate by 1,000 b/sd by simply reducing the pumparound rate by 1,000 b/sd.

The vendor that designed, fabricated, and installed the vacuum tower internals was contacted. Vendor specialists, to their credit, immediately discovered that the collector tray, as designed, could pass only 17,000 b/sd of product plus pumparound before it would begin to internally overflow through the three downcomer boxes (Fig. 2). Note that the design rate for product plus pumparound was 44,400 b/sd.

Both the drawoff nozzle and the drawoff sump were adequately sized for the 44,400 b/sd flow. The access to the drawoff sump, however, was restricted by the large downcomer boxes shown in Fig. 2. Because the sides of these boxes were elevated only 3 in. above the collector tray deck, they easily overflowed.

An additional problem was created by the 6-in. gap between the last downcomer box and the edge of the vapor chimney. This, in turn, caused an impossibly high (i.e., calculated) head loss of 20 in. of liquid, at the design liquid drawoff rate.

Fig. 3 shows a properly designed collector tray. In this standard design, the drawoff sump is fed from a center trough which runs the entire length of the collector tray. Small downflow pipes are used, rather than the larger downcomer boxes. The center trough provides easy access for the liquid to reach the drawoff sump.

The reason the tray vendor chose to discard the standard center drawoff sump design is beyond the scope of this article. The important point, however, is that even the most reputable and experienced tower internal vendors can make mistakes.

The refinery engineering staff, or the process design consultant, must review the hydraulic calculations for each tower internal. Further, any deviation from standard design practice should be questioned. One should never accept "we've done it this way before and it worked," or think "they are the world leaders in vacuum tower internal design and their designs should not be questioned."

ATM. TOWER TRAYS

The refinery had a long history of dislodging the bottom steam stripping trays in the atmospheric crude distillation tower. The principal function of these trays, as shown in Fig. 4, is to reduce the 400-650 F. boiling range material that has to be condensed in the vacuum tower overhead precondenser.

Such stripping trays typically are damaged by water in the stripping steam, Process Improvement Engineering's critical flow, restrictive steam sparger, as shown in Fig. 4, was installed to help prevent tray damage by preventing slugs of water from entering the tower with the stripping steam.

When the tower was commissioned, however, stripping efficiency was less than predicted (Table 1). (The lower the initial boiling point and 5%-point temperature, the poorer the stripping efficiency.)

The pressure drop across the bottom four stripping trays was checked. This analysis showed that the trays were flooding at 75% of their design vapor and liquid loads. An Isoscan survey (i.e., a radiation survey of density at various tower elevations) showed that Tray Deck No. 2 was not intact and that the downcomer from Tray No. 3 (the tray above No. 2) was empty (Fig. 5).

Because the calculated pressure drops for Trays 1-4 were similar, there was no reason to consider operational damage to Tray 2 only. In other words, it would be unlikely for the middle tray, Tray 2, to have been damaged in operation while Trays 1 and 3 remained intact (Fig. 4). It was concluded that Tray 2 had not been installed properly.

Corrosion was also ruled out as an explanation because of the short time between the unit start-up and the Isoscan survey.

This tray installation error is believed to have increased the amount of precondenser hydrocarbon condensate by 1,100 b/sd, and the precondenser duty by 20% (Fig. 1).

The lesson here is obvious: check all tray clips for tightness, and check that the tray deck manways are properly replaced.

PRECONDENSER FOULING

The pressure at the inlet to the primary steam jet was 70 mm Hg, as compared to a design pressure of 55 mm Hg. The design pressure had been demonstrated 2 years before the revamp. Also, the shell-side pressure drop (_P) across the precondenser was measured at 10 mm Hg, as compared to a design _P of 2 mm Hg.

An increase of 10 F. on the precondenser outlet temperature caused the high pressure at the steam jet inlet, as compared to previous experience with comparable cooling water temperatures. Both the excessive exchanger shell-side pressure drop and the poor heat transfer performance are likely consequences of a failure to pull the precondenser tube bundle and clean the outside.

Note that the vacuum tower had been retrofitted at a cost of several million dollars, with structured-type packing replacing valve trays. A primary objective of the revamp was to reduce the pressure drop between the flash zone and the tower top from 45 mm Hg to 12 mm Hg.

The measured _P between these two points did decline, as predicted. The majority of the benefit of reduced vacuum tower flash zone pressure, however, was dissipated because of the fouled precondenser. This fouling apparently worsened in the 2 year interval between the initial process design field test work and the crude unit turnaround.

The lesson here is that, if it is worth $3 million to reduce the flash zone pressure by retrofitting tower internals, it is worth another $30,000 to make sure that the precondenser is cleaned and that the vacuum tower overhead system will perform as originally designed.

GUIDELINES

Although proper inspection prior to restreaming is vital, the inspection process of a revamp is a continuous one that begins with the first idea to revamp the unit.

Immediately before the revamp design work begins, the unit must be checked and optimized in the field. All parties must then have adequate opportunity to review and challenge any new designs or operational changes.

Finally, it is necessary to check that all new equipment is correctly installed, piece by piece, and that all existing equipment is in good working order before restreaming. Short cuts waste time and especially money.