DESIGN, INSTALLATION PITFALLS APPEAR IN VAC TOWER RETROFIT

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

Retrofitting a vacuum tower with structured packing can improve gas oil quality and increase both resid throughput and gas oil recovery. However, many vacuum tower retrofits achieve less than predicted improvements.

Observations made during a vacuum tower turnaround illustrate some of the problems that can be encountered. A simplified sketch of the revamped vacuum tower is illustrated in Fig. 1.

TOWER NOT ROUND

One of the first surprises of the turnaround was the discovery that the vacuum tower was out-of-round.

Structured packing arrives at the construction site in precut blocks. Extra sheets of packing are available to insert into minor gaps in the assembled packing blocks that occur when the tower is not truly round. Therefore, the out-of-roundness did not create a big problem.

The internal supports, however, were quite another matte r.

For example, because the 30-ft diameter center section (Fig. 1) was out-of-round, an I-beam support below the heavy vacuum gas oil (HVGO) chimney tray was found to be several inches too short. As a result, the slotted holes in the I-beam did not line up with the predrilled holes in the brackets, which had already been welded to the vessel wall.

The workmen who installed the supports dealt with this problem by drilling new holes in the I-beam. The I-beam was then bolted into place and the chimney tray center trough laid across the I-beam.

Unfortunately, the new holes they drilled were not slotted. Slots are required to permit differential rates of thermal expansion between the I-beam and the vessel wall.

HOLDDOWN MEMBERS

The operating company had specified that the packing holddown grids be designed to withstand an uplift of 1.0 psi. This translates to a pressure drop of 40 in. of liquid per bed, or over 20 times the normal pressure drop through the bed.

In a 30-ft diameter bed, the 1.0 psi uplift force is 10,000 lb. Added structural safety factors and conservative mechanical design practices resulted in a massive hold down structure with several large I-beam members to keep the holddown grid intact.

The placement of these I-beams below the pumparound return spray header interfered with the spray cone from about 25% of the spray nozzles and would therefore reduce efficiency of heat transfer in the light vacuum gas oil (LVGO) and HVGO pumparound sections.

More seriously, the wash oil section would also have its wetting uniformity degraded, causing coke formation in the wash oil grid to increase.

Installation of the massive holddown I-beams could begin only after the structured packing was installed. In practice, this meant that the workmen dragged the heavy beams across the structured packing.

While it is possible to walk carefully across structured packing without damage, manipulating weighty beams inside the tower will damage the packing unless its entire surface is covered with precut plywood sections.

While holddown grids should be used above beds of structured packing (Fig. 1), they should be designed for a few inches of liquid pressure drop per foot of packing, without any added structural safety factors. The designer should also consider the ease of installation for all tower internals.

WELDING VS. BOLTING

Welding inside the cramped confines of a tower is difficult, expensive, and time-consuming. The fumes, heat, and light from the welders arc also interfere with the activities of other workers inside a column.

To minimize welding, it is best to fabricate the spray headers and lateral piping with 150# pipe flanges that are easily bolted together inside the tower. Further, the flanged connection assures that each lateral arm of the spray header is level.

Similarly, the hats across the tops of the chimneys should be bolted rather than welded into place. (A typical vacuum tower installation with four chimney trays may have 3,000 chimney tray hat attachments.)

In the design of this vacuum column, the packing vendor achieved considerable fabrication savings by specifying that both the spray header lateral arms and chimney tray hat attachments be welded rather than bolted into place. This factor substantially extended the packing installation time and cost.

Incidentally, never use square, four-bolt flanges to assemble the lateral spray arms. They will leak. Use more expensive six-bolt circular, 150# pipe flanges with gaskets.

SPRAY HEADER INSTALLATION

Cold pumparound return and wash oil are distributed across the packed beds with spray nozzles that screw into the lateral distribution arms. The ends of these arms must not be attached rigidly to the vessel wall. This permits the necessary movement resulting from differential rates of thermal expansion.

The supports for the spray header in this vacuum tower revamp had been properly designed. The lateral arms rested on shelves welded to the vessel wall. To keep the lateral arms from slipping off their shelves, U-bolts were provided.

It was probably the designer's intention to leave a small gap between the lateral arms and the U-bolts to permit movement of the arms. If so, this information was not communicated to the workmen, because the U-bolts had been forcefully tightened against the ends of the lateral arms.

INTERNAL OVERFLOW PIPES

A modern vacuum tower produces LVGO for hydrocracker feed and HVGO for fluid catalytic cracker feed. Efficient fractionation between these two products is a major objective of most vacuum tower revamp projects.

To obtain this objective, the internal reflux below the LVGO drawoff pan must be distributed by a gravity distributor. Typically, six liquid-distribution points per sq ft are required. This is not practical with spray nozzles.

There are many designs for gravity distributors. In the distributor employed in this vacuum tower, liquid from the LVGO drawoff sump overflowed into the orifice distribution trough (Fig. 2). As can be seen, the end of the overflow pipe was 8 in. above the bottom of the distribution trough.

Pilot plant tests have shown that when the pressure drop through the chimney tray exceeds 1 in. of liquid, vapor flow up the overflow pipes will retard liquid flow down the pipes, unless the bottom of the overflow pipes are sealed in the liquid in the orifice distribution trough.

The pressure drop through the chimney tray is calculated as shown in this equation:

[SEE FORMULA]

While the liquid level in this trough is normally 12 in., during unit upsets or start-up, it will be less than 6 in., thereby unsealing the bottom of the overflow pipe. Overflow downpipes must have a positive liquid seal to prevent backup of liquid onto the chimney tray and overflow of liquid down the chimneys.

SIEVE TRAY INSTALLATION

The stripping trays in the bottom of the vacuum tower are often subject to damage-usually because of unexpected slugs of water in the stripping steam supply. The water flashes violently to steam, and the resulting pressure surge upsets the stripping trays.

One way to reduce the force of this pressure surge is to introduce the steam through a restrictive pipe distributor. The holes in this distributor are sized for about 6 psi (i.e., the flow through the distribution holes is in the critical flow regime) in vacuum service.

The distributor is made out of triple-x thick pipe. The mass of distributor pipe promotes flashing of water to steam which pressures up the distributor and trips off the steam supply to the vacuum tower.

To further protect the stripping trays, it is recommended that:

• The trays be fabricated from 10 gauge 410 chrome (normal trays being thinner 14 gauge).

• The edge of each tray section be bolted or welded to cross I-beams, which are in turn clipped to the vessel wall.

• The trays be securely clipped to the tray ring, ensuring that the tray clips are installed "square" to the tray ring (i.e., at 90 to the tangent of the tray ring).

The reason that 410 chrome steel is recommended, rather than the more corrosion-resistant 316 (L) S.S. or 317 S.S., is that the 410 chrome steel series has a coefficient of thermal expansion similar to that of the carbon steel vacuum tower shell.

The 300 series stainless steels have a much greater coefficient of thermal expansion than the 400 series chrome steel or the carbon steel vacuum tower shell.

Even when the sieve trays and tower have a similar coefficient of thermal expansion, the trays will heat up much more quickly than the vessel wall. In large-diameter towers, care must be taken to provide for this differential rate of thermal expansion between tray and vessel wall.

In small-diameter towers, it is sometimes permissible to fix the trays rigidly to the vessel wall by welding, but in most towers it is customary to clip the trays. This allows for the differential thermal expansion and eases the installation.

In this design, however, the vendor had marked his drawings with the following instruction for the tray installers: "Drill holes in the tray ring using the installed tray as a template." This clearly implies that the trays were to be bolted to the tray ring.

This method of installation, which was used in some older designs, represents an ideal in "explosion resistant trays." But in practice, it is rarely seen today, except in small-diameter towers, because of the difficulty of engineering out the problems caused by differential rates of thermal expansion.

In this design, the tray vendor had recognized this problem and had placed 1 1/2-in. slots, instead of bolt holes, around the edge of the trays. This would allow the tray to slip, relative to the stationary tray ring, as the tower was heated.

A trained engineer would know that the holes in the tray ring should be drilled at the outer edge of the slot in the tray because the tray would expand faster than the vessel wall. He would also loosely bolt the tray to the tray ring, using a torque of say, 10-15 ft-lb.

The bolts would then be locked into place by using a second locking nut or by damaging the bolt thread behind the torqued nut.

The problem with the recommended procedure was that the tray installers were not trained engineers and it was not communicated to them. They did not have torque wrenches. Moreover, in the 16-ft diameter stripping section, they had 500 bolt holes to drill in the tray rings, which could only be done after the trays were installed.

To facilitate the job, the tray installers burned out the holes in the tray ring, which distorted the shape of the slots in the tray. The holes were burned through randomly along the length of the slot in the tray. The tray bolts were then tightened as much as possible with an ordinary socket wrench, then the bolts were locked into place.

There are also likely to be problems with tower "out-of-roundness" associated with this type of fixing when engineering the slotted bolt holes on the tray perimeter to provide a consistent allowance for thermal expansion-especially on large-diameter towers.

MATERIAL OF CONSTRUCTION

Because of high-temperature sulfidic and naphthenic acid attack, carbon steel cannot be used in the stripping section of vacuum towers. Therefore, it was specified that the trays be constructed of 410 chrome (i.e., magnetic) steel.

However, after the trays were installed, sections of each downcomer were observed to have some bolted-in rusty plates. The composition of these plates was checked and found to be carbon steel. The vendor's tray component list called for 410 steel for this part.

Should this plate (which was only a small part of the downcomer) have failed because of corrosion, the downcomer liquid seal would have been lost and the downcomer back-up would have caused the stripping section sieve trays to flood.

The time and expense required to correct the aforementioned problems can easily be envisioned.

NEW MANWAYS

Complicating all other difficulties was the requirement to install several new manways and nozzles in the vessel. While new nozzles are costly, they do not greatly complicate the packing installation. New manways are another matter.

The manways are required to install the tower internals. This means, in theory, that the new manways should be completed before the tower internals are brought into the column. In practice, however, the welding, weld X-raying, grinding, and rewelding was ongoing while the manways were used for access by the workmen inside the vessel.

It is probably better to adjust packing heights by a foot or so and use existing manways than to optimize packing heights and install new manways, unless vacuum tower turnaround time is not a factor.

WATER TESTING

With this particular turnaround running several weeks behind schedule, a highly unpopular request that the pumparound circuit be water-circulated prior to startup, was made.

This is done by establishing a water level in the chimney trays and filling the pumparound pump's suction with water. Using this pump, water is circulated through the heat-exchanger pumparound circuit, through the in-line filters, and into the lateral spray headers. One should leave the spray nozzles off at first, and cover the packed bed with a canvas sheet. This will prevent scale and dirt from plugging the nozzles and contaminating the structured packing.

Unfortunately, the nozzles on this unit's spray header had been installed, and they had to be unscrewed and cleaned after they plugged. Also, the orifice distribution trough (Fig. 2) had to be cleaned after water flushing, which was also very difficult.

In addition to plugged nozzles, it was found during water testing that several of the nozzles were missing their internals. Also, all of the chimney trays were leaking and had to be repaired before water testing was completed.

CONCLUSION

The problems outlined in this article are not uncommon. They may even represent the norm. Regardless of the time and expense required to correct these difficulties, it is far better to correct them before the unit starts up, than to ignore them and hope for the best.

The lesson to be learned from this project is that close communication between the process designer, mechanical engineer, tower internal supplier, and craft installation supervisor is mandatory during the entire course of the project.

The scope and complexity of revamping vacuum towers with structured packing makes this of critical importance.

Copyright 1991 Oil & Gas Journal. All Rights Reserved.