Decommissioning concrete C 3 tank poses safety concerns

April 5, 1999
These steel tanks at Winnepeg's Wilkes Ave. peak-shaving station were decommissioned without problem; no structural defects were found (Fig. 1). This concrete tank is the first in LPG service to be decommissioned (Fig. 2).[29,551 bytes] The recent, first-ever decommissioning of a concrete storage tank in LPG service indicates the need for greater safety precautions for tanks that lack a vapor barrier in contact with the stored product. Such precautions are particularly necessary if tank
Zoher Meratla
CDS Research Ltd.
Whistler, B.C.


Gerry Sawchyn
Centra Gas (Manitoba) Inc.
Winnipeg, Man.
These steel tanks at Winnepeg's Wilkes Ave. peak-shaving station were decommissioned without problem; no structural defects were found (Fig. 1).
The recent, first-ever decommissioning of a concrete storage tank in LPG service indicates the need for greater safety precautions for tanks that lack a vapor barrier in contact with the stored product.

Such precautions are particularly necessary if tank entry is contemplated without an inert atmosphere.

In the successful decommissioning of such a tank in La Salle, Man., the quantity of propane released from the prestressed concrete shell was much larger than expected, suggesting entrapment of both vapor and liquid.

Two steel tanks were also decommissioned as part of the same project. Procedures used for them provide a contrast to the ones required for the concrete tank.

For storing LNG, LPG, and other refrigerated products, prestressed concrete tanks are widely used on what is called "double-integrity" tanks; both the inner and outer shells are capable of containing the liquid.

The most common design has a 9% Ni inner container and prestressed concrete outer shell. In this design, however, the inside of the outer concrete shell has a metal vapor barrier, and the perlite insulation used is easily purged.

The second type of design has a stainless-steel membrane inner tank with polyurethane insulation between the membrane and the concrete shell. The concrete is provided with a brushed-on vapor barrier on the inside.

If the membrane develops a leak, however, the polyurethane insulation supporting the membrane presents similar hazards to the La Salle tank because it absorbs flammmable gas which cannot be displaced by purging. Only the Preload tank design has no vapor barrier on the inside.

Three tanks

Deregulation in the natural-gas industry led to decommissioning of the Wilkes Ave. propane peak-shaving facilities in Winnipeg, Man., and at La Salle, 10 km away.

In the first phase, two steel, 24-year-old, 20,000-cu m LPG tanks and a refrigeration plant were decommissioned at Wilkes Ave. Each tank was double-walled, including the roof of the inner tank. Foamglas insulation covered the bottoms.

The shell and dome annulus space was insulated with perlite. The annulus space was maintained under nitrogen purge while in service. Both tanks had external send-out pumps.

In the second phase, one prestressed concrete, 40,000-cu m LPG tank with external send-out pumps and associated refrigeration plant at La Salle, as well as a peak-shaving plant and 12 pressure vessels at Wilkes Ave., were decommissioned.

In operations, a send-out heater at La Salle warmed the stored LPG before transfer to the peak-shaving plant at Wilkes Ave. All three tanks had an elevated concrete foundation slab.

Because of costs, CO2 was used for purging in both phases.

The two steel tanks (Fig. 1) were drained until the send-out pumps lost prime. The residual liquid heel was then vaporized with partial boil-off gas recycle.

Liquid CO2 was vaporized with electric vaporizers. To safeguard against propane leaking through the inner tank, the entire annulus space and bottom insulation were purged.

A grade manway on each tank provided access to survey the atmosphere inside the two tanks when the purge was completed. A visual internal inspection indicated that both tanks were free of any structural deficiency or hydrocarbon deposits.

One tank had a small localized lube-oil stain on the floor that was cleaned with absorption pads. A small amount of lube oil was also found in the LPG-withdrawal line used for venting during part of the purge cycle.

Both tanks were left under CO2 atmosphere until eventual sale for an alternate service or teardown. The top-fill penetration into each tank was left open to atmosphere to allow breathing.

The purge of the refrigeration plant was carried out concurrently with the two tanks. The tanks were completely isolated from the process plant piping. Also, the decommissioned refrigeration plant was fully isolated from the Wilkes Ave. peak-shaving plant, still in operation.

Phase 2

The 40,000-cu m LPG prestressed concrete tank had been in service for 21 years.

Originally, it was built as a single shell. After approximately 10 years in service, the external urethane insulation showed significant degradation. To prevent further degradation, it was decided to add a second shell with a dome without decommissioning the tank (Fig. 2).

The annular space between the inner and outer shells was filled with expanded perlite. A nitrogen purge system was also retrofitted.

LPG, in direct contact with the prestressed concrete, was stored at -45° C. A carbon-steel vapor barrier was provided on the outside of the prestressed concrete (Fig. 3) [62,680 bytes]. The tank roof and bottom were made of steel.

The send-out propane heater was used to vaporize the CO2 needed for purging. Purge holes for the bottom Foamglas insulation were added during decommissioning. The entire annulus space and tank-bottom insulation were purged.

Throughout the decommissioning operation, the project team was mindful of the Staten Island incident Feb. 10, 1973, when an out-of-service LNG tank exploded during internal hot work.1 The accident was attributed to entrapment of flammable gas in the polyurethane/Milar insulation.

The tank at Staten Island had been taken out of service because of a leak in the inner-containment Milar membrane. The outer shell was made out of prestressed concrete designed by Preload Inc., New York City.

The La Salle tank is the first large LPG concrete tank to be decommissioned after long service. Because all of CDS' previous decommissioning experience had been on LNG and LPG steel tanks, the effect of concrete porosity on entrapment of propane after purge completion was unknown.

An information note released in 1992 by Preload indicated that there was no risk of fire or explosion from entrapped flammable vapors because the quantity of such vapors is very small.2

According to Preload, the maximum volume of the air voids in the concrete is 5%, representing 36.5 normal cu m for the La Salle tank.

The company further stated that liquid entrapment was impossible because the natural moisture in concrete freezes during cool down, thereby providing sealing.

During decommissioning, the minimum concentration of CO2 at the end of the purge was set at 98.5%.

Because this tank had top entry only, with a continuous vertical shell-mounted access ladder, personnel entry with life support under a CO2 atmosphere was not attempted. The plant operator preferred aeration of the tank, after establishing safe conditions inside it.

After full aeration of the inner tank, with an air mover placed on the bottom 200-mm diameter fill line, samples taken from the venturi discharge and analyzed with a gas chromatograph indicated that the end propane concentration in air was unstable. When the air mover stopped, the propane concentration in air inside the tank increased.

Because there was no compelling reason to enter the tank at the decommissioning stage, the primary focus was placed on maintaining a safe atmosphere inside it.

Several preservation options were considered:

  • Restore the CO2 atmosphere and leave the tank in the same condition as the two Wilkes Ave. steel tanks. This was rejected because the purging was completed at the beginning of August 1996 when the CO2 steady-state temperature inside the tank was around 30° C. In winter conditions, the tank site can experience ambient temperatures below -40° C. Without CO2 replenishment, such a temperature drop reduces the volume of CO2 inside the tank by approximately 24%. This has the potential of creating an explosive mixture in the air layer above the denser CO2. Monitoring the atmosphere inside the tank will also be difficult.
  • Restore the CO2 atmosphere and keep the inner tank under a small positive pressure. Unless the next option were found to be ineffective, this option was undesirable because it would be costly.
  • Maintain the tank aerated. A 500-mm diameter wind-driven air turbine was installed on each of the east and west dome manways with diameters of 500 mm and 910 mm, respectively.
Both manways were located close to the top of the shell. The predominant wind directions were in the north-south axis.

(The aeration system used on this project may not be suitable for other sites or nozzle arrangement on other storage tanks.)

Air purge

With a single 150-mm diameter air mover placed on the 200-mm diameter bottom fill, a steady-state flammable-gas concentration of 50 ppm vol (ppmv) was measured at the air-mover discharge immediately after purge completion.

This represents 0.23% of the lower explosive limit (LEL) of propane in air (2.2 vol %). Assuming full mixing of the propane-air mass inside the tank, this is a safe flammable-gas concentration.

While delivery of the 500-mm diameter wind-driven turbines was awaited, two 300-mm diameter turbines were installed on the dome manways shortly after aeration. Over a period of 2 weeks when the wind was calm, the propane concentration measured on the turbine inlet reached 4% LEL.

Although still safe, it was considered unacceptable because complete mixing inside the tank cannot be controlled. Undisturbed air masses inside the tank would have higher propane concentrations. Forced ventilation was re-instated until the permanent wind turbines were fitted.

The installed 500-mm diameter wind turbines have a combined design flow rate equivalent to 3.5 volume changes/day under an average wind speed of 13 km/hr.

The variations in wind speed and direction were considered conducive to better air and propane mixing inside the tank than continuous forced ventilation. The following additional provisions were made:

  • The internal valves on the two 200-mm diameter bottom fill and withdrawal nozzles were locked open.
  • A sampling point was fitted on the accessible dome manway on the west side.
  • The air intake to the two bottom nozzles was extended above the snow line.
  • A 75-mm diameter dome nozzle, the only opening on the highest point of the roof, was left open and protected against rain and snow.
A weekly monitoring schedule for the gas concentration was begun. If the flammable-gas concentration reached a steady 75 ppmv (0.3% L

EL) or higher, forced ventilation was to be activated. Fig. 4 [56,303 bytes] shows the variation of the flammable-gas concentration from Nov. 14, 1996, to Oct. 9, 1998, and the ambient temperature.

Over the first winter, as the average ambient temperature rose, the rate of propane release increased as a result of the expansion of the entrapped liquid and vapor. With the warm plateau of the summer months, the propane release began to decline, typical of a diffusion process, whereby the initially high rate of release decreased with time.

For illustration, the minimum volume of propane released to atmosphere from November 1996 to August 1997 is estimated at 1,640 normal cu m. This is derived from the following:

  • Average propane concentration in air at turbine inlet: 34.5 ppmv
  • Typical annual average wind speed based on Environment Canada records: 18 km/hr
  • To allow for fluctuations in wind direction, one turbine was assumed to be operating at full capacity, while the second at one-third capacity only.
If the tank were not ventilated during this period, this partial propane release represents a 3.4 vol % average concentration of propane in air inside the tank that is higher than the LEL and therefore unsafe.

Thus, the assumption postulated that only vapor is entrapped in the concrete pores is not supported by the measurements made. Furthermore, technicians suspected that LPG liquid had accumulated in the voids behind the lead bearing plate shown in Fig. 3.

Using forced ventilation from the tank bottom and controlled conditions, inspectors carried out an internal visual inspection from a small platform inside the dome located directly under the 910-mm diameter manway.

The concrete shell appeared as if new, and there was no evidence of defects. This confirmed that prestressed concrete is a suitable containment structure for refrigerated products. Also, there were no hydrocarbon deposits on the concrete shell.

Insulation

The perlite insulation in the annulus space was found to be wet. This condition was attributed to ice thawing on the outer surface of the inner tank as a result of failed urethane insulation.

In the course of a normal turnaround, the insulation will need to be purge-dried prior to recommissioning.

Structurally, all three tanks were found to be free of any visible defects.

References

  1. Davis, John C., "LNG: Growth or Safety?" Chemical Engineering, May 28, 1973, pp. 50-52.
  2. Questions regarding Preload's design of double-wall prestressed concrete tanks for LNG, Preload, Jan. 13, 1992.

The Authors

Zoher Meratla is principal of CDS Research Ltd., Whistler, B.C. From 1974 to 1977, he was group leader, dynamics and systems, with Vickers Engineering, Southampton, U.K. Since then, he has been principal of CDS, where he worked on low temperature and cryogenic facilities. Meratla holds a BEng in mechanical engineering from Sheffield University and a PhD from Southampton University, both in England. He is a member of several committees including, CSA Z276, ISO TC 197, and WG4 of International Standards Organization on LH2 fueling of aircraft.
Gerry Sawchyn is supervisor of propane operations with Centra Gas (Manitoba) Inc., Winnipeg. He has 35 years' experience with the company.

Copyright 1999 Oil & Gas Journal. All Rights Reserved.