TECHNOLOGY Converting LPG caverns to natural-gas storage permits fast response to market

N. Graeme Crossley
TransGas Ltd. Regina, Sask.

• Steps in Cavern Conversions

• LPG Caverns: Normal Operation*

Deregulation of Canada's natural-gas industry in the late 1980s led to a very competitive North American natural-gas storage market.

TransGas Ltd., Regina, Sask., began looking for methods for developing cost-effective storage while at the same time responding to new market-development opportunities and incentives. Conversion of existing LPG-storage salt caverns to natural-gas storage is one method of providing new storage.

To supply SaskEnergy Inc., the province's local distribution company, and Saskatchewan customers, TransGas previously had developed solution-mined salt storage caverns from start to finish.

Two Regina North case histories illustrate TransGas' experiences with conversion of LPG salt caverns to gas storage.

Beginnings

TransGas pioneered de velopment of underground solution-mined bedded salt caverns in Canada and the U.S. in the early 1960s at Melville, Sask. The sole purpose was for dry or brine-free storage of natural gas.

Salt caverns permitted the highest rate of gas delivery, provided a secure and environmentally safe storage, permitted full product recovery, and cost less to construct. All these factors hold true today.

TransGas currently owns, leases, and operates 24 bedded-salt caverns, 6 being converted LPG caverns, representing more than 14 bcf of working-gas capacity.

These washed salt caverns are scattered across Saskatchewan, all developed in the halite portion of the Middle Devonian Prairie Evaporite salt bed (Fig. 1). Steps in the process for cavern conversions are listed in an accompanying box.

The Prairie Evaporite bedded salt is an Evaporitic sequence of Middle Devonian Age (approximately 380 million years) that occurs over most of the eastern part of the Western Canada Sedimentary Basin.

It is underlain by the dolomites of the Winnipegosis formation and overlain by the Second Red Beds and the dolomitic limestone of the Dawson Bay (Fig. 2;Table 1).

In Saskatchewan, salt thickness varies from 0 to 660 ft; formation depths range from 1,640 to 8,860 ft. At Regina, depth to the top of the formation is about 5,250 ft, and total thickness is 460 ft. The Regina site also lies only a few miles from the solution edge of the salt.

The Prairie Evaporite is largely halite in composition; other major constituents include sylvite, carnallite, and anhydrite.

Regina North conversions

All Regina North caverns are leased and were originally used to store propane and butane. TransGas desired to store natural gas at a wellhead pressure (WHP) of 3,000 psig (3,442 psig bottom hole).

The single-entry propane and butane LPG caverns were first used for natural gas service in 1992. All cavern testing, conversions, logging and component changeouts occurred in 1991.

Cavern Nos. 1 and 5

Company records indicate Cavern Nos. 1 and 5 were originally drilled through the bottom of the salt bed into an underlying pressured up, fluid-bearing formation called the Winnipegosis. The original wells were not cement plugged.

The LPG caverns all have only one cemented outer production casing string (usually 7 in.). (All of TransGas' caverns have only one outer cemented production casing string.)

The unplugged drill hole at the Winnipegosis was not critical for LPG operation. An inflow of fluid would help with the requirements for product displacement (cut down on total surface brine requirements). Original in-service LPG operating conditions are identified in an accompanying box.

Brine was used for LPG product displacement. The pillar-to-diameter ratios here are between two and three. Cavern spacing is approximately 600-800 ft between centers.

With this in mind, TransGas was concerned about cavern leaks affecting natural-gas storage. All of TransGas' cavern well casings are cathodically protected; all LPG caverns were to have been cathodically protected also.

The owner company conducted cavern hydrostatic tests to the limits of its water pumps. These tests (short and to a lower bottom hole pressure than the maximum operating gas pressure) were performed on both caverns with a moderate volume of LPG in storage and indicated no leakage.

The initial rate of pressure drop was 1.67 to 3.3 psi/hr. This pressure drop was attributable to nonstabilization effects and some additional water desolutioning of the salt. These results indicated that both caverns had internal leaks.

The remaining LPG product was removed (vented/flared) following these tests. Each cavern was subsequently worked over.

The 4.5-in. casing was pulled, inspected, and bad joints replaced, and heavy wall 5-in. Hydril casing was run over the cavern proper for debrining. A gamma ray-neutron log and sonar survey were run on each cavern.

The wellhead components (valves, seals, extended neck hangers, bonnets, other equipment) were replaced and tested. Production-casing integrity logs or new cement-bond logs were not run.

Nitrogen testing was then done by TransGas to confirm whether the cavern roof areas were structurally sound; nitrogen was used because no pipeline carrying natural-gas, the usual test medium, ran close to this facility. Nowsco Well Service Ltd., Calgary, was brought in to inject a nitrogen gas pad into Cavern Nos. 1 and 5. This would allow for a 3-6 in. roof blanket.

Each cavern was then pressured to the new maximum allowable bottom hole pressure (BHP; equivalent gas pressure) by pumping water down the casing. The cavern pressures were allowed to sit shut-in for approximately 26 days while the pressure decline was monitored.

Each cavern was then repressured with water to the original test pressure and left for 18 days, again while the pressure decline was monitored.

These results indicated the cavern leak rates to be 2.7-3.2 bbl/psi or a brine leakage rate of 26-32 b/d or an equivalent roof gas loss of 35-43 MMcfd (370 Mcf for a sidewall leak; fluid pressure drop of 10 psi/day or equivalent to 0.15 psi/day in a gas-filled cavern due to greater gas compressibility; Table 2 and Fig. 3).

Nowsco was then called back for venting (metering) the nitrogen out of each cavern:

• In Cavern No. 1, 60,000 std. cu m of nitrogen had been injected; 38,688 std. cu m were vented, with 20% of the originally injected volume accounted for in explicable losses.

• In Cavern No. 5, 64,639 std. cu m had been injected; 50,388 std. cu m were vented, with 21% of the original volume accounted for in explicable losses.

Each cavern has a very flat roof span (Figs. 3c and 4).

Recovery of much of the nitrogen that was injected confirmed the soundness of the cavern roof and wellbore areas and that the fluid leaks were occurring lower in the cavern sidewalls, bottom, or floor areas (as suspected from the original drilling and completion records).

A leak in the cavern bottom would pose no operating problem for natural-gas use (brine pond to be monitored and maintained in cavern bottom as a buffer). A leak in the sidewall, however, could have meant abandonment of this project.

Searching for leaks

An industry search was therefore undertaken to discover scientific methods for determining the location of cavern leak points.

The most promising method for leak detection was the new Schlumberger DSI (Dipole Shear Imager) hydrophone tool. This tool was used to pick up any acoustic emissions (microseismic pressure waves) and determine if the leak points originated from the cavern bottom or sidewall areas.

The DSI tool was run in Cavern No. 1 only at maximum bottom hole fluid pressure. The resulting frequency analysis, evaluated from 1 to 40 khz and 200-2,000 hz, was unsuccessful in detecting any acoustic energy (noise) produced by the leak.

The results raised the following questions:

1. Do leaks generate acoustic noise in an incompressible fluid?

2. Would the layer of debris (insolubles) on the floor have blanketed the noise from a floor leak?

3. Would a more sensitive hydrophone array have picked up the signal?

Because the heating season was near and the location of the leak point was uncertain, TransGas decided to defer construction of new facilities to the summer of 1992, which meant postponing construction of the new natural-gas pipeline and the initial filling of the caverns with natural gas.

In December 1991, each cavern was repressured (hydrostatically tested) by pumping brine down the casing to achieve the maximum fluid BHP equivalent to gas storage. The extended pressure decline interval was allowed to continue until August 1992.

The resulting stabilized equilibrium pressures basically matched the hydrostatic brine column to cavern floor. This time period can be considered a mechanical integrity pressure test of the total cavern (Fig. 6).

For communication with the underlying fluid-charged Winnipegosis formation, the equilibrium cavern pressure was 2,682 psig bottom hole or 2,333 psig wellhead (no fluid movement in or out).

This was the "fail safe" pressure level if a sidewall bottom leak existed with no gas being lost. TransGas wanted to store natural gas at 3,442 psig BHP (3,000 psig wellhead) if it could be proven each cavern had a bottom leak that could be kept covered by a small brine pond.

The salt remaining be tween the cavern roof and the top of the salt bed, no evidence of induced secondary permeability in the overlying formations, and the fact that original drilling penetrated into a porous underlying dolomite in both caverns, all indicated the strongest possibility that the caverns were leaking through the cavern floor.

The leakage pathway through the cavern bottom was believed to be very restricted, possibly because of the many feet of insoluble and rubble material covering the cavern floor. Fluid or gas would have to pass through this material to get to the original borehole, through an anhydrite-dolomite lens, and migrate some unknown distance to a Winnipegosis reef with likely very low permeability and restricted reservoir conditions.

With this scenario, there is often a threshold pressure required before any significant fluid movement will take place. This pressure can be much higher or lower than the underlying formation pressure.

The fluid inflow when the pressure is lower than formation may be higher than the outflow when the pressure is higher. As one operates closer to equilibrium pressure, leakage rates should also be lower than at the extremes.

The leak mechanism may function something like a check valve (opening-closing) at different pressure regimes. Although the maximum operating pressure is well under normal fracturing pressure, the presence of an opening such as a cavern changes the stresses in the rock in the area. What the actual leak mechanism might be is unknown.

Gas testing

After examining various options for confirming the leak location and possible gas loss during cavern debrining (initial gas filling), TransGas decided to proceed with a gas test whereby the cavern was initially filled with natural gas down to near the end of the brine casing. The cavern would be pressured to maximum operating pressure with gas and then let sit idle.

A series of density interface logs plus temperature logs would then be run (starting with a base set of logs) over a period of 4 months while pressure was monitored daily. A drop in the fluid interface would confirm the presence of a bottom leak and allow for calculation of the fluid outflow leakage rate.

Both caverns were debrined to the bottom of the inner casing string from Aug. 26 to Nov. 15, 1992, and let sit at their debrining gas pressure level of 2,650 psig wellhead (stabilized) until first produced on Dec. 31, 1992. This was additional confirmation that gas would not leak from the caverns and that no side wall leaks existed.

They were produced down to a low of 740 psig in February 1993 and within days brought back up to higher intermediate pressures of 900-1,700 psig.

On June 8, 1993, cavern leak testing was started by raising the wellhead pressure levels to the maximum operating pressure of 3,000+ psig. Initial base logs were run (interface/temperature). Continuous pressure monitoring was undertaken.

Periodically the pressure was bumped to maximum to keep the driving force active and fluid outflow at its highest level. Two subsequent sets of logs were run over a 4-month time span with the second set run on Oct. 13 confirming that the fluid interface had definitely moved down a significantly measurable amount and proving that the leak points were actually in the cavern bottoms.

Since then, these caverns have been operated at their maximum pressure levels of 3,000 psig wellhead (0.70 psi/ft of depth to cavern roof). Based on finite element stress modeling studies, TransGas limits the minimum operating pressure level to 0.20 psi/ft of depth to cavern roof for the first three operating cycles before allowing the pressure levels to be dropped to the lower limit of 0.15 psi/ft of depth to cavern roof (Table 3).

From actual operating experience, TransGas has found that fluid will enter these two caverns easier at low pressures than fluid will be forced out of the cavern at high pressure.

Cavern No. 1 actually has a fluid inflow rate of 450-285 b/d while Cavern No. 5 has an inflow rate of only 50 b/d. In 1994, TransGas found that fluid inflow had been such that excess brine had to be removed in 1995 from each cavern (along with normal leakage outflow) by displacing it with gas into the existing surface brine/lagoon system in order to regain the maximum cavern usable volume. This is expected to be required every few years.

Once these caverns were debrined, the inner casing strings remained inactive (except for occasional excess brine removal) with all gas injection/production occurring inside the 7 x 4.5 in. annulus. Individual caverns are capable of being produced at 50 MMcfd or greater. Production, however, is limited to 40 MMcfd by surface process equipment limitations (line heater, filter separator, dehydrator).

Cavern No. 8

The third LPG cavern held ethane and was converted to gas storage in 1993 (Fig. 7; Table 3). This cavern has dual-entry wells.

The first lessor drilled a second offset well into this cavern in the 1970s for ease of product recovery. TransGas is the second lessor of this cavern.

Because the original cavern well (plus the offset well) was not drilled below the bottom of salt, leakage from this cavern was not suspected. Past LPG operations seemed to confirm this fact.

Again, both wells have only one outer cemented production casing string. Brine was used for product displacement.

The owner company ran a new cavern sonar survey which indicated a good cavern profile for gas storage. The owner also completed the initial cavern hydrostatic pressure test (run to maximum anticipated bottom hole pressure) which indicated that no leaks were present, a stabilized pressure regime.

TransGas then conducted two separate cavern hydrostatic pressure tests of its own. Both tests were done by pumping water into the cavern to equal the anticipated maximum bottom hole natural-gas pressure (3,000 psig wellhead).

The first test was 24 days' duration while the second test ran for 37 days. Continuous pressure data were recorded.

In both cases, the test results showed a flat pressure-vs.-time graph indicating stabilized conditions with no signs of any cavern leaks. This cavern appeared to be tight.

Based on this information, TransGas decided to extend the natural-gas pipeline 0.25 mile to this third cavern site and perform a natural-gas pad roof pressure test.

In the meantime, the cavern wells were worked over to convert them for natural-gas storage. The inner casing string was pulled, inspected, and replaced, and gamma ray-neutron (GR/N) logs were run. Heavy-wall pipe was installed over the cavern proper for debrining, and all the wellhead components, such as valves, seals, extended neck hangers, and bonnets, were tested and replaced as needed.

No production-casing inspection logs or new cement-bond logs were run.

With completion and tie-in of the new gas-transmission pipeline, the gas pad test was undertaken. Approximately 8,000 bbl of brine were removed by gas injection and the roof gas pad pressured up to 3,060 psig shut-in wellhead pressure.

Within 5 days, the pressure stabilized at 3,050 psig (10 days' test duration). The maximum operating pressure is 3,000 psig wellhead. The results of this cavern integrity test again confirmed that this cavern was tight with no obvious leaks.

This cavern was then completely debrined and put into service for the 1993-1994 heating season. Follow-up fluid interface logging performed in 1994 and 1995 confirmed that after two seasons of cycling gas from maximum to minimum pressures, the fluid interface has not changed position and the cavern is still tight for gas storage (no leaks).

The new inner 5.5-in. casing installed in the offset well was used to debrine this cavern and became a dead piece of pipe once debrining was completed. The 7-in. production casing from the main well is used for all gas injection/production.

Cavern production capabilities are in the 100+ MMcfd range. Actual production, however, is limited by surface production equipment to 40 MMcfd.

Modeling studies

Computer modeling, rock-mechanics testing, and finite element (stress) analysis studies done on the Prairie Evaporite salt of Saskatchewan have revealed several critical parameters applying to cavern geometry and operating pressure levels (for bedded salt caverns):

1. There exists a highly stressed crust or failure and cracking zone around the cavern perimeter approximately 25 ft thick. The saltback and saltbed thickness, therefore, should be greater than 25 ft.

2. A hydraulic gradient or up thrust (force/pressure) exists across the intact salt zone from below the cavern floor; brine leakage into the cavern from underlying fluid-bearing formation may occur. An ideal safe limit would be a saltbed thickness of 50-100 ft.

3. A desirable critical minimum saltback thickness would be 60-80 ft.

4. Safe salt cover thickness (below cavern invert) would be: L/D = 0.21-0.25 (or greater); where L = safe cavern floor thickness (salt), and D = cavern diameter at floor.

5. Ideal saltback floor and roof thickness would be: H/D = 1/3 of cavern width; where H = salt roof thickness, and D = cavern span (diameter).

6. For newly developed caverns or caverns recently converted to gas storage, the minimum cavern pressure will be maintained at 0.2 psi/ft of dept h to cavern roof for the first three operating cycles.

7. The maximum operating pressure limit will be based on a design factor of 0.70-0.80 psi/ft of depth to cavern roof.

8. The minimum operating pressure limit will be based on a design factor of 0.15 psi/ft of depth to cavern roof.

9. Maximum cavern pressure drawdown limits will be based on 360 psig/day (at the high pressure end) and 250 psig/day (at the low pressure end, less than 800-1,000 psig BHP).

10. Minimum cavern-design criteria as set out in Canadian Standards Association (CSA) Standard Z341-93 "Storage of Hydrocarbons in Underground Formations."

The Author