TECHNOLOGY Site tests validate benefits of cavern probes

N. Graeme Crossley
TransGas Ltd.
Regina, Sask.

• Effect of 10° F. Error at 2,400 psig [12622 bytes]

Acquired data have helped determine gas-in-place inventory volumes, confirm spatial volumes, and assess changes in spatial volumes that may have resulted from cavern creep (shrinkage or closure) or downhole abnormality such as fluid infill or collapse of the side walls or roof areas.

This conclusion of two articles presents details and results of a specific storage-site. The first article (OGJ, Mar. 3, 1997, p. 74) presents background and many of the details and lessons of TransGas' cavern gas-storage probe program.

Melville site

Before TransGas could commit to a full-scale probe installation, a test location was selected where one dual probe could be installed and tested to ensure that data would be accurate and usable.

Melville, Sask., was chosen as the test location because the cavern had a reliable history and was under no excessive demand for system operations.

On the other hand, there was enough activity at this cavern to prove the utility of the probes. Additionally, Melville permitted unrestricted access to equipment for all phases of installation and testing.

Melville storage can be used for both storage injection and production and for transmission-line boosting (Fig. 1 [9588 bytes]). Before probe installation, measurement accuracy was questioned because of pulsations from the reciprocating compressors. Errors of up to 5% were verified.

Whenever the compressor was running, measurement accuracy was assumed to be poor. Table 1 [46003 bytes] identifies the effect of inventory error accumulation.

Verification procedure

Determining the effective volume of a gas reservoir by producing the reservoir, measuring or calculating the reservoir pressure, then plotting the data on a graph of P/ZT vs. Q (Q = cumulative production) has long been a standard in the natural-gas industry.

The same principle can be applied to a storage cavern. The only difference is the lack of internal thermal capacity because the cavern space is completely void of solid material.

Although the spatial volume (VS) was determined at the time of cavern mining by measuring the volume of brine removed from the completed cavern, changes in the subformations may cause this figure to be somewhat inaccurate.

Because it would be too expensive to remove all gas and refill the cavern with brine for an updated VS measurement, the pressure/temperature probe and surface metering equipment can be used to recheck the original Vs figure.

For this recheck, there must be a period of controlled production with the following characteristics:

• Sufficient production to provide good graphical separation on a P/ZT vs. cumulative production (Q) plot.

• Highly accurate surface measurement during the test period, including no boosting operations or disturbances from any source that could interfere with measurement accuracy.

• Accurate cavern pressure and temperature data from inside the cavern.

A two-point (or more) plot on the P/ZT vs. Q plot will give a straight line intersecting the Q line at P/ZT = 0. This will yield VS which equals the effective reservoir volume.

A linear relationship should be present between P/ZT and the cumulative production if the spatial volume does not change and if there are no leaks in the cavern.

The new VS figure can then be compared with the original or any subsequent determinations of VS. Over several years, small variances (typically <5%) can be expected with this approach, provided that the cavern formation is sealed and remains stable.

With the original VS, or any value considered to be more realistic, along with measured pressure and temperature values, the volume of gas in the cavern (Vb) can be calculated to determine the volume of total cavern gas at base conditions.

When this calculation is performed at two different points in time, the difference between the calculated volumes will equal the net change in cavern inventory. This result can be compared directly to the results of surface measurement, for the same period of time, and adjustments made, if necessary.

If this calculation is performed by supervisory control and data acquisition (scada) software, the results are available to gas control continually.

This information allows confirmation of metered information already on the scada system (real cavern inventory) as well as more accurate determination of the "full" condition.

This confirmation is particularly valuable during extended periods of excessive sub-zero temperatures when storage is being produced at maximum rates or if the integrity of the surface-metered values and the amount of remaining usable inventory remain in doubt.

Gas temperatures within the cavern can vary by as much as 20-30° F. from the formation-salt temperature, depending upon the previous operation (injection, warmer; production, cooler) and the elapsed time.

Because the contact area between the gas and the surrounding formation is limited to the cavern wall, stabilization time is usually too long and too costly to have the cavern out of service.

At 2,400 psig, an estimating error of 10° F. for the cavern gas temperature will have an approximate 3.25% error variance on the theoretical inventory calculation (see accompanying box "Effect of 10° F. error at 2,400 psig). This variance is too excessive to offer significant confidence in using this method.

Cavern testing

On the Melville project, a controlled withdrawal test was undertaken to determine the actual rate of temperature change with various rates of withdrawal over specified time periods. Rates of temperature recovery were also documented after each withdrawal period.

The production cycle was selected for the test because accurate volume measurement was required. This meant that the pulsation factor had to be eliminated.

By concentrating on a production interval (gas at high pressure), no compression was required to flow this gas volume into the transmission pipeline. The surface measurement equipment required inspection and calibration just before the test to ensure accuracy.

In order to achieve good graphical resolution, at least two points were required with enough withdrawal to give good separation. Therefore, one half of the usable volume was scheduled to be produced.

Melville North Cavern No. 1 is a very small cavern (only 290,294 bbl spatial volume); therefore, the produced gas volume was somewhat lower than for a full-sized cavern.

The straight line was extended to intersect with the x-axis and establish the cavern spatial volume. This was compared to the original spatial volume determined from the volume of liquid brine removed.

Data from the testing appear in Fig. 2 [20412 bytes].

The temperature changes and the recovery rates are of particular interest. Pressures and temperatures both dropped during production periods and both increased during the shut-in intervals.

Fig. 3 [9771 bytes] represents the same data plotted on a graph of P/(ZT) vs. Q. Note that the line joining the plotted points is exceptionally straight showing a linear relationship between P/ZT and cumulative production.

The spatial volume determined from this graph is calculated to be 1,635.6 Mcf. The original spatial volume determined from the volume of brine removed was 1,629.9 Mcf. This gives a difference of 0.35%.

Since this cavern was completed in 1963, this high degree of correlation indicates a very stable cavern environment (that is, no change in spatial volume in more than 30 years' operation). This conclusion is drawn from the results of a single test.

Future tests will confirm whether the 0.35% figure is realistic or coincidental. This test also verified the accuracy of the probes in providing reliable temperature data which were used in the above determination.

Applications

Not all solution-mined caverns are as ideal in characteristics as the Melville cavern. Following are some problems which can occur:

• Cavern shrinkage ("creep") is caused by the plastic motion of the salt formation surrounding the cavern. This occurs generally when the cavern pressures are lower than the lithostatic pressure.

• Gas migration is caused by permeable strata, poor-quality casing cement, and salt spalling.

• Water migration causes the brine level at the bottom of the cavern to change, thereby changing the cavern spatial volume.

Water can migrate into or out of the cavern if adjacent formations contain water and a path exists as a result of conditions occurring as described for gas migration.

Future testing will be scheduled either during a normal seasonal withdrawal cycle or at a special time when the cavern can be taken out of regular service and "exercised."

Between each test, cavern spatial volume will be assumed to be constant. At the end of each month, surface-metered inventory will be compared to the theoretical inventory. Adjustments will be made, as required, to keep the records accurate continually and eliminate major inventory annual adjustments or write-offs.

Several forms of continual cavern "health" monitoring will also be available with the data provided. Fig. 4 [7602 bytes] shows some of the more obvious conditions that may be encountered; Fig. 5 [27887 bytes] presents sample readings.

• The dotted line represents the normal cavern behavior as determined from previous testing.

• If new tests reveal behavior as characterized by Line A, it must be concluded that brine (a) has been forced out of the cavern into an adjacent formation by the gas pressure, leaving a larger spatial volume, or (b) is entering the cavern as the gas is being withdrawn, keeping the pressure artificially high.

• If the behavior is as characterized by Line B, additional brine has entered the cavern and the spatial volume has been reduced; or, gas is migrating out of the cavern into adjacent formations, causing the pressure to reduce at a rate greater than would be expected from the measured withdrawal.

• Lines C and D represent the shut-in condition where gas is neither injected nor withdrawn and obvious changes inside the cavern are taking place.

Many more scenarios, interpretations, and uses for the data will emerge as time goes on. In any event, we should have more accurate gas inventory records and a greater understanding of each cavern's integrity for long-term usage.

Results, conclusions

The test program at Melville yielded the following results:

1. Regular cavern exercising (that is, controlled production or injection testing approximately once every 3 years) provides an indication of the cavern stability and its suitability for continued or restricted operations.

2. Instantaneous pressure and temperature readings, through periods of nonactivity, provide more conclusive evidence when a cavern leak is suspected.

3. A monthly comparison of calculated volumes to surface-measured volumes yields a monthly adjustment to metered volumes and avoids large volume adjustments at yearend.

4. Year-end audits of cavern inventory are simplified, more easily explained, and available earlier.

5. The effects of pulsation and other factors on the accuracy of surface measurement are a nonissue because the probe information is relied upon for absolute inventory and the surface measurement used for flow rate indication.

6. An accurate measurement of the natural gas volume present in a storage cavern can be found at any time using the current downhole pressure and temperature information and the spatial cavern volume calculated.

In TransGas' experience, the probe data have proven reliable and, when the program is fully implemented, should provide timely and accurate gas-inventory verifications as well as data to monitor the integrity of the company's caverns.

The faster information is received on cavern conditions, the quicker the response to a potential problem to protect a valuable asset.

Also, future storage service requires instant and more accurate volume information to respond to the needs of customers.

As cavern space is now marketed to outside companies, it is imperative that TransGas know the state of its natural-gas inventory. Cavern probes are useful in helping to provide this information.

TransGas will continue to learn more about probes by installing them at key locations.

TransGas' experience also indicates that by developing specialized software that will provide user-friendly information to end users like measurement, accounting, gas control, operations planning, and process and storage engineering, more uses for the information will evolve.

Finally, the need for supplementing probes with surface flow measurement requires further evaluation. At present, surface meters will continue to be required for periodic spatial volume confirmation.

Copyright 1997 Oil & Gas Journal. All Rights Reserved.