Kenneth W. Mei
Chevron Research & Technology Co.
Richmond, Calif.
Accurate tank-temperature measurement is as important as accurate tank-level measurement in determining the volume of liquid in a tank.
For accurate tank volume measurements using automatic tank gauging, multiple-point or averaging automatic tank thermometers (ATTs) should be used.
These ATTs must be factory calibrated and field verified to within the tolerances in the American Petroleum Institute (API) standard. This procedure provides accuracy comparable to that of the automatic tank gauge (ATG) on the tank and a high degree of accuracy for tank volume measurement.
BACKGROUND
Although much has been written about how the accuracy of level measurement affects tank-volume measurement, little has been written about the effects of the accuracy of tank-temperature measurement.
Temperature error leads to the use of the wrong volume-correction factor from the American Society for Testing Materials (ASTM)/API petroleum-measurement table. Use of an incorrect factor results in an error in the net volume (i.e., volume at the reference temperature).
With ASTM D1250/API Table 6A as an example, a 1 F. error in temperature produces a 0.04% error in net volume for that single measurement, regardless of the size of the tank. As a rule of thumb, every 2.5 F. temperature error causes a 0.1% volume error.
TEMPERATURE MEASUREMENT
Determination of the liquid volume in a storage tank involves five major steps:
- Measuring the liquid level by either manual or automatic tank gauging.
- Measuring the free water by water cut, if free water is present.
- Measuring the average tank temperature by either a manual or an automatic method.
- Taking tank samples to allow laboratory determination of the gravity or relative density of the liquid.
- Calculating the quantity of the liquid with the tank-calibration table and appropriate petroleum-measurement tables.
While the equipment and procedures for manual tank gauging have not changed significantly in the past 20 years, new technologies in automatic tank-level gauging and temperature measurement have been developed and are widely used.
Many ATGs claim to have 1/16 in. or 1 mm inherent accuracy, which equals or exceeds the resolution of manual gauge tape. Other benefits compared to manual gauging include reduced manpower, increased safety (less climbing on the tank), and real time inventory information. Many precision, noncontact-type ATGs have been installed in recent years.
Considerations of the accuracy of tank volume measurement rarely discuss temperature-measurement accuracy. This absence results primarily because it is difficult to correlate the inaccuracy of tank-temperature measurement and the error that results from it.
It is easy to visualize a slice of liquid, say 1/8 in. thick, in a large tank and translate it to a volume. It is difficult to visualize how much the liquid in the tank would expand or contract if the temperature were mismeasured by 1 F.
There also has been little emphasis placed on temperature measurement because, until recently, practical industry measurement standards for both manual and automatic tank-temperature measurement have been lacking. Because there was no modern industry standard for accurate ATTs, the ATT was assumed to be part of the ATG system.
In many cases, the overall volume measurement accuracy of an expensive, high-performance ATG is seriously degraded because the ATT does not provide comparable performance.
ATTs that accurately measure average temperature are readily available. Installation can be expensive in floating-roof tanks if a separate stilling well for the temperature sensing element(s) is required.
The impact on volume measurement caused by error in both level and temperature measurement is illustrated in Table 1.
STRATIFICATION
Horizontal temperature differences are typically less than 1 F. for low and medium-viscosity liquids. Somewhat larger differences may be expected in high-viscosity petroleum liquids.
The horizontal temperature differences that occur adjacent to the tank shell do not have a significant effect on the average temperature of the liquid in the tank.
Field test data suggest that the region subject to ambient cooling is usually less than 3 ft from the tank shell, and the region subject to ground temperature is usually limited to less than 3 ft from the tank bottom.
Temperatures in large storage tanks (5,000 bbl or larger) are usually vertically stratified unless the tank contents are thoroughly mixed.
Based on numerous field temperature surveys, vertical temperature differences as high as 5 F. or 3 C. are normal for pipeline terminals receiving products from multiple suppliers. Because this is a common occurrence, the API standard sometimes requires more than one temperature measurement.
This vertical temperature difference may be smaller for crude-oil tanks receiving from a dedicated source or for tanks with adequate mixing.
In tanks with vertical temperature stratification, the temperature gradient is rarely linear. Vertical temperature stratification is caused by the lack of, or insufficient, mixing while the tank receives products of different temperatures and gravities from multiple sources.
The same field surveys also indicate that the hotter parcel does not immediately rise to the top of the tank. This condition complicates the determining of the average tank temperature by interpolating one or two spot temperatures.
In heated tanks, the liquid temperature near the heating coils (normally near the tank bottom) is often much higher than the liquid temperature further away from the coils.
CUSTODY TRANSFER
In Europe, ATTs are commonly used for custody-transfer tank measurements. The ATTs are also used for inventory tax purposes where ATGs have been approved by the local standards' regulatory agencies.
Also common in Europe is the use of multiple-spot resistance temperature detectors (RTDs) to determine average tank temperature. Usually, the temperatures are transmitted by the level transmitter of the ATG "at no extra cost."
The transmitter may average the spot temperatures or the averaging can be done by the tank gauging computer.
The problem of tank temperature stratification is well recognized. Over the past several years, attempts have been made by International Standardization Organization and Institute of Petroleum to address this concern in international standards.
Following the publication of standards on tank-temperature determination by manual method using either mercury-in-glass thermometers or portable electronic thermometers (PETs), API completed the standard on tank-temperature measurement by fixed ATTs for aboveground bulk-storage tanks.
API's philosophy on the use and qualification of ATTs is very similar to its standard on automatic tank gauging:
- An ATT system may be used for either custody transfer or inventory control. The use of an ATT system for custody transfer normally is a matter of mutual, contractual agreement between the buyer and seller, subject to federal, state, or local regulations.
- If an ATT system passes the factory calibration test and the field verification test (after its installation), the ATT system in considered suitable for custody transfer. The test involves comparison with a reference temperature measurement.
- If the ATT is used for custody transfer tank temperature, it must be checked or verified against a reference on a regular basis.
SPOT VS. MULTIPLE
The accuracy of an RTD-based ATT system depends on the following:
- The tank temperature stratification and the location of the temperature sensing elements
- The resistance/temperature characteristics of the RTD
- The accuracy of the ATT system readout equipment.
In general, a spot tank thermometer should only be used when the liquid temperature is considered uniform or when the temperature stratification is small and acceptable. Stratification may be minimal for small production tanks, for tanks storing material of uniform temperature, or for tanks with adequate mixing equipment.
For large tanks, temperature stratification is likely to exist unless the contents of the tank are thoroughly mixed. To meet or exceed the accuracy of the ATG on the tank, the ATT should use either multiple RTDs or an array of averaging RTDs for which high accuracy is needed (e.g., for custody transfer).
(A hydrostatic tank gauge may require only one temperature sensor located between the two pressure sensors to provide comparable volume measurement, provided the liquid in the tank is homogeneous, i.e., the gravity or relative density is constant throughout the tank.)
An average tank temperature is determined by the multiple spot or variable-length RTDs. This practice is consistent with the procedure described in the API standard on manual tank temperature measurement.
ATT TOLERANCES
The overall accuracy of an ATT system cannot be better than the accuracy of any component in the system. The key components of an ATT system are the temperature sensors, signal converter/transmitters, and readout devices.
If the ATT is used for custody transfer or applications requiring a high level of accuracy, the ATT should be first calibrated in the factory.
Two methods are considered acceptable for single-point ATTs:
- The ATT system as a whole is calibrated with constant temperature baths at three or more temperatures covering the operating range. The temperature measured by the ATT system should be calibrated to within 0.5 F. or 0.25 C. at each temperature. The bath temperature is measured with calibrated reference thermometers.
- The components of the ATT system are separately calibrated. The resistance of the temperature sensing element is measured in the bath. The bath temperature and the temperature equivalent to the resistance should be within 0.25 F. or 0. 15 C. at each temperature.
Separately, precision resistors or a reference temperature calibrator should be used to simulate temperature input to the temperature converter/transmitter. The temperature input and the temperature displayed by the readout should be calibrated to within 0.25 F. or 0.15 C. at each temperature.
For field verification, two methods are acceptable:
- Verification by components: The temperature sensing element is checked against a reference PET. The temperature measured by the ATT temperature sensor and by the PET should be within 1 F. or 0.5 C.
The rest of the ATT system (excluding the temperature element) is checked by temperatures simulated by precision resistors or a reference temperature calibrator at three or more temperatures covering the expected tank-operating range. The ATT output or readout should agree with the reference within 0.5 F. or 0. 25 C. at each temperature.
- Verification of a system: The ATT system as a whole is verified against a PET. The PET is lowered near the ATT temperature sensing element to approximately the same depth. The temperatures measured by the ATT and the PET should be within 1 F. or 0.5 C.
This simple method has two limitations.
First, it may generate erroneous results and therefore is unsuitable for heated tanks where uneven heating (e.g., by heating coils) is often encountered.
Second, if the gauging platform at which the manual temperature is taken is too far from the temperature-sensing element of the ATT and if horizontal stratification is large (0.5-1.0 F.) because of sunshine or other reasons, the ATT may fail the test.
The field verification procedure for multiple point and variable-length ATTs, if tested as a system, requires that all temperature elements be submerged in the liquid. Ten temperature readings are taken, evenly spaced, or every 2 ft, covering the entire liquid level.
The appropriate average temperatures measured by the ATT system and the PET are then compared. The tolerances should be within 1 F. or 0.5 C.
FACTORY TESTS
Based on shop test data provided by a number of suppliers of tank temperature sensors several randomly selected, off-the-shelf RTDs were found to meet the factory test tolerance described previously (Figs. 1 and 2).
Of the commonly used RTDs, copper RTDs exhibit better linearity, but the change in resistance is small for the narrow operating temperature range of most tanks (e.g., 50-120 F.). This change in resistance may affect the resolution of the output signal.
Platinum RTDs provide stable performance and less long-term drift. They have better resolution but less linearity compared to copper RTDs.
Nickel RTDs are also used but less commonly than platinum and copper RTDs.
The test data in Figs. 1 and 2 reflect only the accuracy of the temperature elements. The accuracy of the analog-to-digital converter, which is usually an integral part of the newer electronic, digital temperature transmitters, is 0.1% or better of the reading.
The performance of the overall temperature measurement can even be improved by use of a "smart" temperature transmitter to characterize the RTD over the tank-operating temperature.
This arrangement allows an ATT to pass the system test tolerance, even if the RTD slightly exceeds the 0.25 F. tolerance for the element itself. Therefore, the factory-calibration tolerances set forth in the standard are practical and technically achievable.
The factory calibrations for both multiple-point and averaging ATT systems use the same tolerances, except each temperature element of the ATT system is tested.
BIBLIOGRAPHY
API Manual of Petroleum Measurement Standards, Chapter 7.1, "Static Temperature Determination Using Mercury-in-Glass Tank Thermometers."
API Manual of Petroleum Measurement Standards, Chapter 7.3, "Static Temperature Determination Using Portable Electronic Thermometers."
API Manual of Petroleum Measurement Standards, Chapter 3.1B, "Standard Practice for Level Measurement of Liquid Hydrocarbons in Stationary Tanks by Automatic Tank Gauging."
Berto, F.J., "Method for volume measurement using tank gauging can be error prone," OGJ, Mar. 13, 1989, p. 57.
Mei, K.W., "Automatic tank gauges can be used for custody transfer," OGJ, Nov. 13, 1989, p. 81.
Copyright 1992 Oil & Gas Journal. All Rights Reserved.
