TECHNOLOGY Review of tank measurement errors reveals techniques for greater accuracy

Frank J. Berto
Consultant
San Anselmo, Calif.

• Fig. 1 [16158 bytes]
• Fig. 2 [15514 bytes]
• Fig. 3 [15774 bytes]
• Fig. 4 [15610 bytes]

This first article in a two-part series traces the history of tank measurement accuracy, reviews the sources of measurement errors, and recommends procedures to increase measurement accuracy. The second article will detail current measurement techniques and describe the advantages and disadvantages of each. The second article also will describe the author's concept of an ideal tank gauging device.

Measurement history

In the 1860s, oil measurement accuracy was +5%, and the price of oil ranged from 10¢/bbl to $10/bbl. In 1866, oil producers agreed to give buyers an allowance of 2 gal for every 40 gal gauged, to cover spillage, evaporation, and measurement errors. In other words, for every 40 gal of crude oil purchased, the buyer received 42 gal. (This is the origin of the 42-gal barrel.)

In the 1960s, crude oil sold for $1-2/bbl and custody-transfer accuracy was about -0.5%. This loss was caused mainly by errors in temperature measurement, sampling, sediment and water analysis, and API's infamous "Table 6," which will be explained later in the article.

During this decade, the buyer paid for 200 gal of crude oil and received 199 gal. This level of accuracy was built into most of the 1965 API measurement standards. The loss was accepted by buyers and sellers. It was standard practice, therefore, to insure a cargo of crude for 99.5% of the quantity listed on the bill of lading.

By the late 1970s, the price of crude oil had risen to $20-30/bbl. In many cases, the seller was a government and the buyer was an oil company. The oil company's profit was a small fraction of the value of the crude oil, and the 0.5% loss was significant.

Oil companies made a concerted effort to improve the accuracy of oil measurement and to avoid measurement losses. As a result, the API Manual of Petroleum Measurement Standards (MPMS) has been rewritten entirely since the 1970s.

Today, if tank-gauged custody transfer is performed according to state-of-the-art procedures, accuracy of ±0.25% can be achieved. With today's accuracy, a producer receives 399-401 bbl oil for every 400 bbl purchased.

Measurement accuracy

To measure oil with the highest possible accuracy, tank measurement should not be used. Tank measurement is a compromise between accuracy and economy.

Meters and provers are the best way to measure standard volume with a high, provable accuracy. The oil industry, however, uses tank gauging to avoid the expense of meters and provers.

The quantity of oil in a tank can be measured manually or with automatic tank gauges (ATGs). Both methods involve a three-step process:

1. Determining the volume by measuring the level of liquid in the tank. This can be done by measuring either the "innage" (the liquid height) or the "outage" (the vapor space above the liquid).

2. Determining the temperature by measuring the average temperature of the liquid in the tank.

3. Determining the quality by analyzing a tank sample or a line sample.

Volume measurement

Regardless of the quality of manual or automatic tank gauging, the accuracy of volume measurement is limited by the inherent imperfection of the tank.

A tank is not a perfect can. Filling a large tank causes the bottom to sink, the shell to bulge, and the roof or top course to drop. The shell dimensions also change with temperature. These small movements are neither predictable nor repeatable.

The following accuracy limitations are listed in the new API MPMS standards for manual and automatic tank gauging:

1. Accuracy of the manual gauging tape or ATG-±1/8 in.

2. Accuracy of the tank capacity table, including the effect of tank tilt and hydrostatic pressure-±1/2 in.

3. Tank shell expansion due to liquid head (A full tank takes the shape of a wooden barrel. This effect is normally included in the calculation of the tank tables. The accompanying movement of the lower datum plate or the upper reference point is not included.)

4. Undermeasurement due to tank bottom movement-±1/4 in. (This varies with the compression strength of the soil under the tank.)

5. Movement of the gauging well ±1 in. (Floating-roof tanks are normally fitted with slotted gauging wells. Vertical movement of the gauging well affects measurement of outage and causes an error when converting measured outage to innage. Radar and servo-operated ATGs measure outage, and they often are installed on gauging wells. If the gauging well is improperly installed, these ATGs cannot deliver their high potential accuracy. Chapter 3, Section 1B of the API MPMS describes how to support gauging wells properly for minimum movement [Figs. 1-4].)

6. Movement of the datum plate affects innage measurements-±1/2 in. (When a tank is filled and the shell takes a barrel shape, the bottom bulges up near the shell. Further inward, the bottom drops because of hydrostatic pressure. The datum plate should be located 18-30 in. from the shell to minimize the effect of bottom movement.)

7. Encrustation. (This may be significant when heavy or waxy products are stored in small tanks. It results in "overmeasurement.")

8. Thermal expansion of the tank shell and the gauging well-±1/8 in. (Thermal expansion causes two errors because it affects both the tank diameter and the tank height. Tank capacity tables are calculated for one temperature [60° F.]. They do not correct for thermal expansion of the tank shell. When outage ATGs convert from measured outage to innage, there is no correction for vertical thermal expansion of the gauging well for floating-roof tanks, or of the tank shell for cone-roof tanks. The amount of error depends on the product temperature and the ambient temperature.)

9. Human errors. (These are greater with manual gauging than with ATGs.)

The preceding error estimates are the author's calculations of the magnitude of typical level errors for transfers from large tanks. It is important to note that the errors are both positive and negative, and sometimes compensate for one another. These errors have been discussed in detail in the articles listed in the bibliography.

In addition to the preceding, tank measurement is more accurate when tanks are full or nearly so. Measuring small parcels by tank gauging leads to serious errors, because the measurement errors account for a much larger fraction of the total volume.

Temperature measurement

It is the author's belief that errors caused by poorly measured temperatures are much greater than those caused by poorly measured levels. In fact, level measurement errors are the third leading cause of errors. A temperature error of 2.5° F. is the same as a level error of 0.1% (0.5 in. in a 40-ft tank).

Both manual temperature measurements and those made with automatic tank thermometers (ATTs) are inaccurate. Manual measurement made by cup-case thermometers normally err by at least 2-3° F. The greater the difference between the oil temperature and ambient temperature, the greater the error.

Oil in unmixed tanks is probably temperature-stratified. Chapter 7, Section 1, the API MPMS standard for manual temperature measurement, recommends taking three readings for tanks taller than 10 ft. The standard also lists lengthy immersion times and an elaborate procedure for obtaining accurate manual temperature readings. Most users do not take the time to follow the standard and, as a result, record inaccurate liquid temperatures.

There is a rule of thumb for estimating the size and direction of this error: A manually measured temperature will be in error by about one tenth the difference between the oil temperature and ambient temperature. For example, if oil at 100° F. is measured when the ambient temperature is 30° F., the manually measured oil temperature will be recorded as 93° F. and the buyer will be billed for 0.3% more oil than he receives.

The simple solution to this problem is to use a digital electronic thermometer. This device costs less than $2,000 and measures oil temperature accurately and quickly. It reaches temperature stability in about 1 min and indicates when the reading is stable.

This kind of thermometer pays for itself in manpower savings alone, and the savings resulting from increased accuracy can be even greater.

It should be noted that the rule of thumb for manual temperature measurement was determined using a digital thermometer to measure the true oil temperatures after gaugers or inspectors measured inaccurate temperatures using cup-case thermometers.

Although ATT measurements are accurate, only about 5% of the oil tanks in the U.S. are equipped with the devices. Chapter 7, Section 4, the API MPMS standard on automatic temperature measurement, describes the required equipment. Most oil companies are not convinced that the equipment represents a necessary investment.

When purchasing a high-performance ATG to obtain accurate level measurements, one should choose an ATG that has automatic average temperature measurement. This adds about $1,500 to the material cost of a level-only ATG system, but is well worth it.

Sampling, analysis

"Undermeasured" sediment and water is the second largest source of error, especially when crude oil is being gauged. Two factors are responsible for this undermeasurement: difficulty in obtaining representative samples and problems with analyzing the samples correctly.

For clean products, manual sampling usually produces representative samples for determining product qualities and specific gravity. Crude oil and heavy fuel oil, on the other hand, contain, in addition to the product for which the buyer is paying, sediment and water, for which the buyer does not want to pay. Of the sediment and water, the water is the important component. Measuring sediment is an old oil industry tradition.

Simply put, one cannot determine the true water content of the material in a tank by collecting tank samples manually and measuring the depth of free water at the gauging well. This, however, does not stop people from performing crude oil custody-transfer in this manner. These measurements of sediment and water are inaccurate, but better than none at all.

If the depth of free water at the bottom of a tank were 3 ft, that would be significant quantity for measurement purposes. A depth of 3 in., however, is meaningless. Chapter 8, Section 1, the API MPMS standard on manual sampling, spells it out:

"Automatic samplers are recommended whenever representative samples are required for sediment and water and density measurement."

Measurement uses

Tank measurements are taken for three purposes:

• Operations-This requires the least accuracy. The main purpose is to avoid overfilling or emptying a tank.

• Inventory control or stock accounting-This requires an intermediate level of accuracy. Accounting and loss-control programs work better with accurate measurements, but there are timing problems associated with measuring changing tank inventories. Many tank measurement errors cancel each other, especially when a large number of readings are involved. Despite these facts, some refiners are successfully using tank measurement for loss-control programs that routinely reveal losses of 0.1%.

• Custody transfer-This requires the highest level of accuracy because money is changing hands. Metering is the most accurate method of custody transfer. Most metering installations can measure to an accuracy of ±0.1%; the best can achieve an accuracy of ±0.05%. In contrast, typical tank-gauged custody transfers have an accuracy of 0.5%; the best possible accuracy is ±0.25% on a full tank parcel.

In Europe, tank measurement practices for custody transfer are different from those in the U.S. Europeans routinely use ATGs for tank level measurement and ATTs for temperature measurement.

Oil taxes are much greater than the value of the oil in Europe, and custody transfers normally are witnessed by government inspectors. The author believes European oil companies were forced to install ATGs and ATTs because government inspectors refused to climb tanks to witness manual measurements.

In the 1970s, European standards for automatic tank level and temperature measurement were more stringent than the equivalent U.S. standards. This is no longer true. In fact, API and International Standards Organization (ISO) standards are being written by the same people.

Current API standards for ATGs and ATTs define two performance levels: one for custody transfer and another for inventory control. In the U.S., ATGs normally are used for operations and inventory control, and custody transfer usually is done by manual tank gauging or metering.

API now has standards that describe custody-transfer ATGs and their installation, but very few tanks in the U.S. are equipped to meet the custody-transfer requirements.

Gauging wells

About 4 years ago, environmental authorities declared war on slotted gauge poles. Covering the slots on a gauge pole converts it to a guide pole, which serves only to keep the floating roof from rotating. Unslotted gauge poles have proven worthless for level measurement, temperature measurement, and sampling.

API established a task force to address this problem. The group conducted wind tunnel tests to measure evaporative losses from slotted gauge poles. The evaporative losses were significant-similar in magnitude to losses from a floating roof with only primary rim seals. For tanks equipped with upgraded rim seals, the slotted gauge pole is the largest source of evaporative loss on a floating-roof tank.

The task force then measured losses from various "improved" slotted gauge poles. Many of these improvements required a tall float in the gauge pole. The float adversely affected the performance of servo and radar ATGs. In addition, it was cumbersome to remove for manual level measurement, temperature measurement, or sampling.

The task force's latest set of tests measured losses for tanks equipped with a gasketed cover, a pole wiper, and a sleeve suspended from the cover. The tank was not equipped with a float. The evaporative loss was reduced to 1/40 of the base case.

The API task force has produced an environmentally effective solution that still permits the use of a slotted gauge pole without a float. The next step will be to convince the various environmental regulators.

API standards

API's tank measurement standards have been significantly revised since the 1960s to reflect the ten-fold increase in the price of oil. Many ISO tank measurement standards are being revised to agree with the API's. These revisions had three main goals:

• To correct standards that contained significant measurement mistakes, such as API's Table 6, the standard for volume reduction to 60° F. (MPMS, Chapter 11.1).

• To incorporate new, more-accurate procedures, laboratory tests, and measuring equipment

• To outline procedures that permit custody transfer by automatic tank gauging rather than manual gauging.

Here is a brief summary of the current API standards:

• Chapter 3, Section 1A, Standard Practice for the Manual Gauging of Petroleum and Petroleum Products, was issued in 1994. It describes the uncertainties of tank measurement and includes a procedure for checking the calibration of gauging tapes.

• Chapter 3, Section 1B, Standard Practice for Level Measurement of Liquid Hydrocarbons in Stationary Tanks by Automatic Tank Gauging, was issued in 1992. It describes the accuracy limitations of tank measurements and provides installation, calibration, and verification procedures for different types of ATGs. It differentiates between custody transfer and inventory control measurement. It describes the gauging wells that are required for custody transfer with outage ATGs.

• Chapter 3, Section 2, Standard Practice for Gauging Petroleum and Petroleum Products in Tank Cars, was issued in 1995. It covers the procedures for level and temperature measurement for both atmospheric and pressurized railroad cars.

• Chapter 3, Section 3, Standard Practice for Level Measurement of Liquid Hydrocarbons in Stationary Pressurized Storage Tanks by Automatic Tank Gauging, was issued in 1996. It is based on European practices for the installation, calibration, and verification of servo and radar ATGs. To provide custody-transfer accuracy, most existing bullets and spheres will have to be revamped to incorporate the necessary internal gauging wells.

• Chapter 3, Section 4, Standard Practice for Level Measurement of Liquid Hydrocarbons on Marine Vessels by Automatic Tank Gauging, was issued in 1995. It has no companion manual standard. Marine measurements normally are not used for custody transfer.

• Standard 2545, Method of Gauging Petroleum and Petroleum Products, was issued in 1965. All sections of this standard have been superseded and it has been canceled.

• Chapter 7, Section 1, Temperature Determination Using Mercury-in-Glass Tank Thermometers, was reissued in 1991. The key change is much longer immersion times. With heavy oil, 45 min is required to read the correct temperature with a cup-case thermometer, if the thermometer is moved up and down continuously.

• Chapter 7, Section 3, Temperature Determination using Portable Electronic Thermometers, was issued in 1985. Digital electronic thermometers read the correct temperatures in less than 1 min. Considering that a 2.5° F. temperature error results in a 0.l% measurement loss, it is surprising that cup-case thermometers are still used.

• Chapter 7, Section 4, Static Temperature Determination Using Fixed Automatic Tank Thermometers, was issued in 1993. Large tanks normally are temperature-stratified, and average temperature measurement should be used for custody transfer.

• Chapter 8, Section 1, Standard Practice for Manual Sampling of Petroleum and Petroleum Products, was issued in 1981 and revised significantly in 1995. In the author's opinion, the 1981 standard better defined the need for automatic samplers.

• Chapter 8, Section 2, Standard Practice for Automatic Sampling of Petroleum and Petroleum Products, was issued in 1982 and significantly revised in 1995. It incorporated the results of an enormous amount of test work by the API working group to develop automatic line samplers and to install them properly where the line contents were well mixed.

• Chapter 10, Section 7, Standard Test Method for Water in Crude Oil by Karl Fischer Titration (Volumetric), and Chapter 10, Section 9, Standard Test Method for Water in Crude Oil by Coulometric Karl Fischer Titration, are rapid and accurate procedures for water determination. The centrifuge tests described in Chapter 10, Sections 3 and 4, do not measure all of the water present in the sample.

• Chapter 12, Section 1, Calculation of Static Petroleum Quantities, Part 1, Upright Cylindrical Tanks and Marine Vessels, was issued in 1996. It provides a consistent sequence of calculations to determine standard volume from tank measurements.

• Chapter 16, Measurement of Liquid Hydrocarbons by Weight or Mass, Section 2, Mass Measurement of Liquid Hydrocarbons in Vertical Cylindrical Storage Tanks by Hydrostatic Tank Gauging, was issued in 1994. This standard was "fast tracked" from ISO Standard 11223-1. It describes how HTGs should be installed, calibrated, and validated for mass measurement.

• Chapter 18, Section 1, Measurement Procedures for Crude Oil Gathered from Small Tanks by Truck, was issued in 1990. It covers simplified procedures for tank level and temperature measurement, tank sampling, and sediment and water analysis for small parcels of crude oil.

Measurement rules

After nearly 40 years in the oil measurement field, the author has developed five essential rules:

1. More oil is mismeasured because of thermometers than because of tank gauges.

2. Measuring small parcels of oil in large tanks is inaccurate.

3. Taking tank samples of crude oil and measuring the water level at the gauging well keeps tank gaugers busy, but is inaccurate. Measuring sediment and water requires a line sampler.

4. Measuring outage is accurate only for measuring the vapor space above the oil. To determine the oil volume, measure the innage.

5. Using an unslotted gauging well is a measurement disaster. It gives inaccurate levels and temperatures and produces a nonrepresentative sample. It also makes it easier to overfill tanks.

Bibliography

Berto, F.J., "Control program halves crude losses," OGJ, Dec. 27, 1982, p. 173.

Sivaraman, S., and Thorpe, W.A., "Measurement of tank-bottom deformation reduces volume errors," OGJ, Nov. 3, 1986, p. 69.

Sivaraman, S, and Holloway, C.J., "Method measures cylindrical storage tank reference height variations," OGJ, Dec. 12, 1988, p. 50.

Berto, F.J., "Methods for volume measurement using tank gauging devices 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.

Sivaraman, S., "Field tests prove radar Tank Gauge Accuracy," Oil & Gas Journal, Apr. 23, 1990, p. 89.

Berto, F.J., "Hydrostatic tank gauges accurately measure mass, volume, and level," OGJ, May 14, 1990, p. 57.

Sivaraman, S., and Sheppard, R., "Minimum Transferred Volume Necessary for Accuracy when Determining Custody Transfer Volumes by Tank Gauging," Petroleum Review, August 1991.

Berto, F.J., "Gauging data pose question on stability of reference gauge heights," OGJ, July 29, 1991, p. 78.

Mei, K.W., "Accurate automatic temperature measurement reduces tank volume errors," OGJ, July 20, 1992, p. 105.

Mei, K.W., "Unslotted gauge wells cause tank-level measurement errors," OGJ, Jan. 30, 1995, p. 89.

The Author