Simplified Apparatus for Detection of Leaks in Pressurized Pipelines

1. Field of the Invention

The present invention relates generally to a method and an apparatus for the reliable detection and quantification of the flow rate produced by a leak from pressurized pipeline systems containing petroleum, solvent, or other chemical liquids.

2. Brief Discussion of the Prior Art

There are a wide variety of pressurized pipeline systems carrying petroleum, solvents, and other chemical products that may contaminate or seriously damage the surrounding environment in the event of a leak. In underground or underwater pipelines, where visual inspection is not possible, a leak can be a significant problem. Small leaks in these pipeline systems (e.g., several tenths of a gallon per hour) can go undetected for long periods of time and result in a large cumulative release of product into the soil or groundwater, or into fresh or ocean water.

The need for leak detection capability in pressurized pipelines associated with underground storage tanks containing petroleum products has recently been identified. This need is an important one because the number of tanks involved is very large, and so is the volume of product dispensed through the pipelines associated with these tanks. The pipeline systems in question are most commonly made of steel or fiberglass; they are typically 2 in. in diameter. 50 to 200 ft long, buried 1.5 to 3 ft below grade, and are pressurized at 20 to 40 psi while product is being dispensed. In September 1989, the United States Environmental Protection Agency (EPA) issued technical standards for the detection of leaks in underground storage tanks containing petroleum or other hazardous chemicals and solvents. This regulation established the minimum performance standards that must be met by any leak detection system designed for testing the integrity of underground tanks and/or the pressurized pipelines associated with these tanks.

The EPA requires that underground storage tank (UST pipeline systems that contain petroleum products be tested for leaks either on a monthly or an annual basis. To satisfy the criterion for monthly testing, a system must have the capability to detect leaks as small as 0.20 gal/h with a probability of detection (P.sub.D) of 0.95 and a probability of false alarm (P.sub.FA) of 0.05. To satisfy the criterion for annual testing, a system must be capable of detecting leaks as small as 0.10 gal/h with the same P.sub.D and P.sub.FA required of the monthly test.

There have been a number of approaches to leak detection m pipeline systems. Some leak detection systems are designed, to operate while product is being moved through the line: others require that the flow of product be stopped for the duration of a test. Leak detection systems generally use one of three methods: they measure the drop in pressure in the pipeline over a period of time they measure the difference in pressure or flow rate at two or more points along the pipeline, or they measure the change in the volume of the ,product over a period of time. Detecting small leaks is difficult because there are many physical phenomena present in pressurized pipeline systems that produce pressure, volume, and flow-rate fluctuations that are as large as or larger than those produced by a leak. These normally occurring to fluctuations degrade the performance of the leak detection system and result in false alarms or missed detections. As a consequence, a number of compensation schemes have been proposed to reduce them.

1.2.1 Pipeline Leak Detectors That Measure Pressure

The most common approach to the detection of leaks in a pressurized underground pipeline containing an incompressible fluid at rest is to relate the pressure drop in the line to the flow rate of the leak. A leak in the line is declared if the pressure drops by a specified amount over a given period. If this specified amount, or threshold, is not exceeded, the line is declared tight. Pressure tests are very difficult to interpret because the pressure drops are coupled with the properties of the pipeline itself. Thus, a similar pressure drop in two different pipeline systems should not necessarily be interpreted in the same way. Experimental measurements with controlled leaks indicate that (1) the pressure decreases exponentially with time as product is released from a line. (2) the volume released from a line decreases linearly with pressure when no vapor is trapped in the line, and (3) the leak rate decreases exponentially with pressure. The relationship between pressure and (1) volume. (2) leak rate, and (3) time are controlled by the elasticity of the pipeline system. The properties of the line are usually measured in terms of the bulk modulus, which is the inverse of the elasticity constant. As the elasticity of the line increases, the time required for the pressure to decay from the operating pressure of the line to zero (or to any other pressure below the operating pressure) increases. In one line it might take 15 rain for the pressure to drop 10 psi when there is a leak of 0.1 gal/h (defined at the operating pressure of the line), while in another line it might take 60 min. If the length of the test is defined as 15 min, the test protocol will prevent the sensor from detecting a 0.1-gal/h leak in some of the lines that are tested.

Some of the pressure changes that occur in pressurized pipelines are not associated with a leak. The most important are those associated with the thermal expansion or contraction of the liquid, the trapped vapor, and the pipe material itself. Experimental measurements in underground pipeline systems containing petroleum indicate that the pressure changes are directly proportional to the temperature changes and the bulk modulus of the pipeline system. These temperature-induced pressure changes occur frequently in both leaking and nonleaking pipelines. When the pressure changes in a leaking pipeline are no greater than these normally occurring temperature-induced changes, it is difficult to detect a leak by monitoring the line for drops in pressure.

Accurate detection of a leak demands (1) that both the instrumentation and protocol have sufficient sensitivity to detect the smallest leaks of interest. (2) that the temperature changes in the line be measured and compensated for and (3) that the pressure changes be related to the flow rate of the leak. All three require that the range of the elasticity properties of the pipelines that will be tested be known. The second requires that the temperature of the product be measured. The third requires that the pressure-volume relationship be measured each time for each line being tested.

1.2.1.1 Bulk Modulus

The bulk modulus of a pipeline is defined by the relationship between pressure and volume within that line. The bulk modulus of both the line and the product must be known before one can convert the pressure and temperature changes to volume changes or before one can interpret the meaning of a pressure drop. One can estimate the bulk modulus by simultaneously measuring the pressure of the line and the volume of product released through a valve in the line. Errors in determining this relationship occur if the line is leaking, if the temperature of the product in the line is changing, or if vapor or air is trapped in the line. Accurate calibration is difficult because the integrity of the line is unknown, as are the temperature of the product in the line and the volume of trapped vapor. Furthermore, the bulk modulus of the pipeline system changes over time as the volume of trapped vapor and air changes, and as the elasticity of the flexible hosing, the mechanical leak detector, and the pipe material changes.

1.2.1.2 Thermally Induced Pressure Changes

Thermally induced fluctuations in pressure are the major source of error in detecting a liquid leak with a pressure detection system. The magnitude of the error depends on the magnitude of the coefficient of thermal expansion and the bulk modulus of the liquid and the line material For gasoline motor fuels, whose coefficient of thermal expansion is 6 to 7 times larger than that of water, even small temperature changes have been shown to produce large pressure changes (e.g., a 0.1.degree. C. fluctuation in temperature can cause the pressure to change by 10 psi). Furthermore, both theoretical and experimental analysis demonstrate that the rate of change of temperature in an underground pipeline system can be high and complicated. From the standpoint of petroleum-dispensing operations, it is difficult to distinguish temperature-induced pressure changes from those that are leak-induced, because the rate of change of pressure vanes exponentially both with the volume of product released through a hole in the line and with the change in the temperature of the product. The temperature of the product vanes exponentially when product from the tank is brought into the line, because the temperature of this product differs from the temperature of the backfill and soil around the pipeline. This temperature difference, which can be many degrees, results in an exponential change as the product in the line attempts to come into equilibrium with its surroundings. In lines that are 100 to 200 ft long and 2 in. in diameter, it may be 6 to 12 h before the rate of change of temperature is low enough to permit accurate testing.

The traditional methods of compensating for temperature effects, which require the measurement of the rate of change of temperature of the liquid and the pipeline, are impractical because (1) the temperature distribution of the product in the line is spatially inhomogeneous, and a large number of temperature sensors would have to be retrofitted along the line in order to measure it: and (2) installing, maintaining, and calibrating a large number of sensors would be difficult. The best method of compensating for the effects of temperature fluctuations is to wait until these fluctuations are small enough to be negligible. For accurate pressure tests, this waiting period should be between 6 and 12 h.

1.2.1.3 Summary

Detecting small leaks in a pressurized pipeline by monitoring the pressure changes in the line is very difficult. High performance requires (1) that the test be long enough to allow the pressure to drop by a specified amount, suitable for detecting the smallest leaks of interest over the full range of pipeline systems to be tested, and (2) that the waiting period between the last dispensing of product and the beginning of the test be long enough for the temperature changes in the line to become negligibly small. To obtain accurate results in the case of the 2-in.-diameter lines found at a typical retail service station, dispensing operations might have to be terminated up to 12 h before beginning the test. Thus, the total time required to conduct a test becomes quite long.

1.2.2 Pipeline Leak Detectors That Attempt to Compensate for Thermal Changes

In U.S. Pat. No. 4,608,857, Mertens describes a method for detecting leaks as small as 1 L/h in a pressurized pipeline without waiting for fluctuations in the temperature of the product to subside. (As we have seen, such fluctuations induce pressure changes that can be mistaken for a leak. ) Mertens establishes three measurement periods of equal length. Initial line pressure is the same during the first and third periods but is lower during the middle period. Pressure changes are measured during all three periods. The middle measurement is then subtracted from the average of the finest and third. The difference is compared to a threshold, and in this way the existence of a leak is determined. Mertens indicates that the volume of product in the line must be small for the method to work properly. Furthermore, according to Mertens, the method accurately compensates for temperature providing that "the sum of the consecutive measurement periods is very small compared to the half value period of a temperature equalization process."

Analysis of this method shows that, when a leak is present in the line, the average pressure change that occurs during either the first or third periods will always be greater than that during the middle period. Furthermore, depending on the bulk modulus of the pipeline system, the actual volume change that occurs during these measurement periods will vary from one leaking line to another, even when these lines have the same initial starting pressures and identical leaks. Mertens's method does not require that the bulk modulus be measured and does not attempt to interpret the test results in terms of the actual leak rate. Mertens's method declares a leak in the pipeline if the difference between the high- and low-pressure measurements exceeds a predetermined threshold value. However, a wide range of volume changes could produce this same pressure change, and therefore, the accuracy of his method will vary from line to line.

1.2.3 Pipeline Leak Detectors That Attempt to Detect Leaks While There Is Flow in the Line

The method described by Mullen in U.S. Pat. No. 3,702,074 detects leaks in pressurized pipelines while product is flowing through the line. Mullen measures flow rate at two different points along the line (either the inlet and the outlet or any other two points sufficiently distant from one another 1 and at two different pressures, one high and one low. The difference in flow rate between the two measurements made at the lower pressure sure is subtracted from the difference between the same measurements made at the higher pressure. The result is then compared to a threshold leak rate, which, if exceeded, is the basis for declaring a leak in the pipeline. Mullen contends that because his measurements are closely spaced in time, he prevents long-term dynamic trends, such as those produced by the thermal expansion and contraction of the product, from affecting the results. However, while the temperature changes, the rate of change remains the same. For example, if measurements are made one minute apart the temperature change is much less than if they are made one hour apart; however, the rate of change is the same over any interval, whether it is a minute or an hour. Mullen's approach does not work because it confuses the rate of change with the actual change, which has no bearing on the results. Mullen's method will effectively compensate for temperature changes only if they happen to be the same during the high- and low-pressure measurements. This is unlikely to be the case, however, because, as stated above, the change in temperature in a pipeline is generally not constant (i.e., it tends to be exponential with time ). Furthermore, the fact that Mullen does not account for inventory changes also affects the accuracy of his method. Mullen minimizes short-term transient effects, such as those due to pressure, by taking several readings at each pressure and averaging them. By isolating different sections of line and by repeating the test at each segment of the line, he can locate the leak. He eliminates false alarms due to faulty equipment by comparing the test results for each segment of pipe tested: if the equipment is faulty., the flow-rate threshold will be exceeded in all of the segments tested.

It is an object of this invention to provide a method, and a device, for the reliable detection of small leaks in pressurized pipelines containing liquids, including water, petroleum, solvents, and other chemical products. Another object of this invention is to provide a method of and a device for quantitatively estimating the volume change and the flow rate of a leak in a pipeline at any pressure in the line. Yet another object of this invention is to provide a method of and a device for compensating for the thermal expansion and contraction of the product in the pipeline and of the pipeline itself.

A further object of this invention is to provide a method of and a device for quantitatively estimating the thermally reduced volume change and flow rate of the product in a pipeline at any pressure in the line.

The invention is designed to detect small leaks in pipelines that contain any type of incompressible liquid and that are either pressurized or can be placed under pressure for the duration of a test. The invention is particularly useful in underground or underwater pipeline systems, but can also be used on pipeline systems located above ground, such as those found in buildings or placed in specialized containment systems. The major application of this invention is for the detection and quantification of the flow rate produced by a leak in UST pipeline systems containing petroleum and other chemical products.

The invention requires that a leak detection test of the line be performed when the fluid in the line is at rest. Because the device will compensate for thermally induced changes in the pressure or volume of the product during a test, it is particularly useful for liquids that have a high coefficient of thermal expansion compared to water.

All references to the pressure of the pipeline system or pressure vessel made in this specification refer to gauge pressure. When the gauge pressure of the pipeline system is zero, the absolute pressure of the pipeline system is equal to atmospheric pressure. The claims made in this patent are based on absolute pressure. Thus, when the pressure is atmospheric, the gauge pressure of the pipeline system is zero.

Briefly, the preferred embodiment of the present invention is a simplified apparatus for detection of leaks in pressurized pipelines which utilizes a large pressure vessel and a small measurement vessel. The measurement vessel magnifies level changes during leak detection tests. The apparatus is connected to a pipeline through the measurement vessel. The entire system can be filled with liquid from the pipeline by opening a valve between the measurement and pressure vessels. Leak detection tests are conducted by measuring changes in volume with the measurement vessel over time while the pressure over the liquid in the pressure vessel and measurement vessel is maintained approximately constant. During tests, liquid communication between the measurement vessel and pressure vessel is prevented by closing the valve between them, but vapor communication between the vessels is permitted.

FIG. 1 illustrates the preferred embodiment of the present invention, in which an acoustic sensor system located inside a pressure vessel measures product-level changes and in which the vapor space above the product is used to maintain constant pressure during a leak detection test;

FIGS. 2(a) and 2(b) show side and top views of the preferred embodiment of the acoustic sensor system;

FIGS. 3(a) and 3(b) show side and top views of an alternative embodiment of the acoustic sensor system;

FIGS. 4(a) and 4(b) show four possible shapes for the cross section of the bar-shaped fiducial, the preferred triangular shape and three alternatives (rectangular, half-circular and circular);

FIGS. 5(a) and 5(b) show side and top views of a second alternative embodiment of the acoustic sensor system:

FIG. 6 shows an alternative embodiment of the pipeline leak detection system in which the acoustic sensor system housed inside the pressure vessel shown in FIG. 1 has been replaced by an electromagnetic sensor system attached to a float that rests on the product surface.

FIGS. 7(a) and 7(b) show side and top views an alternative placement of the electromagnetic sensor, in this case outside and along the neck of the pressure vessel;

FIG. 8 shows a second alternative embodiment of the pipeline leak detection system in which a pressure regulator and a container of inert gas are used to keep the pressure constant in the vessel that contains the acoustic sensor system;

FIG. 9 shows a third alternative embodiment of the pipeline leak detect system in which the acoustic sensor system shown in FIG. 8 has been replaced by an electromagnetic sensor system;

FIG. 10 shows a fourth alternative embodiment of the pipeline leak detection system in which the pressure vessel and pressure regulator FIGS. 8 and 9 have been replaced by a pressure sensor and a positive displacement pump that is used to pump liquid into or out of the pipeline me as a means of keeping the pressure constant:

FIG. 11 shows a fifth alternative embodiment of the pipeline leak detection system in which the positive displacement pump in FIG. 10 has been replaced by a piston that is used to displace a volume of liquid in the pipeline as a means of keeping the pressure constant.

FIG. 12 shows a simplified version of the preferred embodiment of the present invention;

FIG. 13 shows an alternative embodiment of the simplified apparatus shown in FIG. 12;

FIG. 14 illustrates a typical leak detection test sequence, with four 5-minute level measurement segments separated by 2-minute intervals;

FIG. 15 illustrates the simplified apparatus implemented with an electronic sensor for measuring level changes; and

FIG. 16 illustrates the simplified apparatus implemented with capability for both manual and electronic level measurements.

In order to measure the volume change that is due to a leak (i,e., the flow rate ), it is necessary to compensate for the temperature-induced volume changes. The present invention compensates for the thermal expansion or contraction of the product in the line without having to measure the temperature of that product. The time it takes to test a line is less than an hour. Unlike most of the methods currently in operation, this new technology is not based on measuring pressure changes in the pipeline system, Instead, it calls for a measurement of the change in the volume of product in the pipeline system. At least two consecutive measurements are made, one at the operating pressure of the line, and the other at zero pressure; for accurate temperature compensation, the pressure must be constant or nearly constant during the measurement. The invention compensates for temperature changes by differencing the volume changes noted during each of these measurements. A high degree of temperature compensation is achieved if the thermally induced volume changes are nearly the same during measurements at each pressure. Since this may or may not be the case, and since there is no way to verify it, a third measurement is made at the same pressure as the first measurement it is then averaged with the first measurement before the volume changes obtained at zero pressure are subtracted.

The methodology used to measure the temperature-compensated volume rate due to a leak takes advantage of the fact that the flow rate (volume change) due to a leak is not linear with pressure, but the flow rate (volume change) due to temperature fluctuation is. The preferred approach is to make one volume measurement when the line pressure is near zero and a second measurement at a higher pressure, preferably in the vicinity of the operating pressure of the line. At zero pressure, the flow rate due to a leak is zero; thus, the only volume change that occurs is due to thermal expansion and contraction of the product, vapor, or pipeline. The difference between the zero-pressure and the nonzero-pressure measurements represents the thermally compensated flow rate due to a leak at the nonzero pressure.

In general, to determine whether a pipeline system is leaking, the mass flow rate should be estimated from the change in mass of the liquid product in the pipeline system measured over the duration of the leak detection test. For detection of leaks in underground storage tanks and pipelines, it is the industry practice to measure and report the volumetric flow rate estimated from the change in volume of the product in the tank or pipeline system over the duration of the test. For the accuracy required for tests on underground storage tank pipeline systems, the mass flow rate and the volumetric flow rate can be assumed to be approximately equal. The difference between the mass flow rate and the volume flow rate is small, because the liquid product is incompressible at the pressures that an underground storage tank pipeline system is operated and the temperature of the product during a leak detection test does not change sufficiently to change the density of the product. The volumetric flow rate can also be accurately estimated from measurements of the change in the level of the liquid product in a pressure vessel, which is attached to and in communication with the pipeline system and contains both liquid product and trapped gas, during a leak detection test, because level changes can be easily converted to volume changes using a calibration factor. This specification measures and reports volumetric flow rate. There is a wide range of devices that can be used to implement the temperature compensation approach described above. Each device requires a sensing system to measure the change in volume of the product in the line. These devices can use any type of mass, volumetric, level, or density sensing system to measure and report volumetric flow rate. The sensing systems described in this specification measure either volume or level, but mass or density measurement systems could be used interchangably.

In the preferred embodiment of the present invention, a test is conducted at the operating pressure of the line and at a pressure near zero. The basic measurement scheme is to divide the test into three segments of equal length, and to make measurements at one pressure during the first and third segments and measurements at the other pressure during the middle segment. The operating-pressure measurement can be made during the first and third segments and the zero-pressure measurement during the middle segment, or vice versa. The averaging of the two operating-pressure measurements, which bracket the lower or zero-pressure measurement, minimizes any nonlinear changes in the temperature field during the total test period. It is acceptable to use more than three test segments providing the three-segment data collection and data analysis procedures are followed; doing so actually improves the accuracy of the system, and for this reason there is no upper bound on the number of tests. Mathematically, there are a number of equivalent ways to process the multiple-segment data. This three-segment approach assumes that the product temperature changes determined from averaging the temperature changes during the first and third segments is approximately equal to the temperature changes that occur during the middle segment. This assumption is valid for underground pipelines because the temperature changes in the line tend to increase or decrease exponentially over time when there is a temperature difference between the product in the pipeline and the surrounding ground.

The accuracy of a single test of the line will depend upon (1) the precision of the instrumentation used to measure the volume or volume-related changes in the line, (2) how constant the pressure can be kept during the measurements. (3) the duration of each measurement, (4) the number of data samples used to compute the volume change at each pressure, and (5) the temperature changes that occur in the liquid over the duration of the test.

To improve performance, a multiple-test strategy is used. This minimizes false alarms and missed detections. Three tests are conducted, although the wailing period described below is applied only to the first test. A temperature-compensated volume change is estimated from each three-segment test, or from an average of two or more three-segment tests. Providing that no product has been dispensed between the first and last test sequences, the rate of change of temperature should be decreasing over time, and the volume rate measurement should approach a constant value.

Accuracy can be somewhat degraded if the test is conducted immediately after new product has been brought into the pipeline and if the temperature of this new product is significantly different from that of the surrounding ground. The initial exponential change in temperature that occurs immediately after product is brought into the line is highly nonlinear. Even though a test conducted during this period can still meet the EPA release detection standards, a short waiting period (approximately 15 min) can minimize this nonlinearity and improve performance dramatically. For UST pipeline systems, the waiting period starts immediately after dispensing has ceased.

There are four approaches that can be used to implement this method. The approach taken will depend on the size of the pipeline, the maximum allowable size of the detector, the accuracy of the test, and the cost tradeoffs. These approaches are:

(a) Level sensor and reservoir (Passive Method). A reservoir, in this case a closed pressure vessel, is filled with fluid from the pipeline until the pressure in the vapor space of this container is equal to the line pressure of interest. A sensor is then used to measure changes in the level of the liquid in the vessel. The vessel is designed so that the level changes, and therefore the pressure changes, remain small during the measurement. Measuring the level changes in the vessel requires a high-precision sensor.

(b) Level sensor and reservoir (Active Method). As in the Passive Method, a closed container is partially filled with fluid from the pipeline. The remaining space is filled with a gas and maintained at a constant pressure equal to the line pressure of interest. Again, a

sensor is used to measure the changes in the level of the liquid in the container.

(c) Piston-displacement device. An object of known volume is inserted into or removed from the liquid in the pipeline to maintain a constant pressure in the line.

(d) Pump and reservoir. A small, two-way pump is used to move fluid back and forth between the line and a reservoir or container to maintain a constant pressure in the line. The volume changes are measured directly by the pump.

The first two devices measure level changes and convert these to volume changes. If there is no vapor in the line, one can calculate these changes from the geometry of the container; otherwise, one can generate a calibration curve by draining the container and measuring the volume of the liquid taken out of the container. The size of the container used to add or remove liquid from the line should be proportional to the size of the line, the amount of thermally induced volume change, the elasticity properties of the pipeline system, the volume of vapor in the line, and the size of the leak (although the leak is generally responsible for only a fraction of the volume changes contributed by all the other factors listed here ). Conversion from level to volume changes is done most easily if the cross section of the container does not change with level. A vertical cylinder is an example of such a container. The reason for keeping the pressure constant during a test is that the pressure changes in the vapor space are small when the level changes are small. The pressure changes in the container can be calculated from the perfect gas law. The vapor acts as a highly elastic spring. Any sensor that can measure liquid level independently of pressure with sufficient precision and accuracy to detect the smallest leak rates of interest will suffice (for example, an acoustic, optical, electromagnetic, or capacitance sensor) . For reliable detection of leaks as small as 0.05 gal/h, these level sensors need to have a precision of approximately 0.002 in. or better.

An automatic pipeline leak detection system (PLDS) is illustrated in FIG. 1 as it would be used in an underground storage tank 220 in accordance with the preferred embodiment of the present invention. The PLDS has three main components: the probe assembly 122, a transducer controller 112, and a system controller 100. The transducer controller 112, which is mounted adjacent to the probe assembly 122 within an explosion-proof housing, controls the acoustic transducer 30. The system controller 100 is mounted to an above-ground support and is in electrical communication with the transducer controller 112 through a cable 90. The cable 90 carries power and command data from the system controller 100 to the transducer controller 112, and acoustic data from the transducer controller 112 back to the system controller 100.

The transducer controller 112 contains the pulse waveform shaping, transmitting and receiving, and digital preprocessing electronics for the PLDS system. The system controller 100 contains the remainder of the hardware and software necessary to control the desired operational modes from the transducer controller 112, acquire the acoustic data, process the data in terms of product level, product-level changes, and leak rate, and display the results. The system controller 100 can also be equipped to control other sensor systems, such as those that provide overfill protection and alert, an automatic tank gauging system, detection of leaks in the annular space of a double-wall tank, detection of petroleum floating on the groundwater outside the tank, and detection of vapors in the soil and backfill outside the tank.

The transducer 30 is in electrical communication with the transducer controller 112 by means of a conductor 92. With reference to FIG. 2(a), the transducer 30 receives command data from the transducer controller 112 and transmits a series of accurately timed acoustic pulses up the probe, through the product, and to the various fiducials (acoustically reflective targets). Fiducials 24 and 22 comprise the bottom circumference of two concentric thin-walled nylon tubes (the "sleeve") separated in the vertical by a known distance; the nylon sleeve fits into a cylindrical robe, preferably a 1.5-in.-diameter plastic tube that holds the probe assembly. The lower fiducial 24, F.sub.1, is preferably positioned at a height, h.sub.1, about 2 in. above the transducer 30, while the upper fiducial 22, F.sub.2, is preferably positioned at a height, h.sub.2, about 4 in. above the transducer. In operation, acoustic pulses emitted by the transducer 30 are reflected from the fiducials 24 and 22 and from the interface between the product and the vapor, whether the product level is high 40 or low 50.

Referring to FIGS. 1 and 2(a), the probe assembly 122 consists of a cylindrical pressurized vessel 10, the acoustic sensor 31 (which includes the transducer 30 mounted on a base 32, a robe 20 with a hole 26 located near the bottom of the robe, and two fiducials 22 and 24 separated by a known distance and mounted on a sleeve comprised of two concentric nylon tubes 28 and 34), and a series of valves, pipes, and cables linking it to the tank, pipeline and controllers. A valve 62 connects a pipe 60 from the pressure vessel to a pipe 70 attached to the pipeline 200; this valve 62 is the means by which product from the pipeline enters the pressure vessel. Another valve 64 connects the pipe 60 from the pressure vessel to a pipe 80 that drains into the tank 220 or another holding container; this valve 64 is the means by which product is removed from the pressure vessel during a test. The trapped vapor 33 in the pressure vessel is used to maintain a constant or nearly constant pressure during the measurements at each product level in the pressure vessel. An electric cable 92 connects the transducer to its controller unit 112. Because the fluid in the pipeline must be at rest during a leak detection test, the pipeline 200 has a flow switch 120 to monitor whether product from the tank enters the line during a test. It also has a high-performance check valve so that pressure in the line can be maintained during a test. Valves 66 and 68 are used in the calibration of the sensor.

The acoustic sensor 31 (FIG. 2 ), which measures level changes within the pressure vessel, is housed in a tube 20 that supports both the transducer 30 and the two reference fiducials 22 and 24. The transducer is located at the bottom of the tube and the fiducials are mounted at a known distance from the transducer on a sleeve (comprised of nylon tubes 28 and 34 ) that is inserted in the tube above the transducer. A hole 26 near the bottom of the tube allows product from the pressure vessel to enter or leave.

FIGS. 3(a) and 3(b) show another alternative embodiment of the acoustic sensor 31. The fiducials 322 and 324, which are affixed to the tube, are thin bars positioned such that their long axes are perpendicular to the transducer. Four of many acceptable cross-sectional shapes 330, 332,334, and 336 for the fiducials 322 and 324 are shown in FIG. 4(a). The triangular bar 330 has the preferred cross-section, because (1) the bottom edge of the bar is flat and perpendicular to the transducer so that the acoustic energy reflected from the fiducial is maximized, and (2) the top edges of the bar are not perpendicular to either the transducer or surface, so that the acoustic energy reflected from the top of the bar is minimized.

In yet another alternate configuration of the acoustic sensor system (FIG. 5), the transducer and fiducials are not housed in a tube: the transducer 30 is mounted on the bottom of the pressure vessel 10 and the fiducials 342 and 344 are mounted on a rod 340 suspended vertically from the top of the pressure vessel. Some acceptable cross-sectional shapes for the fiducials 342 and are shown in FIG. 4(al ).

It is convenient but not absolutely necessary for the pressurized vessel 10 to have a cross-sectional area that does not change with height. A cylinder is preferred, because the height-to-volume conversion factor is then the same regardless of the level of product within the vessel. If the cross-sectional area changes from top to bottom, such as in a spherical vessel, the height-to-volume conversion factors is a function of the level of the product, and a table of conversions is required.

Referring to FIG. 2(a), three measurements are made with the acoustic sensor subsystem 31. First, an estimate of the speed of sound through the product between the transducer and the liquid surface is made; when the speed of sound is known, an acoustic pulse can be used to measure the height of the product in the pressure vessel and to measure the rate of change of level. The pulse travels from the transducer to the fiducial closest to the product surface. The lower fiducial 24 is used to measure sound speed when the product surface is at the lower level 50, and the upper fiducial 22 is used when the product surface is at the higher level 40. The speed of sound with one fiducial can be measured by

where

U.sub.1 =the speed of sound in inches/second between the transducer and either the upper fiducial 22 or the lower fiducial 24

h.sub.1 =the known distance in inches between the transducer and the upper fiducial 22 or the lower fiducial 24

t.sub.1 =the round-trip travel time in seconds between the transducer and the upper fiducial 22 or the lower fiducial 24

i=either 1, which represents measurements made between the transducer and the lower fiducial 24, or 2, which represents measurements made between the transducer and the upper fiducial 22

If the product surface is above the upper fiducial 22, both fiducials can be used, with the following algorithm, to estimate the speed of sound U.sub.1-2. ##EQU1##

Second, an estimate of the surface level of the product in the pressure vessel is made; this ensures that the product is at the correct level, either the higher level 40 for measurements made at the operating pressure of the pipeline or the lower level 50 for measurements made at zero pressure. The liquid level in the pressure vessel changes because of the contraction or expansion of the vapor in the pressure vessel as the pressure in the pipeline system increases or decreases, respectively. This estimate is repeated for each segment of a leak detection test. The height of the surface above the transducer in inches, h, is then calculated from

where

h.sub.s =the measured distance, in inches, between the transducer and the product surface 40 or 50

t.sub.s =the round-trip travel time in seconds between the transducer and the product surface 40 or 50

U.sub.s =the speed of sound in inches/second between the transducer and the product surface; U, is estimated from either U.sub.1 (the speed of sound in inches/second between the transducer and the fiducial 22 or 24 that is closest to the product surface) or U.sub.1-2 (the speed of sound in inches/second between fiducials 22 and 24)

The speed of sound through the product varies as the density of the product changes. For a product of uniform chemical composition, the change in density is dependent on the change in the temperature of the product. As a consequence, the speed of sound through a given product can be accurately determined from the average temperature of the product over the propagation path of the acoustic signal. For the liquids of interest, changes in the speed of sound vary linearly over the range of ambient temperatures that will be encountered during underground pipeline tests and can be determined from

where

U=speed of sound speed in meters/second

T=temperature in degrees Centigrade

m=dU/dT in meters/second/degrees Centigrade

b=sound speed in meters/second at T=0 degrees Centigrade Third, the change in the level of the product is determined from ##EQU2## where U.sub.s =U.sub.1 =speed of sound in meters/second between the transducer and the surface; the fiducial closest to the product surface (either fiducial 22 or 24) is used in estimating the speed of sound

.delta.t.sub.s =the change over time in the round-trip travel time in seconds between the transducer and the surface

.delta.t.sub.i =the change over time in the round-trip travel time in seconds between the transducer and the fiducial closest to the product surface (either fiducial 22 or 24)

The first term in the square brackets in Eq. (5), .delta.t.sub.s, is a measurement of the product-level changes, and the second term, .delta.t.sub.1, is used to correct the level changes for errors due to sound speed changes. The product in the pressure vessel is subject to thermal expansion and contraction. In general, however, no correction is made for this phenomenon because the error associated with it is usually smaller than the precision required of the sensor for measuring level changes. If the pressure vessel were large or if the temperature changes of the product in the vessel were great, the height changes would be estimated with the following equation, which compensates for the thermal expansion and contraction of the product in the pressure vessel: ##EQU3## where V=volume, in cubic inches, of the product in the pressure vessel at a surface height of h

h=height, in inches, of the liquid surface in the pressure vessel above the transducer

A=cross-sectional area, in square inches, of the surface of the product in the pressure vessel at a height of h above the transducer

C.sub.e =coefficient of thermal expansion of the liquid in the pressure vessel

.DELTA.T=change in the average weighted temperature between the transducer and the fiducial that is located closest to the product surface during the measurement

An estimate of the average temperature change is made from ##EQU4## where t.sub.i is the round-trip travel time between the transducer and either fiducial 22 or 24. The third term in Eq. (6), involving .DELTA.T, is the one that compensates for the thermal expansion and contraction of the product in the pressure vessel.

An alternative yet similar equation that can be used to estimate the temperature-compensated level changes in the pressure vessel is ##EQU5## The only difference between Eqs. (6) and (8) is the term that is used to correct the level changes for sound speed. Once the speed of sound through the layer of product between the transducer and the fiducial 22 or 24 has been estimated, the quantity (t.sub.s /t.sub.1) in Eq. (8) is used to extrapolate that estimate to the layer of product between this fiducial 22 or 24 and the surface. Another method of estimating .delta.h.sub.s, is to use the speed of sound through the layer of product between the two fiducials 22 and 24 when the product is at the higher level 40 and above the higher fiducial to estimate the sound speed changes in the layer above this upper fiducial 22. This method uses ##EQU6## where U.sub.1-2 =speed of sound between fiducials 22 and 24.

The protocol for conducting a pipeline leak detection test with the preferred embodiment of the invention shown in FIGS. 1 and 2 is as follows:

1. During the installation of the leak detection system, it is determined what the height of the liquid in the pressure vessel 10 will be (1) when the pressure is zero and (2) when the pressure is at another, higher level that will be used during a test. This is done as follows. The first step is to establish the height of the product when the pressure is zero (i.e., atmospheric). All valves 62,64,66 and 68 are closed except for the one 62 that allows product to enter the pressure vessel from the pipeline 200 via connecting lines 70 and 60. Valve 62 is then closed and valve 66 is opened, allowing the vessel to come to atmospheric pressure. The valve 64 at the juncture of the connecting lines 60 and 80 is then opened, allowing product to drain from the pressure vessel into the tank or other appropriate holding container until the level of the product in the vessel falls to a point as close as possible to, but still above, fiducial 24. Valve 64 is closed. Next valve 62 is opened and a submersible pump 240 is turned on and allowed to pressurize the pipeline 200. The pressurized product from the pipeline flows into the pressure vessel via connecting line 70 and rises to the upper level 40 (pipeline pressure greater than zero ). If the pressure vessel and the fiducials have been properly designed, the level should rise above the upper fiducial 22 until it is approximately the same distance from this fiducial as it was from the lower fiducial 24 when the pressure was zero. Once the levels have been checked by means of Eq. (3), a calibration can be performed to establish the height-to-volume conversion factor for the system.

2. The height-to-volume conversion factor, which relates the level of the product in the vessel to a corresponding volume is then determined. When the pressure in the vessel 10 is zero, valve 68 is opened and a known quantity of product is removed from the container. The change in level resulting from this change in volume is measured with the acoustic transducer 30. The height-to-volume conversion factor is obtained by dividing the volume change by the level change.

3. A leak detection test is initiated from the system controller 100. The system controller instructs the submersible pump 240 via cable 110 to pressurize the pipeline 200. However, no product is dispensed from the line. The system controller then opens valve 62 via cable 102 to allow product from the pipeline to enter the pressure vessel until it reaches the upper level 40 and the pressure in the vapor space 33 is the same as that in the pipeline. The pump is then turned off via a command from the system controller 100 via cable 110. The height of the product in the pressure vessel is then checked. It should be above the upper fiducial 22 at the upper level 40.

3. The system controller 100 then instructs the transducer controller 112 to collect data on level changes over a specified period of time, nominally 5 min. The rate of change of the level is calculated by fining a least-squares line to the data. The slope of the line, when multiplied by the height-to-volume conversion factor, is the rate of change of volume at the higher pressure.

5. The system controller 100 then lets the pressure in both the pipeline 200 and the pressure vessel 10 drop to zero by opening valve 64 via cable 104. When zero pressure has been reached, another check on the height of the product is made. It should now be above fiducial 24 at the lower level 50. The system controller 100 then instructs the transducer controller 112 to collect data on level changes over a period of time identical to the one used in step 4 (for the high-level measurements). The rate of change of the level is calculated by fitting a least-squares line to the data. The slope of the line, when multiplied by the height-to-volume conversion factor, is the rate of change of volume at the zero pressure.

7. The system controller then closes valve 64 via cable 104, opens valve 62 via cable 102, and instructs the submersible pump 240, via cable 110, to pressurize the pipeline system. Again, the level of product in the pressure vessel rises until the pressure in the vessel is the same as that in the pipeline. The height of the product in the pressure vessel is again checked. Then the system controller 100, via cable 110, turns off the submersible pump and instructs the acoustic transducer controller l 12 to collect data on level changes over a specified period of time. The rate of change of the level is calculated by fitting a least-squares line to the data. The slope of the line, when multiplied by the height-to-volume conversion factor, is the rate of change of volume at the higher pressure (being measured now for the second time).

8. Via cable 108, the system controller 100 checks the flow switch 120 continuously throughout the test to determine whether any product is being dispensed. If there has been no flow, data analysis can proceed. If there has been flow, the test is terminated.

9. Thermal fluctuations in the rate of change of volume must be compensated for. The temperature-compensated volume rate is calculated as follows. The average of the two measurements of the rate of change of volume obtained when product is under pressure and at the upper level 40 is subtracted from the rate obtained when the pressure is zero and the product is at the lower level 50.

10. If the temperature-compensated volume rate exceeds a predetermined threshold, the pipeline may be leaking.

11. If in the first test the threshold is exceeded, two more tests identical to the one described above are conducted. It is determined on the basis of the last two tests whether the line should be declared leaking In this way the possibility of false alarms is reduced.

The test procedure is identical for the alternative embodiments, shown in FIGS. 3 through 5, that use an acoustic sensor, and for the alternative embodiments shown in FIGS. 6 and 7 in which the level changes are measured with a linear variable-differential transformer (LVDT) and LVDT controller 312, rather than with an acoustic system.

The primary function of the acoustic sensor 31 is to measure the level of the product in the pressure vessel and to detect and measure any changes in this level. Any level-measurement sensor system can be used providing it has sufficient accuracy, precision and resolution to meet the performance standard for measuring level and changes in level. The acoustic sensor 31 shown in FIGS. 1 through 5 has been replaced in FIG. 6 by a system 380 that uses an electromagnetic float 360 to track the surface. The float 360 is attached to a vertical rod 370 that in turn is attached to a cylinder 386 with a ferromagnetic core; the cylinder moves up and down inside a linear variable-differential transformer (LVDT) 382, a commercially available device. The LVDT measures the change in the magnetic field as the cylinder 386 moves up and down. Level changes of better than 0.001 in. can be measured. The length of the LVDT depends on the difference in the levels 40 and 50 to be measured. The LVDT can be completely contained within the pressure vessel as shown in FIG. 6, located on top of the pressure vessel as shown in FIG. 7, or placed in any position in between. The LVDT measures only level changes, so in order to measure absolute height, the changes in level must be continuously summed by the system controller 100. Although the shape of the float does not affect the performance of the measurement system, the vertical dimension of the pressure vessel can be minimized if the shape of the float matches the shape of the vessel. To minimize evaporation and condensation effects, the float should have a cross-sectional area nearly as wide as that of the pressure vessel.

The present invention quantitatively estimates the flow rate from a leak at the operating pressure of the pipeline; compensates for thermal expansion and contraction of both the product and the pipeline without the need for measuring temperature directly; can conduct a leak detection test in a short time (approximately 15 min); is self-calibrating, because it measures volume directly, or measures level changes, which can easily be converted to volume changes from the height-to-volume calibration measurements or from the cross-sectional area of the pressure vessel; only requires level or volume measurement sensors are required in order for the invention to measure volume changes, to change the pressure in the pipeline, and to keep the pressure constant during a test; and increases its performance of the invention increases when a multiple-test strategy is used.

Four alternative embodiments of the present invention are shown in FIGS. 8 through 11. Each of these embodiments measures the volume changes at zero pressure and at the operating pressure of the line, keeping the pressure constant during these measurements. The three-segment procedure is then used to compute the temperature-compensated volume rate. The devices shown in FIGS. 8 and 9 measure level changes in a pressure vessel, while those in FIGS. 10 and 11 measure volume changes directly. The devices shown in FIGS. 8 and 9 are nearly identical to the ones shown in FIGS. 1 and 6, except that a pressure-regulating subsystem 414 is used to raise and lower the pressure within the vapor space and to maintain a constant pressure within the vessel. The devices shown in FIGS. 1 and 6 maintain constant pressure without any additional devices, while the ones in FIGS. 8 and 9 include a device intended for this purpose. The pressure-regulating subsystem 414 consists of a pipe 420 connecting the gas in the pressure vessel to either a high- or low-pressure regulator. When the valve 410 is opened, an inert gas enters the pressure vessel 350. This gas is used to establish and maintain constant pressure at the higher level. If the pressure in the vessel drops below this level, a regulator 430 raises it; if the pressure rises, another regulator 432 lowers it appropriately. When the valve 412 is opened, the inert gas is allowed to escape; thus, the gas is also used to establish and maintain a pressure of zero in the vessel 350. If the pressure in the vessel drops below zero, a regulator 440 raises it back to zero; conversely, if the pressure rises above zero, a regulator 442 lowers it back to zero. The acoustic sensor subsystem in FIG. 8 requires only one fiducial 462, which is located as dose as possible to the underside of the surface of the product 300 in the pressure vessel. The device in FIG. 9 is identical to the one in FIG. 8 except that the acoustic transducer measurement system has been replaced by an electromagnetic float system like the one described in FIG. 6.

The test procedure for the alternative embodiments shown in FIGS. 8 and 9 is very similar to that for the embodiments shown in FIGS. 1 and 6, except that the pressure in the pipeline 200 and the vapor space 33 in the pressure vessel is controlled with a pressure-regulating subsystem 414. The higher pressure is maintained constant during both measurements (Steps 4 and 7) by the high 430 and low 432 regulators, and the zero pressure (Step 6) is maintained by two additional high and low regulators 440 and 442.

FIGS. 10 and 11 show a measurement system that maintains a constant pressure by adding or removing a known volume of product from the pipeline, and that then sums the volume changes. In FIG. 10, the sensor 122 (shown in FIG. 1 ) is replaced by a positive-displacement pump 500, a motor 510, and a pressure sensor 520. Product is added to the pipeline 200 through a pipe 82 or removed from the pipeline through a dram line 80. The valve 501 is used to calibrate the positive-displacement pump 500 and motor 510. The pressure sensor 520, which is connected to the pipeline via a pipe 528, is used to maintain the proper pressure conditions during a test. The controller 100, which is connected electrically to the pressure sensor via a cable 522 and to the motor 510 and positive-displacement pump 500 via cables 524 and 526, establishes both the high pressure and the zero pressure and maintains pressure by reading the pressure sensor and turning the displacement pump's motor on or off. When the pump removes product from the pipeline 200, the pressure in the line drops. When the pump adds product to the pipeline, the pressure rises. The volume change in the line at the higher pressure or at zero pressure is measured directly by the pump 500.

The device in FIG. 11 is substantially the same as the one in FIG. 10, except that the motor 510 and positive-displacement pump 500 have been replaced by a displacement piston device 530 and a linear actuator 540. The linear actuator is connected to the system controller 100 via a cable 542 and to the displacement piston 530 via another cable 544. The motor and positive displacement pump have been replaced by a displacement device 530 that uses a piston 532 to displace a known volume of fluid in the containment volume 534. The volume changes are determined directly from the movement of the piston 532.

The alternative embodiment of the invention shown in FIG. 10 measures volume changes directly. The protocol for conducting a pipeline leak detection test with this embodiment is as follows:

1. The motor 510 and positive-displacement pump 500 in FIG. 10 are calibrated over the range of volume measurements expected during a test. This is done by withdrawing known amounts of liquid from the line 82 via valve 501 and measuring the resulting change in volume. An alternative location of the valve 501 is in the line 80 connecting the pump 500 to the container or tank 220. A calibration curve is then generated by fitting a least-squares line to the data that identify the measured volume and the actual volume withdrawn from the line.

2. A leak detection test is initiated from the system controller 100. The system controller instructs the submersible pump 240 via cable 110 to pressurize the pipeline 200. Once the test pressure is reached ("test pressure" meaning the one higher than zero), a command from the system controller turns off the submersible pump via cable 110.

3. The system controller 100 then instructs the positive displacement pump controller 512 112 to record data on the change in volume over a specified period of time, nominally 5 min. The pressure in the pipeline is measured via the pressure sensor 520, and the system controller maintains a constant pressure in the pipeline by providing instructions to the motor 510 and positive-displacement pump 500 to add or remove product to maintain this pressure at a constant level during the test. The rate of change of volume is calculated by fitting a least-squares line to the data. The slope of the line is the rate of change of volume at the higher pressure.

4. The positive-displacement pump 500 removes product via the drain line 80 until the pressure in both the pipeline 200 and the pressure vessel 10 drops to zero. This is confirmed by the pressure sensor.

5. The system controller 100 then instructs the positive displacement pump controller 512 to record data on the change in volume over a period of time identical to the one used in Step 3 (for the high-pressure measurements). The rate of change of volume is calculated by fitting a least-squares line to the data. The slope of the line is the rate of change of volume at the higher pressure.

6. The system controller 100 then pressurizes the pipeline 200. Via cable 110 it instructs the submersible pump 240 to add product to the pipeline until the desired pressure has been reached. The system controller then turns off the submersible pump and activates the motor 510 and the positive-displacement pump 500, which adds or removes product from the pipeline 200 over the specified period of time so that the pressure remains constant during this measurement, as verified by the sensor 520. The rate of change of level is calculated by fining a least-squares line to the data. The slope of the line is the rate of change of volume at the higher pressure (being measured now for the second time).

7. Via cable 108 the system controller 100 checks the flow switch 120 continuously throughout the test to determine whether any product is being dispensed. If there has been no flow, data analysis can proceed. If there has been flow, the test is terminated.

8. Thermal fluctuations in the rate of change of volume must be compensated for. The temperature-compensated volume rate is calculated as follows. The average of the two measurements of the rate of change of volume obtained when the line is pressurized is to subtracted from the rate obtained when the pressure is zero.

9. If the temperature-compensated volume rate exceeds a predetermined threshold, the pipeline may be leaking.

10. If, in the first test, the threshold is exceeded, two more tests identical to the one described above are conducted. It is determined on the basis of the last two tests whether the line should be declared leaking. In this way the possibility of false alarms is reduced. The alternative embodiment of the invention shown in FIG. 11 also measures volume changes directly. The protocol for conducting a pipeline leak detection test with this embodiment is similar to the embodiment shown in FIG. 10 and is as follows:

1. The displacement piston device 530 and linear actuator 540 in FIG. 11 are calibrated over the range of volume measurements expected during a test. This is done by withdrawing known amounts of liquid from the line 82 via valve 501 and measuring the resulting change in volume. A calibration curve is then generated by firing a least-squares line to the data that identify the measured volume and the actual volume withdrawn from the line.

2. A leak detection test is initiated from the system controller 100. The system controller instructs the submersible pump 240 via cable 110 to pressurize the pipeline 200. Once the test pressure is reached ("test pressure" meaning the one higher than zero), a command from the system controller turns off the submersible pump via cable 110.

3. The system controller 100 then instructs the displacement piston device and linear actuator controller 552 to record data on the change in volume over a specified period of time, nominally 5 min. The pressure in the pipeline is measured via the pressure sensor 520, and the system controller maintains a constant pressure in the pipeline by providing instructions to the displacement piston device 530 and linear actuator 540 to add or remove product to maintain this pressure at a constant level during the test. The rate of change of volume is calculated by fitting a least-squares line to the data. The slope of the line is the rate of change of volume at the higher pressure.

4. The displacement piston device 530 removes product from the pipeline via the line 82 until the pressure in both the pipeline 200 and the pressure vessel 10 drops to zero. This is confirmed by the pressure sensor.

5. The system controller 100 then instructs the displacement piston device and linear actuator controller 556 to record data on the change in volume over a period of time identical to the one used in Step 3 (for the high-pressure measurements). The rate of change of volume is calculated by fitting a least-squares line to the data. The slope of the line is the rate of change of volume at the higher pressure.

6. The system controller 100 then pressurizes the pipeline 200. Via cable 110 it instructs the submersible pump 240 to add product to the pipeline until the desired pressure has been reached. The system controller then turns off the submersible pump and activates the displacement piston device 530 and linear actuator 540, which adds or removes product from the pipeline 200 over the specified period of time so that the pressure remains constant during this measurement, as verified by the sensor 520. The rate of change of level is calculated by firing a least-squares line to the data. The slope of the line is the rate of change of volume at the higher pressure (being measured now for the second time.

The remaining steps necessary to complete a test are the same as steps 7 through 10 are identical to those described above for the motor 510 and displacement pump 500 in FIG. 10.

There are many alternative embodiments of the basic method of collecting and analyzing volumetric data for the conduct of a pipeline leak detection test. It is required that the volumetric data be collected at two or more different pressures and during two or more measurement periods (i.e. segments) that are contiguous or nearly contiguous and have the same or nearly the same duration. For a high level of performance, data should be collected at two pressures using three measurement segments in which the pressure separation is maximized. A two-pressure leak detection test with only two segments can be conducted, but in order to achieve a high level of temperature compensation, and therefore a high level of performance, the rate of change of temperature must be uniform or nearly uniform. The accuracy of the two-segment approach, which can be effective under some pipeline temperature conditions (especially if one observes a waiting period before conducting a test), is controlled by the ambient rate of change of the temperature of the product at the time of the test. If the rate of change of product temperature is not constant during a test, the results will be in error. When a third measurement segment is added, accurate temperature compensation can be achieved even when pipeline temperature conditions during a test are not uniform, provided that the rate of change of temperature during data collection is monotonic (i.e. that it only increases or only decreases during a test). Such temperature conditions are typically encountered in underground pipelines, especially those associated with storage tanks at petroleum fuel storage facilities; thus, the two-pressure, three-segment volumetric data collection and analysis method will have wide application.

The two-pressure, three-segment approach works well because the rate of change of temperature of the product in the pipeline does not change with pressure but the flow rate due to a leak does. The flow rate due to a leak, which is usually defined at a specific line pressure, can be determined directly by testing first at some operationally useful test pressure and then at atmospheric pressure (i.e. zero gauge pressure). The value chosen for the test pressure, which can be either below or above atmospheric pressure, is usually the same as the operating pressure of the pipeline, or it is some pressure specified in an environmental regulation or professional society testing standard. The flow rate due to a leak at a specific test pressure can be determined even if neither of the two pressures is atmospheric, provided the relationship between leak rate and pressure is known or can be estimated empirically. For flow under pressure through an orifice or hole, the relationship between leak rate and pressure is well known; it is proportional to the square root of pressure for liquids with viscosities near water and is linear for more viscous liquids.

Since it takes a finite time to change the nominal pressure in the line, the two (or three) measurement segments will not be absolutely contiguous. For a variety of reasons, the time interval between measurement segments may need to be longer than the time required to change the pressure. For example, the added time may be needed to allow and pressure transients and instabilities produced by the pressure change to subside before a volume measurement is made, which is particularly important when testing long or complicated lines. Additional time might also be required to insure equal intervals between measurement segments when the pressure changes are done manually. For best performance, the segments should be of equal duration, as should the time interval(s) between them. If they are not, it does not mean that a leak detection test can not be conducted, but it will produce an error in the final test result. The magnitude of the error is dependent primarily on the rate of change of temperature of the product in the line.

It is important to note that more than three measurement segments may be used. When the three-segment pattern is repeated, multiple estimates of temperature-compensated volume rate can be made and averaged, a technique that increases the accuracy of the final test result. It is also important to note that the nominal pressure can be different during each measurement segment, regardless of the number of segments. If this is the case, however, more stringent precision requirements of the sensor used to measure volume may be needed, and the magnitudes of any errors in the final test result will also be larger than a test conducted when the pressures of the first and third measurement segments are the same and equal to the highest (or lowest) pressure of a three-pressure test and the pressure of the second measurement segment is equal to the lowest (or highest) pressure of a three-pressure test. Whether this error is significant depends on what performance must be achieved by the leak detection system.

The two-pressure, three-segment volumetric measurement method described above can be modified either for other applications or to accommodate some important operational constraints that may be placed on a test. One especially useful application, in which pressure is maintained at the same level during any two consecutive measurement segments in a three-segment test, has operational advantages for several scenarios: (1) when pressure transients occur each time the pressure is changed, as is the case when testing large pipelines; (2) when the storage, handling, or disposal of product, which might occur when pressure is changed, is difficult; and (3) for simplification of the test protocol when a manual or semi-automatic implementation of the test methodology is used. The pressure sequence may be high-low-low, low-high-high, high-high-low, or low-low-high, which hereafter will be referred to as the `high-low-low` protocol. In the implementation previously described, the pressure is kept constant or nearly constant in the first and third measurement segments, but is changed during the second, or middle, segment. The pressure sequence can be either high-low-high or low-high-low. The same algorithm for computing the magnitude of the temperature-compensated flow rate for the "high-low-high" implementation is applied for the "high-low-low" implementation, i.e. averaging the volumetric flow rate estimated during the first and third measurement segments and subtracting the volumetric flow rate computed during the second segment. If one of the pressures is atmospheric, the absolute value of the magnitude of the temperature-compensated volume rate computed according to the "high-low-low" protocol is equal to one-half the flow rate due to a leak. The absolute value of the magnitude of the temperature-compensated volume rate computed according to the "high-low-low" protocol is equal to one-half the flow rate computed with the "high-low-high" protocol if both test sequences are conducted using the same pressures. If both pressures are non-zero (non-atmospheric), the temperature-compensated flow rate is equal to one-half the difference in the flow rates that are estimated at both pressures. If a leak is present, the flow rate is due to the leak and any errors in the test; if the line is not leaking, then any nonzero flow rate is due to these errors. Again, the pressures can be positive or negative or a combination of both.

A manual or a semi-automatic implementation of the two-pressure, three-segment test using the "high-low-low" protocol is attractive for underground pipelines associated with underground storage tank (UST) and aboveground storage tank (AST) facilities that dispense petroleum with a pump, where in order to change pressure in the line one must normally have control of the dispensing pump. For convenience, the leak testing apparatus can be attached to the line at a valve connection located at or near the fuel dispensing pump. When this pump is turned on, which can be done manually, the line is pressurized. For example, this is easily done using the pump handle on a dispenser at a retail motor fuel service stations. A high-low-low test can be initiated once the pump is turned off. The decreasing pressure levels required for the second and third segments can be attained by removing product from the line or pressure vessel or by releasing the vapor trapped in the pressure vessels of three of the four test apparatuses described above. If possible, sufficient product (or vapor) should be removed so that the pressure drops to atmospheric level (i.e zero gauge pressure). This can be done automatically or manually by an operator. Once this has been accomplished, the test can be completed without any further pressure changes. No further access to the fuel dispensing pump is required, a factor that has cost, safety, and operational benefits. This approach is particularly amenable to portable and/or battery operated implementations.

The difference in flow rate (also called the temperature-compensated volume rate, or TCVR) between the average pressure of the first and third segments and the second segment is obtained by averaging the rate of change in the volume of product as measured during the first and third segments and subtracting this average from the rate of change computed for the second segment. The TCVR obtained in this manner is

where P.sub.1, P.sub.2, and P.sub.3 are the line pressures that are maintained constant or nearly constant during each measurement segment and VR.sub.1 (P.sub.1), VR.sub.2 (P.sub.2), and VR.sub.3 (P.sub.3) are the volumetric flow rates determined for each measurement segment at pressures P.sub.1, P.sub.2, and P.sub.3, respectively. Assuming that the volume changes due to the thermal expansion or contraction of the product in the line are perfectly compensated for by the analysis algorithm, then Eq. (10) reduces to

where LR.sub.1 (P.sub.1), LR.sub.2 (P.sub.2), and LR.sub.3 (P.sub.3) are the volumetric leak rates determined for each measurement segment at the designated pressures. If P.sub.1, P.sub.2, or P.sub.3 is equal to atmospheric pressure, then LR.sub.1, LR.sub.2, or LR.sub.3, respectively, would be zero.

In practice, there will be some type of error due to residual compensation errors, as well as sensor noise and other sources of noise not compensated for by this technique. For best performance, the difference between the lowest and highest pressures should be as great as possible, and the precision (i.e. volumetric sensor noise) should be smaller than the error in estimating the TCVR that is determined in Eq. (10) and that is due to the leak rate of interest. In all of the implementations described above, whether they are based on two pressures and three segments or on three pressures and three segments, the flow rate can be derived from Eq. (10).

The error in compensating for the thermal expansion or contraction of the liquid product in the pipeline can be estimated as part of the test procedure if there is time to permit the addition of one or more measurement segments to the leak detection test. The approach is to use the same analysis algorithm, but to apply it to three segments in which the pressure is the same or nearly the same. Since the flow rate due to a leak would be the same in each segment, any non-zero estimate of flow rate would be due mainly to the error in compensation. This calculation would also quantify other errors (e.g. sensor errors). There are many ways to incorporate this error calculation into the test procedure. For example, if a low-high segment is added to the end of a high-low-high two-pressure, three-segment test, then the first, third, and fifth segments can be used to estimate the error in temperature compensation and the first three, middle three, last three, or an average of any of these three segments can be used as the basis for the leak detection test. If the error is substantial, especially in comparison to the estimated TCVR, the test should be considered invalid and should be repeated. Another example is to add a low-pressure segment at the end of a high-low-low sequence. The first three segments are then used as the basis for the leak detection test, and the last three to estimate the error in temperature compensation. The estimate is less meaningful this way, because the data used as the basis for the test have been collected over a somewhat different period in time than those used to estimate the temperature compensation error.

Another method for obtaining an error estimate uses the difference between two temperature-compensated volume change or flow rate tests. If two tests are conducted (whether or not the tests have overlapping measurement segments), the volume of flow rate due to a leak should be the same if the leak rate remains constant over the period of time in which the two tests are made. Therefore, any difference between the two test results will provide an estimate of the error in the test results. Comparisons of multiple test results will provide a more accurate error estimate, but will also take more time to obtain.

The technology described above can be implemented very simply if (1) the level changes that occur in the pressure vessel shown in FIGS. 1 and 6, which are due to volume changes in the pipeline, are magnified so that they can be made manually with a ruler or electronically with an inexpensive, low-precision level sensor; and (2) the need to transfer liquid in and out of the pressure vessel for the purpose of changing line pressure from one measurement segment to the next can be eliminated. Both goals can be accomplished with the embodiment shown in FIG. 12. This device is simple and inexpensive to make because there is no need for a high-precision level sensor or for electronic access to the transfer pump; it is also inherently safe to use in pipelines containing potentially combustible liquids such as petroleum products, because it does not use electrical power. Furthermore, these advantages can be realized without any sacrifice in the overall performance of the system in detecting pipeline leaks. The leak detection system can be permanently attached to a pipeline or it can be a portable system attached to a line only for the purpose of conducting a test.

The key functional features of the embodiment shown in FIG. 12 for conducting a leak detection test are as follows: (1) a large-diameter, cylindrical pressure vessel (pressure cylinder 610), (2) a small-diameter, cylindrical pressure vessel of similar height (measurement cylinder 612), (3) a sight glass with a ruler or tape marked in increments of approximately 1/16 inch for measuring the level of liquid in measurement cylinder 612, and (4) three manually operated valves V1, V2, V3. Measurement cylinder assembly 616 includes both measurement cylinder 612 and the level gauge (i.e. sight glass 614 with measurement tape). Pressure cylinder 610 and measurement cylinder 612, as well as connecting piping, are preferably constructed of steel; however, any material that can withstand the maximum expected pressure can be used. Sight glass 614 is preferably constructed of glass, which is clear to permit the liquid level to be read but sufficiently robust to withstand the pressures at which the system is used (typically 30-100 psi).

The purpose of pressure cylinder 610 in FIG. 12, which is also the main purpose of the pressure cylinder in the embodiments shown in FIGS. 1 and 6, is to maintain constant pressure in the pipeline during each of the measurement segments that comprise a leak detection test. The purpose of measurement cylinder 614 is to magnify those volume changes occurring in the pipeline that are due either to a leak or to thermal influences in such a way that small changes in volume are converted into large changes in level that can be measured easily through sight glass 614. For this purpose, sight glass 614 (with its measurement tape) is attached to measurement cylinder 612. If measurement cylinder 612 is 1-5/8 inches in diameter and sight glass 612 is 5/8 inch in diameter, then volume changes of 3 ml occurring in the pipeline can be measured by reading the level changes in sight glass 612 to 1/16 inch. A crucial difference between the embodiment shown in FIG. 12 and those shown in FIGS. 1 and 6 is that during a test the liquid in the pipeline communicates only with the liquid in measurement cylinder 612 and not with the liquid in pressure cylinder 610. The function of measurement cylinder 612 is not to measure level changes, but rather to magnify small changes in volume occurring in the pipeline. In the embodiment in FIG. 12, it is sight glass 614 and its measurement tape that are used to measure the level changes in measurement cylinder 612. It should be noted, however, that measurement cylinder 612 itself could be used for this purpose if it were made of a transparent material and a tape measure were attached to it.

Each of the three valves on the embodiment shown in FIG. 12 has a specific purpose. V1 must be open to pressurize the pipeline and leak detection system for a test and to conduct a test. When V1 is closed, the leak detection system is completely isolated from the pipeline. When open, V1 lets liquid from the pipeline into measurement cylinder 612 or from measurement cylinder 612 into the pipeline. Pressurization is accomplished by pumping liquid product into the measurement and pressure cylinders 612 and 610; V2 must also be open during the pressurization process. For best performance during the ensuing leak detection test, both cylinders should be nearly empty before the pressurization process begins; this results in the maximum volume of vapor being trapped at the test pressure, which means that the pressure can be better maintained at a constant level during each measurement segment. V2 is used to isolate level changes that occur in pressure cylinder 610 from changes in the level of liquid in measurement cylinder 612 and pipeline. This valve must be closed during each measurement segment of a leak detection test. As noted above, when measurement cylinder 612 is isolated from pressure cylinder 610 by closing V2, very small changes in the volume of liquid in the pipeline can be accurately measured at a constant pressure. V3 is used to lower the pressure in both the pipeline and the leak detection system (i.e. the measurement and pressure cylinders 612 and 610). When V3 is opened, vapor is released from both cylinders through vapor release tube 618; if all of the vapor is released, the pressure in the pipeline returns to atmospheric level (i.e. zero gauge pressure). For safe operation, the apparatus also includes pressure relief valve 620 to prevent inadvertent over-pressurization of the system and liquid check valve 622 to prevent the possibility of liquid overflowing during a test.

The embodiment shown in FIG. 12, as well as those in FIGS. 1 and 6, uses a pump to pressurize the pipeline. A slight modification to the embodiment in FIG. 12 will allow pressurization by means of bottled gas such as nitrogen, thus eliminating the need to use the transfer pump or another pump dedicated to this purpose. This modified embodiment is shown in FIG. 13. One advantage of pressurizing the line with bottled gas 624 such as nitrogen is that the operator can conduct the entire test without leaving the site of the leak detection system. Another advantage is that once the pressure and measurement cylinders 610, 612 have been filled with an amount of liquid sufficient for conducting a test, they do not have to be emptied and refilled again to repressurize the line for the conduct of a test. The bottled gas 624 can be permanently attached to the leak detection system or it can be a portable unit that is attached to the system any time a test is conducted. When V5 and V3 are open and V4 is closed, the pipeline can be pressurized. Once the line has been pressurized, V5 and V3 are closed and V4 is opened; from then on the steps for performing a leak detection test are identical for the embodiments shown in FIGS. 12 and 13.

Using the bottled gas pressurization technique, however, requires that there be sufficient product already present in the apparatus before a leak detection test is initiated. It should be noted that once the test pressure is reached, the gas bottle 624 is isolated from the pressure cylinder by means of a valve, and no further pressurization of the pipeline is required. In contrast, in the embodiments shown in FIGS. 8 and 9, gas is continuously added or removed as the means of maintaining constant pressure during each measurement segment. In the embodiment shown in FIG. 13, the volume of vapor in the pressure cylinder 610 (which is in direct communication with measurement cylinder 612) is large enough to maintain approximately constant pressure in the pipeline without any active adjustments in pressure.

The embodiments in FIGS. 12 and 13 can be designed for applicability to a wide range of pipelines of different sizes and diameters. One of the most important is the kind of large underground pipeline found at terminals for aboveground storage tanks (ASTs) that contain petroleum products. These lines are typically 4 to 10 inches in diameter and 100 to 500 feet in length. Both embodiments are also applicable to the smaller lines associated with underground storage tank (UST) facilities, for example, the lines found at retail gas stations. These are typically 2 inches in diameter and 50 to 200 feet in length. These embodiments can be used to test lines larger than those found at terminals. The main difference between the embodiments for use on smaller lines and those for use on larger lines is in the size of the pressure and measurement cylinders 610, 612. Both the diameter and the height of the two cylinders must be increased as the diameter and/or length of the pipeline increases. The correct size of the pressure and measurement cylinders 610, 612 for a given line can be calculated analytically using the perfect gas law.

The diameter and height of the two cylinders should be such that they can (1) accommodate the anticipated changes in the volume of product in the line, regardless of whether these changes are due to a leak or to thermal expansion and contraction of the product, and (2) limit, during any given measurement segment, the pressure changes due to such volume changes while (3) still maintaining an acceptable level of precision. The pressure cylinder 610 must be large enough that any pressure changes occurring within it (and resulting from volume changes in the line) are small in relation to it. Similarly, the measurement cylinder 612 and sight glass 614 must be large enough height-wise to handle the product volume changes that might occur during a measurement segment, yet small enough in diameter that level changes can be measured precisely. If the diameter of the measurement cylinder 612 is too large, there is a loss of precision in the ability to measure volume changes in the pipeline. If its diameter or length is too small, the apparatus may not be able to accommodate the volume changes that occur during a measurement segment (whether due to a leak or to thermal influences). The measurement cylinder 612 can be configured so that it is longer than the pressure cylinder 610, extending above and/or below it; this will permit the leak detection system to accommodate even larger leaks and larger thermally induced volume changes.

As indicated above, the size of the pressure and measurement cylinders 610, 612 will depend on the application. Several examples of typical cylinder geometries are given below. When used to test the larger-capacity pipelines associated with ASTs containing petroleum products, the pressure cylinder might be approximately 16 inches in diameter and 36 to 48 inches high, and the measurement cylinder 2 to 2-1/2 inches in diameter and at least the same height as the pressure cylinder. When used to test the smaller-diameter lines associated with most USTs containing petroleum products, the pressure cylinder would typically be 4 to 6 inches in diameter and 12 to 24 inches high, and the measurement cylinder 1 to 1-1/2 inches in diameter and approximately the same height as the pressure cylinder. Both of these typical cylinder geometries can be shown to accommodate a wide range of testing conditions at AST and UST facilities.

The general approach to testing a pipeline with either of the two embodiments of the device shown in FIGS. 12 and 13 is, as with the other embodiments shown in FIGS. 1 and 6, to divide a test into three or more distinct "measurement segments". The measurement segments are used in different combinations to obtain both the temperature-compensated volume rate between the average of the highest and the lowest pressures (test result) and an estimate of the error in the test result (test error). For example, when two of three measurement segments are conducted at two different pressures, an estimate can be made of the difference in temperature-compensated flow rate between the two pressures; when three contiguous or three alternating measurement segments are conducted at the same nominal pressure, an estimate can be made of the error in temperature compensation volume rate. This error estimate includes the error in temperature compensation, level readings, and uniformity of leak rate. The number of segments, the level of pressure during each segment, and the sequence of high- and low-pressure segments are all application-dependent.

For a manually conducted leak detection test, a high-low-low-low data collection protocol can be used, where the higher pressure is the test pressure and the lower one is atmospheric (i.e. zero-gauge pressure). The test pressure can be established by means of either a pump or bottled gas, and the lower pressure is obtained by simply opening V3 and allowing the vapor to be released. Provided that the pressure cylinder 610 has a large enough diameter and that it contains a large enough volume of vapor, pressure in the line will not change significantly during any of the four measurement segments unless there is a very large change in the volume of liquid in the line.

When the apparatus in FIG. 12 is to be used to conduct such a high-low-low-low test, V1 and V2 are opened, V3 is closed, and the transfer pump is turned on until both the pipeline and apparatus have been pressurized. Product will enter the pressure cylinder 610 until the pressure in the vapor space of that cylinder is equal to the pressure in the line. (As stated above, if pressure cylinder 610 already contains enough liquid for the conduct of a test, it can be pressurized with a bottled gas such as nitrogen using the apparatus in FIG. 13.) Once the liquid in sight glass 614 reaches a constant level, the apparatus is fully pressurized and the test can begin. At this point, the transfer pump (or the connection to the bottled gas) is turned off and V2 is closed. Any volume changes in the line during the measurement segment are reflected as level changes in the measurement cylinder assembly 616. V3 is then opened, releasing vapor, and the pressure is lowered to zero. Once the system has been depressurized, V3 is closed, and V2 is opened so that the liquid in the measurement cylinder 612 comes into equilibrium with that in the pressure cylinder 610. V2 is then closed in preparation for the next segment, conducted at atmospheric (zero gauge) pressure. The process is repeated twice more to complete a test.

FIG. 14 illustrates a typical test sequence. The data collection portion of this leak detection test takes 26 minutes. There are four 5-minute segments during which level measurements are made. Each 5-minute segment is separated from the next by a 2-minute interval that allows the test operator to depressurize the line by opening and closing the appropriate valves. This 2 minute interval also serves to let the level of liquid in the measurement cylinder stabilize after the line has been depressurized.

At the beginning and end of each level-measurement segment, the test operator reads and records the level of the liquid product in the sight glass 614 attached to the measurement cylinder 612. For high performance, this reading should be made to the nearest 1/16 inch or better; even if the manual readings are not this accurate, system performance may still be acceptable. This leak detection test requires eight measurements in all, one at the beginning and one at the end of each of the four measurement segments. The first six (made in this test sequence at 0, 5, 7, 12, 17, and 21 minutes after the start of a test and shown on the time line in FIG. 14) are used to compute the test result. The last six (made in this test sequence at 7, 12, 17, 21, and 26 minutes) are used to compute the test error.

FIG. 15 illustrates an embodiment of the device implemented with an electronic sensor for measuring the level changes in the measurement cylinder. This embodiment is similar to the one in FIG. 12, but can be implemented using the apparatus shown in FIG. 13 as well. In this electronic embodiment, a differential pressure sensor 626 is used; however, any type of level sensor with acceptable precision would be usable. The only difference between the embodiment in FIG. 12 and that in FIG. 15 is that the former use a tape and a sight glass 614 for measuring level changes and the latter uses an electronic sensor 626. This sensor has one port 628 connected to the measurement cylinder below the liquid level and one port 630 connected to the measurement cylinder above the liquid level. The difference in pressure can be used to determine changes in liquid level, and the sensor need only have a precision equal to that obtained by a person reading an inexpensive ruler. As mentioned above, when the measurement cylinder is properly sized, it makes little difference whether the level readings are made manually or electronically; the performance of the leak detection system will be very nearly identical in either case. FIG. 16 shows an embodiment with a capability for both manual and electronic measurements. This embodiment allows a manual test to be conducted and the level-change data to be collected and archived for later data quality assurance and auditing.