CA2286439A1 - Improvements in time domain reflectometry - Google Patents

Abstract

A time domain reflectometry method for determining positions of upper and lower interfaces of a first liquid floating on a second liquid within a vessel, using a system includes a probe in the form of a transmission line extending vertically within the vessel from a point above a highest level to be measured of the upper interface to below a lowest level to be measured of the lower interface, and an impulse radar system connected to the upper end of the transmission line, the micro impulse radar system including a microwave pulse generator. The method includes the steps of constructing and installing the transmission line so as to have a pulse response substantially free of internal reflections except at its ends, matching the radar system to the upper end of the probe sufficiently well to avoid the generation of noise or spurious echoes having an amplitude which is comparable to the magnitude of first and second return pulses due to reflections occurring at the upper and lower interfaces, while generating a readily detectable reflection forming a fiducial return pulse marking the end of the probe, and determining the positions of the upper and lower interfaces of the liquid being calculated by determining the time of flight between the fiducial return pulse and the first return pulse and calculating therefrom distance along the probe to the first interface, and determining the time of flight between the first and second return pulses and applying thereto a scaling factor dependent of the dielectric constant of the first liquid so as to determine a distance between the first and second interface. The method is preferably implemented using an MIR system, and sampling of reflected energy is preferably under control of a time base generator whose output delay is linearized by negative feedback.

Description

IMPROVEMENTS IN TIME DOMAIN REFLECTOMETRY
FIELD OF THE INVENTION
This invention relates to time domain reflectometry (TDR) and, more particularly, to techniques for the use of TDR to determine the depths of superposed layers of immiscible liquids in a tank or other vessel.
BACKGROUND OF THE INVENTION
A typical application for such techniques is in the processing of petroleum products, where it is commonly found that a layer of such a product will float on an accumulated aqueous layer in tanks or other vessels containing such products. It is, of course, desirable that quantities of both the product and the aqueous layer be ascertainable;
such quantities can be determined from the depths of the layers provided that the characteristics of the vessel are known.
US Patent No. 5,400,651 (Welch) describes a method for measuring the water level at the bottom of a storage tank using multiple sensors located at different distances from a tank bottom. This requires multiple sensors and multiple mountings.
US Patent No. 5,811,677 (Cournanc) utilizes a vertical transmission line sensor within a vessel, a swept frequency being applied to the transmission line, and the spectrum of reflections occasioned by reflections along the lin~ being analyzed to determine the spatial distribution of electrical impedance along the sensor. This requires more expensive equipment and sophisticated analysis than can be justified in most process control applications.
US Patent No. 3,812,422 (De Carolis) measures changes in impedance along the path of a co-axial cable connected to a pulse generator and short-circuited at its distal end with a view to determining interface levels between liquids and the dielectric constraints of the liquids.
US Patent No. 5,376,888 (Hook) relies on the generation of markers at predetermined points along a probe to locate the origin of other reflections occurring along the probe.
The probe structure is complex, and the system appears primarily intended for investigating the moisture content of soils.
US Patent No. 5,610,611 (McEwan) discloses the application of micro impulse radar (MIR) to liquid level detection applications. MIR is a technique developed by McEwan at the Lawrence Livermore National Laboratory of the University of California, which provides a low cost means of applying radar techniques to a variety of useful applications. There is, of course, a ~~trade-off~~ as compared to more complex and expensive radar techniques in that the pulses are of very low power and broadband, which limits both range and the selectivity of the simple sampling detection technique utilized, while accuracy is limited by the stability of the time base utilized to control the sampling. Applications such as that of the present invention strain the technique to its limit, both because the size of tanks used, for example, in the petrochemical industry can be quite large, and because of the energy absorption that occurs when seeking to penetrate to interface below the surface of stored liquid. Aqueous layers in particular present problems because contamination may render their absorption very high, while the high dielectric constant of water greatly increases time of flight, thus aggravating absorption problems and increasing the maximum time taken for a pulse to return from the far end of a probe. The broadband signals also tend to aggravate the generation of spurious signals over the line of flight due to unwanted reflections to discontinuities in the properties of the propagation medium other than those to be detected and mismatches to the propagation medium.
US Patent Nos. 5,841,606 and 5,884,231 (Perdue et al) are directed to processing the return signal in an MIR TDR
system.
Proposals have also been made for the use of TDR to determine the dielectric constant of liquids. These have usually been based upon measurement of the characteristics of the reflection at the interface at which the signal enters the liquid; the amplitude and width of such pulses are related to the change of dielectric constant which occurs at the interface.
SUMMARY OF THE INVENTION
It is an object of the present invention to enable the positions of upper and lower interfaces of a layer of first liquid floating on a second liquid in a vessel to be determined. If the dimensional characteristics of the vessel are known, this enables the volume and depth of both the first and second liquids to be determined. This is preferably carried out by TDR using a micro-impulse radar system matched to a transmission line extending through anticipated ranges of position of the interfaces. The transmission line is designed to minimize the generation of reflections of microwave energy except at a lower end of the line and at the interfaces, and the matching of the radar system to the line is made such that only a relatively small proportion of the microwave energy generated by the system is reflected at this point, with a view to ensuring that the reflections generated at the interfaces and at the matching to the line are large compared with spurious reflections occurring in the same time domain. Since the dielectric constant of the first liquid, which will usually be a petroleum product, will frequently not be known, it must be determined in order to calculate the position of the lower interface and the propagation velocity of the pulses is a function of the dielectric constant. This may be determined by a separate measurement, or by measuring the amplitude of the reflected pulses at the upper interface, which will be approximately proportional to the change in dielectric constant occurring at the interface, or by operating the system with the second liquid substantially absent, and detecting a pulse reflected at a lower end of the probe, as well as the pulse from the matching interface and from the upper interface. Since the length of the probe is known, the dielectric constant of the first liquid can be calculated, since the dielectric constant of the air or vapor above the first liquid is close to 1, allowing the height of the first interface to be calculated, and thus also the distance travelled along the probe in the first liquid. Since the time of flight within the first liquid is also known from the delay between the receipt of reflections from the first interface and the end of the probe, the dielectric constant can then be calculated.
It is found that, in TDR systems operating over long ranges and under widely differing environmental conditions, it is difficult to maintain stability of time bases used to control sampling of a stream of return signals from successive pulses, as is carried out in MIR and similar pulse radar systems. It is therefore a further object of the invention to provide an improved time delay generator to control sampling in such a system.

SHORT DESCRIPTION OF THE DRAWINGS
Figure 1 is a diagrammatic representation of the principal components of a system in accordance with the invention;
Figure 2a is a diagrammatic illustration of the system installed in a tank containing a liquid petroleum product floating over a layer of water;
Figure 2b is a graph illustrating the amplitude of the return signal received from the probe of system such as that of Figure 2 ;
Figure 3 is a block diagram of the probe transceiver and digitizer;
Figure 4 is a more detailed circuit of the time delay generator block in Figure 3;
Figure 5 is a timing diagram for the circuit of Figure 4;
and Figures 6-10 are flow diagrams illustrating aspects of the operation of the system.
DESCRIPTION OF THE PREFERRED EMBODIMENT
Referring to Figure 1, the system comprises a probe 2 in the form of a transmission line, mounted in a tank 4 to be monitored by top and bottom fittings 6 and 8, a probe transceiver and digitizer unit 10 being mounted at an upper end of the probe both mechanically and electrically. The unit 10 is connected by a cable 12 to a control and display unit 14 which houses a control computer and a power supply for the unit 10. The unit 14 may be connected to further remote control and display equipment by a cable 16.
We have found that it is desirable to optimize the performance of the probe 2 as a transmission line, both within the line and at its termination at its bottom end and its connection to the unit 10 through interface 18 (See Figure 4). A probe constructed as set forth in US Patent No. 5,781, 019 (Telder), the text and drawings of which are incorporated herein by reference, provides a balanced line which minimizes spurious reflections. High quality coaxial probes constructed to admit fluid between the coaxial conductors can also give good results. Single wire transmission lines, which they can be used in less demanding application, are very prone to developing spurious reflection signals which may render difficult the detection of wanted signals, particularly in deeper tanks.
It must be stressed that successful operation of the invention, particularly at longer ranges, is heavily dependent upon the quality of the reflected signal from the probe. Even in systems primarily intended to detect the interface between air and a liquid in a tank, a noisy reflected signal can present substantial difficulties in detection, as is well illustrated in US Patent Nos.
5,841,666 and 5,884,231 which are primarily concerned with overcoming these difficulties. Although these patents discuss the theoretical possibility of detecting further liquid/liquid interfaces beneath an air/liquid interface, it will be readily apparent that the detection difficulties apparent from the examples would be compounded in such a case, particularly systems based on MIR, the use of which is highly desirable for economic reasons, but whose immunity to spurious signals is low.
We have further found that it is important that the transmission line is well-matched to the MIR system so that reflections occurring at the interface 18 between the system and the line are limited such that the proportion of pulse energy test at this interface is kept to a low level. Some reflection is desirable to provide a fiducial or marker signal marking the beginning of the probe, but the reflection of excessive energy at this point results in the generation of spurious signals and noise distributed in the time domain, which may mask wanted signals.
Referring to Figure 2a, a tank 20 is shown schematically in which the probe 2 is mounted to extend from the top of the tank, through a vapor or air space 22, a layer 24 of a first liquid, typically a petroleum product, and a layer 26 of a second liquid, typically water, to the bottom of the tank. Referring to Figure 2b, a pulse from the unit 10 produces a reflection 30 at the interface 18 at the top of the tank, a first interface reflection 32 as it passes from air or vapor above the first liquid into the first liquid, and a second interface reflection 34 at the interface between the first and second liquids. Typically, the dielectric constant of a petroleum product will not be very high, so the pulse formed by the reflection at the first interface is of relatively small amplitude. The dielectric constant of water is much higher, and thus the pulse formed by the reflection at the second interface is of substantially greater amplitude. The velocity of propagation within the probe in water will be greatly reduced by the high dielectric constant of water, such that any significant depth of water will result in any reflection from the enc of the probe falling beyond the right hand extremity of the graph of Figure 2b. Furthermore, the rate of absorption of the pulse will be increased in water, particularly if it is impure, such that there may be no detectable reflection from the far end of the probe.
Referring to Figure 3, the unit 10 includes the interface 18 to the probe 2. The interface is connected to a microwave pulse generator 40 driven by an oscillator 42 which determines the pulse repetition rate through a time delay generator (TDG) 44 described further below. A gate pulse generator 46 is also driven by the generator 46 so as to produce gate pulses with a progressively changing delay relative to the microwave pulses controlled by a ramp signal from a ramp generator 45. The gate pulses control a sampling gate 48 which samples signals reflected from the probe. The samples are digitized by a digitizer 50 which provides an output to the control and display unit 14.
Referring now to Figures 4 and 5, a nominally 50% duty cycle signal (PRF) is filtered by the network R-R-CG and introduced to the input of U5 Isignal VCT in Figure 5). The peak to peak voltage of signal VCT is typically 0.5VDD. The voltage produced at VX applies an offset to signal VCT.
When the voltage VX is increased from 0.5V~,n, the positive duty cycle of the signal at VE becomes greater than 500. A
larger portion of the filtered waveform VCT becomes greater than the threshold of gate U5. This is the transmit path of the TDG circuit.
The PRF signal is also complemented by inverter U1.
This inverted signal, denoted VA, is also applied to an R-R
CG network similar to that employed in the transmit path.
The filtered signal at the input of U2, denoted VCG, will be the complement to that of VCT. Since this R-R-CG filtering network is also tied to VX, the same offset occurs at VCG
and VCT. When the voltage VX is increased from 0.5Vpp, the positive duty cycle of the signal at VB becomes less than 50%. A larger portion of the filtered waveform at VCG is greater than the threshold of gate U2, as with the transmit path, but only a single inversion is used to create waveform VB. This branch of the TDG circuit is the gate path.
The gates used for this part of the circuit may be of type 74HCU04. These are unbuffered High Speed CMOS

inverters. This particular device is used because it has no limitation on the input rise/fall time and, for slow inputs near the switching threshold, it acts as an amplifier more than a logic gate. This characteristic is desirable because the filtered waveforms VCT and VCG are representative of long rise/fall time inputs and the use of a buffered CMOS
device would tend to exhibit erratic switching behavior with these slow inputs. The capacitance CG of the filtering networks can be provided by the input capacitance of the HCU04 gates which is typically about lOpF.
The signals at VB and VE are then introduced to gates U3 and U7 (respectively) through a series resistor RHI. The resistors RHI and RHF, along with the associates gates U3, U4 (gate path), and U7, U8 (transmit path) effectively form hysteresis comparators. Hysteresis is used to help square up the signals VC, VF, GATE, and TX to prevent any erroneous switching due to slow inputs at VCT and VCG. The use of hysteresis becomes especially important for TDGs that are required to create long time delays (greater than 100nsec or > 15m range).
The introduction of the PRF signal to gates U3 and U7 is used to provide a common reference for the transmit and gate paths. The rising edges of VB and VE propagate through the circuit to form the rising edges GATE and TX
(respectively) which determine the time delay. The TX
rising edge will precede the GATE rising edge when the positive duty cycle of VE is greater than that of VB. This requires the voltage at VX to be greater than 0.5Vpp. The falling edges of VB and VE do not propagate because the falling edge of waveform PRF effectively resets the outputs before this can occur. This operation prevents the delay between the falling edges of VB and VE from contributing to the feedback control of the time delay. This reference ensures that the only parameter monitored and controlled by the feedback mechanism is the time delay generated.
The gates used for this feedback part of the circuit are 74AC00. They are Advanced CMOS buffered NAND gates.
These are used because of low input capacitance (~4pF), fast output rise times (~750psec), and low propagation time delay (~lOnsec) at low supply voltages. The low input capacitance is useful to minimize the delay introduced by the hysteresis resistor network (RHI and RHF) and reduces the power consumption. The fast output rise time is required to generate the large band width pulses required for TDR. The low propagation time delay is useful to assert the hysteresis effect.
A difference amplifier constituted by amplifier U10 and its associated parts monitors the time delay. The DC values at VC and VF are subtracted and the difference voltage oV
will be proportional to the time delay generated. The capacitors CF are used to extract the DC value. The difference amplifier does not monitor the TX and GATE

outputs directly. This minimizes the load at the outputs of U4, U8 which helps to shorten the rise-time.
A servo amplifier constituted by amplifier U11 and its associated parts will adjust the control voltage VS to force the difference voltage oV to be equal to the RAMP voltage.
With the bias network (2R-2R), the circuit will generate zero time delay when VS equals the digital ground (VX=O.So~,). Zero time delay occurs when both VB and VE have 50o duty cycles. The resistor RSC and capacitor CSC are used to help compensate the frequency response of the feedback loop.
The invariance of the time delay generated to component mismatch, temperature, and aging is primarily achieved by the use of the common reference (PRF) introduced to gates U3, U7 and the use of the differencing amplifier to monitor the time delay. If no reference were used, the time delay generated would depend on the matching of the R-R-CG
networks and the propagation time delay balance of the transmit and gate paths. The differencing amplifier helps eliminate any variations introduced by the net propagation time delay for either path that manifests itself in the corresponding common mode duty cycle at VC or VF.
The propagation time delays for the transmit and gate paths need to be balanced for other reasons. If the two paths were not balanced, the control voltage VS would have to offset the additional delay. This is nit desirable since the propagation time delay of HCMOS has a significant temperature dependency and the offset required for any in-balance could create instabilities in the feedback mechanism and erroneous control. The extra gate U9 attached to the output of U1 is included to help balance the propagation time delays of both paths. The gate capacitance of U9 mirrors the capacitive load that U6 has on U5 and helps to balance the propagation time delays of U1 and U5. All the additional gates for each path are equally loaded and the propagation time delay for each path after U1, U5 should be balanced.
It should be understood that circuits operating on the same principles as that of Figure 4 may also be used to control other microwave transceivers depending on sensing of return signals after a progressively changing delay, for example, a transceiver such as that disclosed in W098/36940 in place of the timing control circuit disclosed in that specification.
Aspects of the processing of the signal received from the unit 10 by the control and display unit 14 are illustrated by the flow diagrams of Figures 6-11.
Referring to Figure 6, the successive digitized echo profiles produced by the digitizer unit 10 will typically be summed, the averaging index specifying the number of profiles to be summe~(overlaid). The routine tests whether a baseline profile is stored (to provide a reference for amplitude measurements), calculates one if necessary, and applies a temperature dependent scaling value for all subsequent measurements.
Figure 7 shows that the fiducial pulse marking the top of the probe is located - this will typically be of negative polarity relative to the other pulses and thus have the lowest absolute amplitude. The location of the fiducial pulse is tested for validity, and if verified, interface pulses are searched for. If any interface pulse is found, a test is made to determine whether calibration has been performed. If any of these tests fails, the routine of Figure 9 is entered. Otherwise, processing progresses to Figure 8, where successive tests are carried out before processing progresses to Figure 9. If the first interface pulse has a high amplitude (the ~~water~~ threshold) , it represents a time of flight corresponding to an empty tank (reflection from end of probe), or to a reflection from water.
If the first pulse has a lesser amplitude, but above the amplitude characteristic of a petroleum product (~~liquid~~), the result is considered invalid. If the pulse has a still lower amplitude, it is considered characteristic of a liquid interface, and a search is made for a further pulse with a greater time of flight. According to whether such a pulse is found, there are considered to be present either upper and lower interfaces (e. g., liquid and water), or only an upper interface (liquid only).
Referring to Figure 9, the routine loads the equivalent velocity of a pulse in air or vapor, and measurement limit values, and then checks the result passed to it for various characteristics as previously tested, as set forth. In this routine, the velocity ratio is calculated from the amplitude of the liquid interface reflection, but it may be varied to substitute a stored velocity ratio (related to dielectric constant), which is calculated when the result of the processing is ~~liquid only~~, since this will enable the velocity ratio to be calculated based on knowledge of the length of the probe and the position of the first interface.
If such a calculated figure is not available, a preset figure may be employed, based on external measurement of the dielectric constant of the liquid.
Figure 10 discloses a preferred pulse search routine.

Claims (10)

1. A time domain reflectometry method for determining positions of upper and lower interfaces of a first liquid floating on a second liquid within a vessel, using a system comprising a probe in the form of a transmission line extending vertically within the vessel from a point above a highest level to be measured of the upper interface to below a lowest level to be measured of the lower interface, and an impulse radar system connected to the upper end of the transmission line, the micro impulse radar system including a microwave pulse generator comprising the steps of constructing and installing the transmission line so as to have a pulse response substantially free of internal reflections except at its ends, matching the radar system to the upper end of the probe sufficiently well to avoid the generation of noise or spurious echoes having an amplitude which is comparable to the magnitude of first and second return pulses due to reflections occurring at the upper and lower interfaces, while generating a readily detectable reflection forming a fiducial return pulse marking the end of the probe; and determining the positions of the upper and lower interfaces of the liquid being calculated by determining the time of flight between the fiducial return pulse and the first return pulse and calculating therefrom distance along the probe to the first interface, and determining the time of flight between the first and second return pulses and applying thereto a scaling factor dependent of the dielectric constant of the first liquid so as to determine a distance between the first and second interface.

2. A method according to claim 1, wherein the radar system is an MIR system.

3. A method according to claim 1, including the step of calculating the scaling factor based on the amplitude of the first return pulse.

4 A method according to claim 1, including the step determining the dielectric constant of the first liquid to provide the scaling factor.

5. A method according to claim 4, wherein the dielectric constant is derived from an amplitude measurement of the first return pulse.

6. A method according to claim 4, wherein the dielectric constant is derived by conducting the method in the absence of the second liquid, detecting a further pulse formed by reflection from the lower end of the transmission line in place of the second pulse, and relaying the time of flight between the second and further pulses to the length of transmission line beneath the first interface so as to derive the dielectric constant of the first liquid.

7. A method according to any one of claims 1-6, in which a digital signal to be processed is formed by applying a series of pulses of microwave energy to the transmission line, and sampling reflected energy in a period following each pulse using a progressively changing delay.

8. A method according to claim 7, wherein progression of the delay is obtained by applying a linear ramp signal to a time delay generator, and applying negative feedback within the generator to linearize the response of the generator to the ramp.

9. Apparatus for carrying out the method of any one of claims 1-8, including a control computer programmed to implement the method steps set forth therein.

10. A time delay generator for use in an impulse radar ranging system containing a transmitter and a receiver, in which a signal to be processed is formed by transmitting from the transmitter a series of pulses of microwave energy, and sampling reflected energy received by the receiver following each transmitted pulse using a progressively changing delay controlled by a delay generator, wherein the delay generator is controlled by a ramp signal, and the generator is configured internally to include a negative feedback loop connected to linearize the response of the generator to the ramp signal.