GAS LEAKAGE METER

Abstract

A gas leakage meter (100) for use onsite in a process plant comprises a housing (110) with an inlet (101), an outlet (102) and an inlet nozzle (111) fluidly connected to a unit to be tested (3) through a downstream connector (1040) and to the inlet (101), the housing (110) being filled with a liquid to a level (10). An inclined pipe (120) is arranged such that, in use, test gas released from the inlet nozzle (111) raises through a liquid and the inclined pipe (120) to a gas-collecting chamber (122) at an upper end of the inclined pipe. (120). The gas-collecting chamber (122) has a gas release valve (150; 170) for releasing test gas before a test period. Embodiments where the inclined pipe (120) is mounted on a pivot and alternative gas release valves (150, 170) are also disclosed.

Claims

1-15. (canceled)

16. A gas leakage meter for use onsite in a process plant comprising: a housing with an inlet, an outlet and an inlet nozzle fluidly connected to a unit to be tested through a downstream connector and to the inlet, the housing being filled with a liquid to a level; and an inclined pipe arranged such that, in use, test gas released from the inlet nozzle raises through a liquid and the inclined pipe to a gas collecting chamber at an upper end of the inclined pipe, wherein the gas-collecting chamber has a gas release valve for releasing test gas before a test period, and wherein the gas leakage meter is equipped with means for measuring test gas.

17. The gas leakage meter according to claim 16, further comprising an internal environment conditioner equipped with means to alter conditions within the housing such as a heating element and/or a pressure regulator.

18. The gas leakage meter according to claim 16, further comprising a bubble counter configured to count bubbles of test gas within the inclined pipe.

19. The gas leakage meter according to claim 18, further comprising a water trap connecting the gas-collecting chamber to a vertical pipe.

20. The gas leakage meter according to claim 19, further comprising a level indicator configured to measure a liquid level in the vertical pipe.

21. The gas leakage meter according to any claim 18, wherein the gas release valve comprises a pilot chamber with a permanently open refill opening, a liquid release orifice in the bottom of the pilot chamber, a filling pipe connecting the pilot chamber to a first seat, a float able to close the first seat if a buoyancy acting on the float is less than a threshold value and able to lift a valve element from a second seat if the buoyancy is equal to or greater than the threshold value.

22. The gas leakage meter according to claim 16, further comprising a lever that can tilt about a pivot, the pivot dividing the lever into a short arm extending toward a lower end of the inclined pipe and a long arm extending toward the gas-collecting chamber, wherein lever may support assembly if it lacks sufficient rigidity; the gas leakage meter further comprising a force sensor configured to measure a buoyancy F(t) caused by test gas in the gas -collecting chamber.

23. The gas leakage meter according to claim 22, wherein the force sensor is located at the short arm of the lever.

24. The gas leakage meter according to claim 22 wherein the force sensor is located at the long arm of the lever

25. The gas leakage meter according to claim 22, further comprising means to cancel the weight of the gas-collecting chamber at the start of each test period.

26. The gas leakage meter according to claim 25, wherein the means to cancel the weight of the gas-collecting chamber is an adjustment nut able to move a spring axially and/or to provide a spring bias by means of the spring.

27. The gas leakage meter according to claim 22, wherein the means to cancel the weight of the gas-collecting chamber is a counter weight axially movable along the short arm of the lever.

28. The gas leakage meter according to claim 16, further comprising a flow divider to divide a flow or a pressure into a large and a small part to prevent inadvertent expulsion of liquid from the inclined pipe, having a first manifold chamber fluidly connected to the inlet nozzle and a second manifold chamber fluidly connected to a bypass pipe outside the housing; the flow divider further comprising a dividing disc with an open section; wherein the dividing disc has several discrete angular orientations relative to the manifold chambers such that, in use, the ratios of test gas flowing through the open section into the first and second manifold chambers are clearly defined.

29. The gas leakage meter according to claim 28, wherein the discrete angular orientations are determined by grooves and associated protrusions at fixed angular positions between the dividing disc and a raceway around the manifold chambers.

30. The gas leakage meter according to claim 16, wherein the gas release valve has a rotatable valve element with an eccentric pin attached to a spring providing a first torque α.sub.1F.sub.1 keeping the gas release valve open when the eccentric pin is located on a first side of the valve element's axis of rotation and a second torque α.sub.2F.sub.2 keeping the gas release valve closed when the eccentric pin is located on a side opposite the first side of the valve element's axis of rotation.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0053] The invention will be described in greater detail with reference to the accompanying drawings, in which:

[0054] FIG. 1 illustrates a first embodiment of a leakage meter according to the invention;

[0055] FIG. 2 is a bottom view of a first end of an inclined pipe,

[0056] FIG. 3 illustrates a check valve suitable for use within the leakage meter in FIG. 1,

[0057] FIG. 4 illustrates a binary ruler for an optical level sensor,

[0058] FIG. 5 illustrates a second embodiment of the leakage meter,

[0059] FIG. 6 illustrates a third embodiment of the leakage meter,

[0060] FIG. 7 illustrates a fourth embodiment of the leakage meter,

[0061] FIG. 8a shows a bi-stable gas release valve in an open state,

[0062] FIG. 8b shows the same bi-stable gas release valve in a closed state,

[0063] FIG. 9 is a vertical section through a flow divider,

[0064] FIG. 10 is a section along plane X-X in FIG. 9 showing a dividing disc from above; and

[0065] FIG. 11 is a top view of a raceway for the dividing disc in FIG. 10.

[0066] FIG. 12 overall system

DETAILED DESCRIPTION OF PREFERRED EMBODIMENTS

[0067] The drawings are schematic and not to scale. Several details known to the skilled person are omitted from the drawings for clarity of illustration.

[0068] The present invention concerns a gas leakage meter for use onsite in a process plant. FIG. 12 shows how a leakage meter 100 is connected to a unit to be tested 3 that is part of a process plant. When closed, the upstream isolation valve 2 isolates the unit to be tested 3 from the process plant without demounting it. A pressure source 1220 is connected to an upstream connector 1020. The pressure source 1220 contains test gas and its pressure can be regulated. The leakage meter 100 is connected to the downstream connector 1040. FIG. 1 shows the leakage meter 100 with an inlet 101, an outlet 102 and a housing 110. The inlet 101 is connected to the downstream connector 1040. For illustration, the inlet 101 and outlet 102 have external threads for connection to external tubing (not shown). External threads is one of several standard connecting means. The outlet 102 does not need connection means to release test gas to the atmosphere.

[0069] The housing 110 has an inlet nozzle 111 in the lower left of FIG. 1. The inlet nozzle 111 has an orifice that is sufficiently wide to avoid significant drops in temperature and pressure during operation. A liquid level 10 illustrates that the housing 110 contains liquid during operation. Suitable liquids include water, some alcohols and mixtures thereof. For example, glycol added to water lowers the freezing point from 0° C. We describe heating as an alternative with reference to FIG. 5. A gas volume over the liquid level 10 contains test gas, e.g. N.sub.2 or air, with pressure p and temperature T at ambient atmospheric conditions.

[0070] The housing 110 contains brackets 115 supporting an inclined pipe 120 with a first, lower end 121 at the left of the drawing, and a second, upper end at the right. A small circle under the first end 121 and another small circle within the inclined pipe 120 at a bubble counter 140 illustrate gas bubbles entering and rising gently through the inclined pipe 120 during a test period.

[0071] Specifically, the inner walls of the inclined pipe 120 enhances coalescence of small gas bubbles into bubbles with diameter approximately equal to the inner pipe diameter. The inclination is a compromise between coalescence, which benefits from longer time and smaller inclinations, and adhesion of gas to the pipe walls, which may become a problem if the inclination is too small. In the drawings, the inclination of pipe 120 is 10° relative to the liquid level 10. In real embodiments, the inclination of pipe 120 depends on the viscosity of the liquid in the housing 110, and may be significantly different from the example 10°.

[0072] The upper end of the inclined pipe 120 is connected to a gas-collecting chamber 122, hereinafter the chamber 122 for short. In use, a leak causes a volume V(t) of test gas with an overpressure Δp over ambient pressure p to collect in the chamber 122. FIG. 1 shows an exaggerated gas volume V for illustration.

[0073] The shape of chamber 122 is arbitrary. However, the inclined pipe 120 is preferably connected to the bottom of the chamber 122 in order to minimize or avoid a dead volume of liquid without function. In use, the chamber 122 is fully submersed in all embodiments of the leakage meter such that liquid may replace test gas in the chamber 122 whenever needed.

[0074] A manometer comprising a U-shaped water trap 123 and a vertical pipe 124 is connected to the chamber 122. An open, upper end of the vertical pipe 124 is exposed to gas pressure p over the liquid level 10. Note that the liquid level in the vertical pipe 124 varies over a much shorter distance h(t) than the heights 53.5 cm and 13.4 cm from examples in the introduction.

[0075] The bubble counter 140 counts bubbles in the inclined pipe 122 as they pass. We consider bubbles/min at STP as a unit of measure equivalent to units of ml/min, ml/s, etc. In particular, the actual bubbles may be adapted to a desired resolution and the bubble count converted to units of ‘standard bubbles’/min, ml/min or some other unit of choice.

[0076] For a numerical example, a bubble with volume 1/16 ml=62.5 μl corresponds to a sphere with diameter 4.92 mm. This might suggest an inner diameter 4.92 mm of the inclined pipe 120. However, a desired additional decimal of precision and available standard sizes for pipes might suggest an inner diameter 2.0 mm, which may contain a spherical bubble 4.2 μl. Corrections due to gas pressure p and gas temperature T will be described later.

[0077] A level indicator 141 measures a difference level Δh(t.sub.i)=h(t.sub.i)−h(t.sub.j)at discrete times ti and tj during each test period. The measurements are not necessarily performed at regular intervals iΔt. The first measurement h(t.sub.S) is at is when the test period starts. Subsequent measurements are relative to h(t.sub.S) rather than to the liquid level 10. Thus, there is no need for an accurate liquid level 10. The liquid just needs to cover the chamber 122. In particular, the actual liquid level 10 may deviate at least a few mm from a nominal filling level without affecting measurements significantly.

[0078] An optical version of the level indicator 141 may comprise a light source such as a LED laser, a light detector, a collimating and/or magnifying lens 142 and a binary ruler 143.

[0079] For use with optics, the binary ruler 143 may be a sheet of cardboard, plastic or metal with a printed or cut out pattern to be described with reference to FIG. 4. Ambient conditions do not significantly affect the binary ruler 143.

[0080] Usually, ambient conditions do not affect the function of light sources, detectors or electronic circuits in optical versions of the bubble counter 140 or the level indicator 141. In comparison, vapor may condensate on a lens such as the lens 142, and reduce visibility and functionality if the lens is colder than the liquid within the housing 110. For example, outdoor temperatures at or below 0° C. occur naturally in large parts of the world, and pure water in the housing 110 might be heated under such conditions. To avoid problems with condense, it is possible to keep the lens 142 at a temperature close to the liquid temperature, or use some principle other than optics in the bubble counter 140 or level indicator 141. Low power commercial devices using electric conductivity, resistance or capacitance may be viable alternatives even in Ex-areas.

[0081] Smaller bubbles, e.g. 2 mm bubbles, increases resolution and reduces the risk for false positives caused by counting undersized bubbles. For example, a 2 mm bubble with volume 4.2 μl provide better resolution than a bubble with 1/16 ml=62.5 μl. N bubbles less than 2 mm cause an error much less than N.Math.4.2 μl. For comparison, N bubbles less than 1/16 ml, each assumed to be 62.5 μl, cause larger deviations. Opposite, 2 mm bubbles increases the risk for bubble trains, for example a ‘bubble’ longer than 2 mm in an inclined pipe 120 with inner diameter 2.0 mm. A bubble train with ten 2 mm bubbles corresponds roughly to 20 mm liquid column in a pipe 124 with 2.0 mm inner diameter.

[0082] Before each test period, the gas release valve 150 at the top of chamber 122 may open fully to ensure that liquid completely replaces gas in the chamber 122. During each test period, the gas release valve 150 permits single bubbles to escape from the chamber 122, but delays a bubble train or burst of test gas long enough to estimate the amount of gas in the bubble train by means of the manometer 123-124 and the level indicator 141-143.

[0083] It is possible to achieve this functionality by means of a simple control loop with an electronic pressure sensor and a solenoid valve. However, electronic pressure sensors have disadvantages discussed in the introduction. FIG. 1 illustrates an embodiment of the gas release valve 150 based on a buoyancy that disappears if or when the gas volume Vin chamber 122 becomes too large. FIG. 3 is a detailed view of the gas release valve 150.

[0084] If the devices 140 and/or 141 are optical devices, at least part of the pipes 120 or 124 must be transparent. There are no large loads on the internal components 120-124, so a transparent thermoplastic, for example PET or acrylic glass, may be a suitable material for one or more of these components. Further beneficial properties of thermoplastic polymers, e.g. mechanical and chemical properties as well as the recyclability of PET are available online and in literature.

[0085] An optional flow divider 130 is essentially a valve able to divide a flow or a pressure into a large and a small part. A first purpose is to prevent inadvertent expulsion of liquid from the inclined pipe 120 and/or the housing 110. For this, the divider 130 initially leads the entire flow from the inlet 101 through a bypass pipe 132. Before the test period starts, part or all of the flow from the inlet 101 is diverted to the inlet pipe 131 by means of a flow actuator 133, in the drawings illustrated by a lever arm with a ball on a distal end. Carefully opening for flow through the inlet pipe 131 prevent the undesired expulsion of liquid. As mentioned in the introduction, the upstream pressure valve in Wilson's test device may be opened slowly and carefully for the same purpose.

[0086] A second purpose of the flow divider 130 is to estimate leaks larger than the maximum leaks required for approval or certification. Assume, for example, maximum allowed leaks as in previous examples and a flow significantly larger than 3.5 ml/min through the inlet 101. We want to estimate the larger leak even if the chamber 122 is designed for leaks less than or equal to 3.5 ml/min. The flow divider 130 divides the incoming flow into an inlet flow through the inlet pipe 131 and a bypass flow through the bypass pipe 132. The actuator 133 controls the ratio of inlet flow to bypass flow, preferably in predetermined steps. This will be further explained with reference to FIGS. 9-11.

[0087] Before we continue with the description of the drawings, we introduce a few more definitions and make some general remarks.

[0088] We have already defined t.sub.S as the start time of a test period, and now introduce t.sub.E as the end time. The duration of a test period is T.sub.test=t.sub.E−t.sub.S. Formally, a parameter such as p(t), h(t) or T(t) during a test period might be denoted (.Math.)(t−t.sub.S). We use the convention that t is reset to zero at the start of each test period such that (.Math.)(t−t.sub.S)=(.Math.)(t) for 0<t<T.sub.test.

[0089] All embodiments of the proposed leakage meter have a ‘pass or fail’ mode, in which ‘fail’ means that the amount of test gas collected during the test period T.sub.test exceeds a limit volume V.sub.limit. For example, T.sub.test=3 minutes times 3.50 ml/min give V.sub.limit=10.5 ml, and the unit to be tested fails the pressure test if V(3 min)>10.5 ml in this example.

[0090] Preferred embodiments perform additional measurements during each test period.

[0091] Ideal gas laws adequately describe conditions in real gases at pressures and temperatures occurring in gas pressure tests. A common form is, in SI units:

pV =nRT   (1)

where p is pressure in Pa, Vis gas volume in m.sup.3, n is the number of mols, R=8.314 J/(mol.Math.K) is the universal gas constant, and T is the temperature in K.

[0092] From (1), it follows immediately that the number of mols

n=pV/RT   (2)

[0093] It may be convenient to use n from equation (2) in external systems because it facilitates comparison between measurements taken at different locations and/or at different times. In particular, n for a particular measurement ‘includes’ the gas ratio pV/T.

[0094] In some numerical examples below, we use resolutions 1 Pa and 1 K. Resolutions in units of hPa (mbar) and 1 K (° C.) are widely used in practice. Inexpensive barometers and thermometers with these resolutions are available for manual or automatic measurements. However, the skilled person knowing the problem at hand must determine the precision and resolution for p, V and T for use in real embodiments.

[0095] For illustration of equation (2), 4.5 ml gas at 1 atm=101325 Pa and 20° C.=293 K contains n=101325.Math.4.5.Math.10.sup.−6/(8.314.Math.293) mol=187.18 μmol. Further, the ideal gas is a good approximation to real gases under all conditions of interest, so n is independent of test gas.

[0096] Storing and using n for comparisons does not exclude storing and using other parameters. For example, ambient temperature and/or icing may be a suspected cause of deviation for a particular type of valve to be tested. Accordingly, ‘(ambient) temperature’ may appear in an ‘Environment’ branch in an Ishikava diagram for the valve to be tested, and be mandatory in a test schema for this reason and/or because T is needed to adjust gas volumes.

[0097] A standard volume V.sub.0 at STP can be computed as V.sub.0=nRT.sub.0/p.sub.0. However, ‘standard’ temperature and pressure depend on local definitions. Varieties include T.sub.0=273 K (0° C.) or 293 K (20° C.), and p.sub.0=1 bar=1000 hPa (10.sup.5 Pa) or 1 atm=1013.25 hPa=760.00 mmHg.

[0098] It is necessary to convert gas volumes V measured at p and T into V.sub.0 or V.sub.limit at STP. For this, we rewrite equation (1) as:

V.sub.0=(p/T)(T.sub.0/p.sub.0)V   (3)

[0099] To illustrate scale of volume corrections, we assume ambient pressure p in the range 920-1050 hPa, which is within recorded atmospheric extremes at 887 hPa and 1085 hPa. Further, we assume T in the range 0-50° C., which also may occur naturally, e.g. if equipment is exposed to sunshine for an extended period. Further, some pressure test standards specify specific temperatures or temperature ranges, e.g. 38° C. (100° F.) or 5-50° C.

[0100] In this example, we use T.sub.0/p.sub.0=293/1013.25 K/hPa. Using limits from the previous paragraph, p/T ranges from 920/323 to 1050/273 hPa/K. From (3), it follows that V.sub.0 or V.sub.limit varies from 0,82 V to 1.11V in this example. Regardless of whether these ranges and numbers are representative, they illustrate that corrections of measured volume V to ‘standard’ volume V.sub.0 may approach 10-20%, and thus that volume corrections according to equation (3) are significant and necessary.

[0101] Next, we consider the vertical position of the bubble counter 140. A bubble at the bubble counter has gas pressure p+Δp, where

Δp=pgh   (4)

[0102] Similar to parameters shown in FIG. 1, Δp is an overpressure over ambient pressure p and h is a height relative to the liquid level 10. As usual, p is liquid density and g≈9.8 m/s.sup.2. In this example, the bubble counter 140 is a fixed distance from the liquid level 10, so the relative difference Δh between measurements at different times is not relevant.

[0103] The ratio V/V.sub.0=(p.sub.0/T.sub.0)T/(p+ pgh) is independent of bubble size, and thus a useful measure for the effect of counting bubbles at different depths h. Table 1 shows results with density p(T) for pure water at h=0, 10, 20 and 30 mm.

TABLE-US-00001 TABLE 1 Effects of ambient p and T compared to depth of measurement within housing 110. h/mm 0 10 20 30  4° C., 1050 hPa 0.979 0.978 0.977 0.976 4° C., 920 hPa 1.117 1.116 1.115 1.114 50° C., 1050 hPa 1.142 1.141 1.140 1.139 50° C., 920 hPa  1.303 1.302 1.300 1.299

[0104] The first row in Table 1 contains values of h in mm, and the first column contains values for temperature and pressure in each row. Each cell in the rest of Table 1 contains values of

[0105] V/V.sub.0=(p.sub.0/T.sub.0)T/(p+ pgh) to 3 decimals We have used p.sub.0/T.sub.0=101325/273 Pa/K, p=1000.00 kg/m.sup.3 at 4° C. and p=988.05 kg/m.sup.3 at 50° C. The acceleration of gravity g=9.81 m/s.sup.2 in this example.

[0106] Starting with the column for h=0, it appears that V/V.sub.0 increases from 0.979 to 1.303 with increasing T and decreasing p. Following each row, it appears that V/V.sub.0 decreases with approximately 1/1000 per cm added depth in pure water. In practice, this means that a deviation plus/minus a few mm from a nominal liquid level over the bubble counter 140 has little effect on bubble sizes and fail criteria. A similar result applies to overpressure Δp in the chamber 122. As before, values of p and T do affect gas volumes significantly.

[0107] Bubble sizes at the bubble counter 140 are expected to vary about a mean value, e.g. measured volume V=4.2 μl from a previous example. Adding bubble volumes reduces uncertainty because deviations from the mean in both directions tend to cancel each other. Sample mean and (signed) sample variance from basic statistics provide estimates for mean and variance if one desires concrete numbers. In such methods, ‘outliers’ such as bubble trains containing several bubbles may be replaced by a maximum bubble size, e.g. corresponding to 5 or 10 bubbles. The latter requires a bubble counter able to estimate bubble sizes to a certain degree, not necessarily accurately. Techniques using a camera may achieve this.

[0108] For a slightly more theoretical example, consider two abstract sensors. The first sensor corresponds to the bubble counter 140, and adds 1 to a bubble count whenever it detects a bubble, regardless of bubble size. The second abstract sensor provides the gas volume in a bubble train whenever a bubble train arrives. The second sensor corresponds to the arrangement with gas release valve 150, manometer 123-124 and level indicator 141.

[0109] The Kalman filter (KF) mentioned in the introduction is a mathematical model outside the scope of the present invention. However, for the later description it is useful to know that a KF can provide accurate estimates from quite coarse measurements. In essence, whenever a measurement arrives, the KF multiplies the measurement with a weight K and a predicted value with a weight (1−K) to obtain a weighted sum. In a formal KF, the Kalman gain K is a matrix. Here and in a modular KF, the first abstract sensor has a scalar Kalman gain K.sub.1 and the second abstract sensor a Kalman gain K.sub.2. The gains K.sub.1 and K.sub.2 change over time and quantify the confidence to put on measurements relative to predictions.

[0110] Incidentally, Poisson processes may conveniently model the first and second abstract sensors for a KF or SPC. We refer the interested reader to Kelly (2006) for a comprehensive description of a modular KF. Kelly (2006) also contains a collection of statistical formulas and derivations that may be used or useful in an external system

[0111] FIG. 2 shows a first end 121 of the inclined pipe 120 seen from below. During operation, a wide shallow spoon shape ensures that test gas is collected and fed into the inclined pipe 120. Studs 125 enhance coalescence of tiny amounts of test gas into bubbles. The studs 125 may be omitted and/or replaced by other means known to promote coalescence.

[0112] FIG. 3 is a detailed view of the gas release valve 150 shown in FIG. 1, and we describe FIG. 3 in the context of FIG. 1 during operation.

[0113] Before each test period, liquid enters a pilot chamber 151 through a permanently open refill opening 152. A liquid release orifice 153 in the bottom of the pilot chamber 151 provides a delay in a state described below. Before each test period, liquid enters the chamber 122 partly through the liquid release orifice 153, but mainly through a filling pipe 154 leading to a first seat 155. In FIGS. 1 and 3, the first seat 155 has the shape of a truncated cone or funnel. Liquid entering through the filling pipe 154 provides buoyancy for the float 156, which lifts a valve element 157 from a second seat 158 enabling gas in the top of chamber 122 to escape. This is the normal situation when bubbles arrive one by one.

[0114] If a burst of test gas arrives, the buoyancy from the float 156 becomes less than a threshold value. A level h.sub.1 illustrate the minimum amount of liquid in the chamber 122 providing the buoyancy needed to keep the valve 150 open. If or when a bubble train or burst of test gas suddenly increases the gas volume, the buoyancy disappears such that the spherical float 156 falls into the cone or funnel, engages the conical wall and thereby temporarily prevents test gas from escaping through the filling pipe 154. At the same time, the valve element 157 drops a short distance onto the second seat 158 to prevent test gas from escaping through the top. The shapes of the elements 156 and 157 and their respective seats 155 and 158 may be altered without inventive effort.

[0115] When the liquid level in chamber 122 is below h.sub.1, i.e. when the buoyancy is too small to keep the gas release valve 150 open, liquid slowly enters the gas volume in chamber 122 through the orifice 153. Liquid entering through the refill opening 152 replaces liquid leaving through the liquid release orifice 153.

[0116] The liquid release orifice 153 causes a delay that depends on several factors, e.g. orifice geometry, viscosity of the liquid and a depth h.sub.2 from the liquid level 10 to the orifice. A practical way to determine a suitable delay is to test different orifice diameters d for a given liquid at an approximate depth h.sub.2. There is no need for an accurate delay. The delay just has to be long enough to perform the necessary measurements of Δh in pipe 125, and short enough to allow gas from the bubble train to escape before the next (few) bubble(s) arrive. Recall that that some known external systems, e.g. a KF, may compensate for measurement errors. In the present context, the gas volume in chamber 122 is proportional to the difference Δh between liquid levels in the vertical pipe 125 before and after the bubble train arrived.

[0117] From the description in the past few paragraphs, it follows that a main purpose of the pilot chamber 151 is to provide walls for the liquid release orifice 153 and the pipe 154. Thus, the pilot chamber 151 may have any shape, and the refill opening 152 may be an open top of a cylinder or simply be a hole in the wall of chamber 122.

[0118] The bottom of the pilot chamber 151 is preferably slightly conical with apex up to avoid a gas trap under the pilot chamber 151. Dashed lines extending downward from either side of the orifice 153 illustrate such a conical bottom of the pilot chamber 151.

[0119] Further, since the buoyancy from float 156 must carry the weight mg of the valve element 157, the mass m of element 157 is preferably small. Since the valve element 157 is submersed during operation, it should have a density slightly greater than the liquid density. For example, acrylic glass has density 1.16-1.18 g/cm.sup.3, which is slightly greater than the density of water with or without additives around 1.0 g/cm.sup.3. It follows that reducing the mass of the valve element 157 amounts to reducing its size. Since small spheres are cheaper to make and less sensitive to orientation than small, truncated cones, a sphere may obviously replace the frustoconical valve element 157 shown in FIG. 3.

[0120] In FIG. 3, a stiff rod transfers lifting force from the float 156 to the valve element 157. The rod is shown as a straight line inclined 5° to the vertical inserted into cylindrical bores in the float 156 and the valve element 157. The bores are wide relative to the rod to reduce the need for accurate horizontal alignment of these elements and their respective seats.

[0121] Preferably, the stiff rod is unconnected or flexibly connected to at least one of the elements 156 and 157. In use, this allows the float 156 and the valve element 151 to sink or fall into their respective seats 155 and 158 at different speeds.

[0122] FIG. 4 illustrates a binary ruler 143. Binary rulers for use with optical readers usually have a pattern printed on a suitable material such as cardboard, metal or plastic. Alternatively, binary rulers for use with optical readers may comprise an opaque material with punched out, transparent fields. The example in FIG. 4 has four columns, where white means ‘0’ and black means ‘1’. A level h(t) crosses a unique pattern, in FIG. 4 ‘black, white, black, white’, which for h(t) in FIG. 4 is interpreted as binary 1010=1.Math.2.sup.3+0.Math.2.sup.2+1.Math.2.sup.1+0.Math.2.sup.0=10 (=1.Math.10.sup.1+0.Math.10.sup.0) in decimal numbers.

[0123] The rightmost column defines the resolution, e.g. by alternating white and black rectangles each 1.0 mm high. To achieve better resolution, a column with alternating white and black rectangles each 2.sup.−1=½ mm high would be added to the right hand side of the binary ruler in this example Similarly, it is obviously possible to add columns to the left hand side of the binary ruler 143 to represent 2.sup.4, etc.

[0124] Later during the test period, we assume that a new level h(t+iΔt) is detected at the pattern ‘black, black, white, black’ corresponding to binary 1101=13 decimally. At 1 mm resolution, the difference Δh=h(t+iΔt)−h(t)=(13−10) mm measures a pressure change in the chamber 122 independent of the liquid level 10. A 3.0 mm column of pure water at 4° C. corresponds to a change Δ(Δp)=10.sup.3 kg/m.sup.3.Math.9.8 m/s.sup.2.Math.3.0.Math.10.sup.−3 m=29.4 Pa. In general, a resolution 1 mm water column corresponds to a fine pressure scale about 10 Pa=0.1 hPa.

[0125] Incidentally, the binary ruler 143 is similar to a binary search from left to right, and the principle is by no means new. The ‘binary principle’ is widely used in non-optical sensors.

[0126] FIG. 5 illustrates an alternative embodiment of the leakage meter 100 that utilizes environment control and a lever principle. We do not repeat the description of components 101-111 and 130- 133 explained with reference to FIG. 1.

[0127] An auxiliary chamber 103 attached to the housing 110 illustrates one or more spaces fit to contain equipment and components as required. Such a chamber without reference numeral contains the upper end of the vertical pipe 124 in FIG. 1.

[0128] An internal environment conditioner 104 represents elements to alter conditions within the housing 110, for example, a heating element and/or a pressure regulator. In terms from control theory, the conditioner 104 is an actuator in a control loop. A temperature sensor 112 and/or a pressure sensor 113 may provide an associated feed forward or feedback.

[0129] In general, internal conditions such as temperature, pressure, humidity etc. may be different from corresponding conditions in the surrounding atmosphere. This approach minimizes adverse effects of atmospheric conditions rather than correcting for them. Insulation, tightness and other features of the housing 110 should of course be adapted accordingly.

[0130] Regardless of internal temperature, the pressure p may be equal to or very close to atmospheric pressure. For example, a gas tight housing 110 may comprise a bellow, a flexible membrane or other gas tight means to equalize internal and atmospheric pressure p. Further, internal pressures p (+ .Math.p) and temperature T will be different from STP in most cases.

[0131] The temperature sensor 112 represents sensors to measure any temperature of interest, for example, gas and liquid temperatures within the housing 110 and/or the temperature of the atmosphere surrounding the leakage meter 100. The temperature sensor 112 supplies data to downstream information systems automatically as opposed to manual measurements entered by means of a keyboard.

[0132] For example, a control system may comprise a heater that changes liquid temperature faster than atmospheric temperature. The control system may need an automatic temperature sensor 112, whereas manual measurements may suffice for atmospheric temperature.

[0133] A major benefit of environment control in the housing 110 is an ability to use sensors calibrated for a narrow range of conditions. For example, the electronic pressure sensor 113 may be accurate to within 0.1 hPa (0.1 mbar) within a controlled, narrow temperature range.

[0134] As in FIG. 1, the inclined pipe 120 is attached to the bottom of the chamber 122. However, the chamber 122 in FIG. 5 may be considerably larger than needed to contain a bubble train, for instance permitted leaks 1.5 ml/min×3 min=4.5 ml or 3.5 ml/min×3 min=10.5 ml compared to 10/16=0.625 ml plus a small safety margin.

[0135] Contrary to the embodiment in FIG. 1, the inclined pipe 120 in FIG. 5 can tilt about a pivot 116. A stopper 126 under the chamber 122 ensures that the inclined pipe 120 leads test gas toward the chamber 122 at all times, i.e. that an angle θ between the inclined pipe 120 and a horizontal plane is always greater than zero. As before, a minimum angle θ reduces the risk for adhesion of test gas to the inner wall of pipe 120, especially at small leaks.

[0136] A lever 127, e.g. a thin beam or sheet with its smallest dimension along the pivot axis, may support the assembly 120-122 if the assembly, in particular the inclined pipe 120, lacks sufficient rigidity. The inclined pipe 120 and the lever 127 may be the same mechanical component, so FIGS. 6 and 7 do not show the lever 127 explicitly. Either way, the pivot 116 divides the lever 127 into a short arm with length a extending toward the first end 121, and a long arm with length b extending toward the chamber 122.

[0137] A gas release valve 170 in the top of chamber 122 replaces the gas release valve 150 shown in FIG. 1. If the chamber 122 is designed to collect test gas during the entire test period, it may suffice to open the gas release valve 170 only between test periods in order to replace test gas with liquid in the chamber 122. The gas release valve 170 may be any suitable valve, for example a commercially available solenoid valve certified for Ex-areas or a manually operated valve. We present an energy efficient alternative with reference to FIGS. 8a and 8b.

[0138] During operation, a buoyancy force F(t) increases as the volume of test gas in the chamber 122 displaces liquid. The buoyancy force of interest equals the weight of liquid displaced since time t.sub.s. With conventions described previously:

F(t)=pgV(t)−F(0); 0<t<T.sub.test   (5)

[0139] F(0) is a downward force exerted on the stopper 126 at t.sub.S, e.g. when the chamber 122 is filled with liquid and the gas release valve 170 is fully closed. In words, equation (5) states that F(t) becomes less negative as the gas volume V(t) increases. Once the buoyancy from V(t) overcomes F(0), F(t) becomes positive, i.e. directed upwards. For accuracy, F(0) should be as small as practically possible.

[0140] By the lever principle, an upward force F at distance b from the pivot 116 exerts a downward force −(b/a)F measured at distance a on the opposite side of the pivot. For example, (b/a)=10 implies that the downward force is ten times the buoyancy force at the chamber 122. This may add a decimal of precision. However, measurement errors will also be multiplied by b/a.

[0141] As mentioned in the introduction, some electronic pressure and force sensors are relatively sensitive to temperature. This is not necessarily a concern in a housing 110 heated to a stable temperature, for instance a typical lab temperature near 20-22° C.

[0142] FIG. 6 essentially shows the lever in FIG. 5 acting with a ‘large’ force F.sub.S=−(b/a)F on a spring 161. At equilibrium, the spring 161 acts on the lever at a with a force −F.sub.S=kx where k is a spring constant and x is a downward extension of the spring in this example. A small downward extension x at a corresponds to a ‘large’ upward displacement X=−(b/a)x at b. Measuring and adjusting ‘large’ quantities is a widely used strategy to reduce relative errors.

[0143] A dashed line around the box 112 illustrates that the temperature sensor 112 is not necessarily part of the leakage meter 100. As noted, instruments outside the leakage meter may measure the ambient temperature and pressure required for correction of gas volumes.

[0144] The spring 161 is a specific embodiment of the force sensor 160 in FIG. 5. It is used as an example in FIG. 6 because the spring extension x has the approximate size of a, b and other typical dimensions in the housing 110, e.g. to within an order or two of magnitude. It is irrelevant that the measured quantity is actually a displacement as long as the displacement represents the force F. For comparison, the element sensing force in another force sensor 160 might be an elastic beam with displacement or strain several orders of magnitude smaller than a, b etc. Here, it is irrelevant that force sensors using strain actually may measure stress, i.e.

[0145] force per unit area, as long as the output from the force sensor 160 represents force.

[0146] An adjustment nut 162 connects the spring 161 to the housing 110. Turning the adjustment nut 162 causes a vertical displacement z, which should not to be confused with the spring extension x. For example, the spring 161 may be a tension spring unable to compress from equilibrium. Then, the entire spring 161 may move downward a distance z.sub.1 until it exerts a force −(b/a)F(0) on the lever at a, thereby opposing F(0) in equation (5) to within acceptable limits. In another example, the adjustment nut 162 might cause a downward spring bias F.sub.S0=k.sub.Z1 when the adjustment nut 162 moves downward the distance .sub.Z1. F.sub.S0=(b/a)F(0) would also cancel F(0) in equation (5) to within predefined limits.

[0147] Either way, we may set F(0)=0 in equation (5) and solve for V(t):

V(t)=(k/pg)X(t) 0<t<T.sub.test   (6)

where X is a displacement at b adjusted such that X(0)=0 at the start of each test period.

[0148] Summarized and neglecting opposite signs on opposite sides of the pivot 116, at a, x(t)=(a/b)X(t) cancels F.sub.S=(b/a)F=kx, so the lever ratio b/a does not appear in equation (6). Still, measuring displacement at b does reduce uncertainty by the factor a/b, so the lever ratio is implicit in the apparatus in FIG. 6. Further, the spring constant k determines the ratio V(t)/X(t) and thereby amplification through the factor k/pg in eq. (6).

[0149] In order to illustrate measurement of vertical displacement at b, we assume a binary ruler 143 as in FIG. 4 where the white fields are transparent and the black fields are opaque. The binary ruler 143 is attached to the chamber 122 at right angles to the inclined pipe 120. A light source 144 and a focusing lens 145 produce a fine focused line across the binary ruler 143. In this example, a light detector 146 has four vertical photosensitive strips such that light passing through a transparent field produces a voltage. Thus, a unique horizontal pattern of opaque and transparent fields produces a unique pattern of voltages from the four strips. A sequence ‘0101’ in Courier normal font illustrates a binary output on line 147. Further treatment of binary and electrical signals is beyond the scope of the present invention, but still well known to those skilled in the art.

[0150] Since the binary ruler 143 forms the angle θ with the vertical in this example, the measured height is X(t)cosθ.In practice, θ may be reduced by an angle corresponding to the start position where the chamber 122 touches the stopper 126. This corresponds to a vertical ruler 143 in the start position, i.e. not 90° from the pipe 122 as shown in FIG. 6. Since cosθ≈1 for small angles θ, angle corrections may be superfluous in real embodiments. For example, 15° is relatively large for a ‘small’ angle, and causes a correction [1−cos(15°)]=0.034 mm for a 1 mm slit in the binary ruler 143. This cosine correction is less than 10% of a height about 0.5 mm of the horizontal line of light produced by a typical focusing lens 145.

[0151] FIG. 7 shows an alternative embodiment in which the spring 161 is attached to the chamber 122 at b. There is no need to repeat explanations of reference numerals 10-161 and symbols p, T, θ, a, b and x. As before, ambient pressure p and temperature T are required for volume corrections, but the associated temperature and pressure sensors 112, 113 are not necessarily parts of the leakage meter 100.

[0152] In FIG. 7, a counter weight 163 can rotate on fine threads 164 on the short arm of the lever, and replaces the adjustment nut 162 shown in FIG. 6. Rotating the counter weight 163 causes an axial displacement that, in effect, alters the lever ratio b/a to exactly oppose the weight of the liquid filled chamber 122 at the start of each test period. In terms defined above, a counter weight 163 with mass m at an adjusted distance a provides a weight mg=(b/a)F(0) that opposes F(0) from equation (5) to within predefined limits.

[0153] The skilled person understands that there is no need for both an adjustment nut 162 and a counter weight 163, and further that a lock screw, a jam nut or equivalent means may prevent undesired rotation of either element 162 or 163. Furthermore, any force sensor 160 may replace the spring 161 without inventive effort.

[0154] The counter weight 163 and the liquid filled chamber 122 are exposed to the same local acceleration of gravity g, so g cancels in a force equation relating gravity acting on the chamber 122 and the counter weight 163. This principle distinguishes a traditional balance from a traditional scale. As is well known, traditional ‘reference weights’ for a balance are usually labelled in units of mass, e.g. kg or gram, rather than in units of force such as Newton.

[0155] The mass of liquid displaced by a gas volume V(t) within the chamber 122 is pV(t). Moreover, equation (6) indicates that V(t) is proportional to the vertical displacement X at b. As discussed, we may safely neglect the cosine correction at small angles θ. Perhaps more important, equation (6) follows from assumed linear relationships between buoyancy, volume and displacement in the sense that second order effects become insignificant. For the spring 161 or an elastic beam, the linear assumption is equivalent to elastic deformation and Hooke's law. For the volume V(t), a linear relationship to X implies that it suffices to ‘measure’ the chamber 122 filled with gas and filled with liquid in order to determine a maximum and minimum X. All values in between are proportional:

V(t)/(V.sub.max−V.sub.min)=X(t)/(X.sub.max−X.sub.min)   (7)

[0156] The liquid filled chamber 122 in previous examples corresponds to V.sub.min=0. However, V.sub.min≠0 is possible in equation (7). Similarly, any X.sub.min≠0 is equivalent provided the difference X.sub.max−X.sub.min remains constant. Setting X.sub.min=0 is just a convention.

[0157] Measuring mass is generally easier than measuring volume, so the exact volume V.sub.max may be determined by filling the chamber 122 with liquid at a known density and measure the mass before and after filling. For a numerical example, we assume a lab and 22° C. at which pure water has density p=997.76 kg/m.sup.3. Further, we assume that a lab balance measures the difference between masses before and after filling to Δm=5.678 gram independent of local gravity. The internal volume V.sub.max=(5.678±0.001)/0.99776=(5.691±0.001) ml. In this example, mg precision on a lab balance corresponds to μl precision in volume. This is comparable to the 4.2 μl resolution of a 2 mm bubble.

[0158] FIGS. 8a and 8b illustrate a bi-stable embodiment of the gas release valve 170 in an open and a closed state, respectively. The valve 170 has a shell 171 mounted in the top of chamber 122. The shell 171 has a vertical bore and contains a rotatable valve element 172, e.g. a ball or a cylinder. The valve element 172 has a secondary bore, which opens for flow through the valve 170 when aligned with the vertical bore as in FIG. 8a.

[0159] FIG. 8a shows the valve 170 in its open state before each test period. Dash-dot lines cross at the valve element's axis of rotation. A spring attached to an eccentric pin 173 acts on the rotatable valve element 172 with a torque a.sub.1F.sub.1. As long as the eccentric pin 173 remains below the valve element's axis of rotation, the torque a.sub.1F.sub.1keeps the secondary bore in a vertical position such that the valve 170 remains open.

[0160] FIG. 8b shows the valve 170 in its closed state during test periods. In this state, the eccentric pin 173 is above the valve element's axis of rotation, and the spring attached to the eccentric pin 173 acts on the rotatable valve element 172 with a torque a.sub.2F.sub.2. As long as the eccentric pin 173 remains above the valve element's axis of rotation, the torque a.sub.2F.sub.2. keeps the secondary bore in a horizontal position such that the valve 170 remains closed.

[0161] It is understood that the same technical effect is achieved whenever the spring forces Fi and F.sub.2 are applied to opposite sides of the valve element's axis of rotation, not necessarily over and under the valve element's axis of rotation.

[0162] A lever 174 connected to the valve element 172 is able to rotate the valve element 172 in the shell 171. In FIG. 8a, the lever 174 is directed 45° from a horizontal axis, i.e. upward to the right, when the valve 170 is open. In FIG. 8b, the lever 174 is directed 135° from the horizontal axis, i.e. upward to the left, when the valve element 172 is rotated 90° in order to close the valve. Suitable stoppers (not shown) limit the rotation.

[0163] The spring attached to the eccentric pin 173 may be a tension spring attached to the wall of the chamber 122 to the left of components shown in FIGS. 8a and 8b. Alternatively, a compression spring or a torsion spring may provide similar torques a.sub.1F.sub.1 and a.sub.2F.sub.2. Either way, the spring attached to the eccentric pin 173 provides the force required to keep the valve 170 open or closed. Thus, the only energy required to operate the valve 170 is the work needed to overcome a limited spring force over a short distance separating the two stable states.

[0164] In terms of force and distance, factors a.sub.2/b.sub.1 and a.sub.1/b.sub.1 substantially reduce the force required to open or close the valve if b.sub.1 is a lever arm substantially longer than a.sub.1 and a.sub.2. A longer lever arm b.sub.1 along the lever 174 increases the distance over which an actuator has to act on the lever. In terms of power and time, smaller power implies longer time to supply the required energy. Thus, it is possible to operate the bi-stable valve 170 in FIGS. 8a and 8b at low power, e.g. supplied from a battery through a low-power inductive actuator. Note that wires from the housing to the gas release valve 170 might affect weight measurements, and that .inductive couplings do not require wires.

[0165] FIG. 9 is a vertical section through the flow divider 130. As explained with reference to FIG. 1, the inlet pipe 131 leads to the inlet nozzle 111, the bypass pipe 132 leads test gas away from the housing 110 and the lever 133 represents an actuator.

[0166] The lever 133 is attached to a rotatable dividing disc 134, which in FIG. 9 separates an upper compartment from a lower manifold. In use, the upper compartment receives test gas through the inlet 101. The lower manifold has at least two manifold chambers 1310 and 1320. One or more first manifold chamber(s) 1310 is/are fluidly connected to the inlet pipe 131, and one or more second manifold chamber(s) 1320 is/are fluidly connected to the bypass pipe 132.

[0167] The orientation of the flow divider 130 in FIG. 9 is not limiting. Real embodiments may have the manifold chambers 1310, 1320 on a side rather than in the lower part, etc.

[0168] The dividing disc 134 may be set in one of several discrete angular orientations by means of protrusions, e.g. pins or balls, fitting in grooves in a raceway 135. A compression spring 136 urges the protrusions into the grooves and the dividing disc 134 against vertical walls within the manifold. A gasket 137 under part of the dividing disc 134 seals for gas with a small overpressure.

[0169] A central shaft 138 is attached to the dividing disc 134 and inserted in a sleeve 139 attached to the shell of the flow divider 130. The shaft 138 in the sleeve 139 prevents that the dividing disc 134 tilts relative to the raceway 135. This facilitates using loose balls instead of pins between the dividing disc 134 and the raceway 135.

[0170] FIG. 10 is a section along plane X-X in FIG. 9, and illustrates the dividing disc 134 viewed from above. The dividing disc 134 has a gas tight closing part 1341 and an open part 1342 through which some internal walls in the lower manifold are visible. A thin rectangle illustrates a mandatory division between the at least two chambers 1310 and 1320. Radial lines illustrate optional internal walls in the lower manifold. In real embodiments, the walls represented by radial lines and by a thin rectangle may have similar designs.

[0171] In use, the angular orientation of the dividing disc 134 determines the ratio of amount of test gas entering the chamber 1310 to the amount of test gas entering the chamber 1320. In FIG. 10, this ratio is 5/3 corresponding to five sectors visible through the opening 1342 to the left of the thin rectangle, and three visible sectors to the right of the thin rectangle.

[0172] In general, the flow divider 130 diverts a precisely known fraction of a ‘large’ leak arriving through the inlet 101 to the inlet pipe 131 such that the ‘large’ leak may be estimated from measurements produced by the leakage meter 100 and the precisely known fraction.

[0173] FIG. 11 is a top view of a raceway 135 for the dividing disc 134. Grooves 1350 may conveniently be drilled holes with a diameter slightly less than protrusions, e.g. pins or balls, located between the dividing disc 134 and the raceway 135. As best seen in FIG. 9, the spring 136 urges the protrusions into the set of grooves 1350, thereby creating the set of angular orientations.

[0174] While the invention has been described by means of examples, the scope of the invention is determined by the following set of claims.

REFERENCES

[0175] Alonzo Kelly: “A 3D State Space Formulation of a Navigation Kalman Filter for Autonomous Vehicles”, Carnegie Mellon University, Rev 2 2006, original from 1994