Method for estimating physical characteristics of two materials

Abstract

An estimate of interfacial areas between the liquid and gas, liquid and wall, and gas and wall in two phase flow is determined using a standard 2D sensor in a fashion to infer 3D information about the liquid/vapor profile when the sensor length is much longer than the diameter. Cross-sectional flow areas for the gas and liquid are also estimated as a function of the axial dimension of the sensor, and the centroid of the mass in the sensor element can also be estimated. An electric capacitance tomography (ECT) system creates tomograms of the flow inside a sensor, and estimates of 3D physical area information are produced from the tomograms.

Claims

1. A method of estimating a physical characteristic of a material, comprising: disposing a flowing mixture of a first material and a second material in a three-dimensional sensor volume; defining within the three-dimensional sensor volume a matrix of parallel voxels, each voxel having x, y and z dimensions; measuring at least one parameter of the mixture of the first and second material within the three-dimensional sensor volume and producing a tomogram in the form of a two-dimensional matrix that is defined within a perimeter and contains multiple values with one value being associated with each voxel, each value representing an amount of the first material in the associated voxel; calculating multiple hypothetical points having x, y and z coordinates in the three-dimensional sensor volume based on the multiple values of the tomogram, the z coordinate of each point being calculated from at least one value of the tomogram and the x and y coordinates of each point corresponding to at least the associated voxel of the at least one value of the tomogram; calculating at least one estimated physical characteristic of the first material based on the multiple points; and changing the flowing mixture of the first and second materials in response to at least one estimated physical characteristic of the first material to change a physical characteristic of the flowing mixture.

2. The method of claim 1 wherein calculating at least one estimated physical characteristic comprises calculating a surface area based on the multiple hypothetical points that is an estimate of the surface area of an interface between the first and second materials within the volume.

3. The method of claim 1 wherein the three-dimensional sensor volume is defined by a circumferential sensor wall and by length dimensions of a sensor and wherein calculating at least one estimated physical characteristic comprises calculating a hypothetical surface area of the interface between the circumferential wall and the first material based on the values in the tomogram along the perimeter of the tomogram.

4. The method of claim 1 wherein the three-dimensional sensor volume is defined by a circumferential sensor wall and dimensions of a sensor and wherein calculating at least one estimated physical characteristic comprises calculating a hypothetical surface area of an interface between the circumferential wall and the second material based on the multiple hypothetical points that are disposed adjacent to the circumferential wall.

5. The method of claim 1 wherein calculating at least one estimated physical characteristic comprises calculating a volumetric void fraction based on the multiple hypothetical points that is an estimate of a volume fraction of at least one of the materials within the three-dimensional sensor volume.

6. The method of claim 1 wherein calculating at least one estimated physical characteristic comprises providing mass densities of the first and second materials within the three-dimensional sensor volume and calculating an estimated centroid of mass of the mixture of the first and second material within the three-dimensional sensor volume based upon the multiple hypothetical points and the mass densities of the first and second materials.

7. The method of claim 1 wherein calculating at least one estimated physical characteristic comprises calculating an estimated cross sectional area of one of the materials within the three-dimensional sensor volume based upon the multiple hypothetical points.

8. The method of claim 1 wherein calculating multiple hypothetical points in the volume comprises: calculating the z coordinate of each point based on a value of the tomogram; and determining the x and y coordinates of each point based on the position of a voxel.

9. The method of claim 1 wherein calculating at least one estimated physical characteristic comprises: mapping a surface to the multiple points to produce a mapped surface; and calculating the at least one estimated physical characteristic of the first material based on the mapped surface.

10. The method of claim 9 wherein calculating the at least one estimated physical characteristic comprises calculating a surface area of the mapped surface to produce a calculated surface area that is an estimate of the surface area of an interface between the first and second materials within the volume.

11. The method of claim 9 wherein mapping comprises: defining a plurality of triangular planar surfaces using a plurality of combinations of three adjacent points within the multiple points whereby the mapped surface is the plurality of triangular planar surfaces; calculating the surface area of each triangular surface to produce multiple area values; and adding the multiple area values to produce the estimate of the surface area of the interface between the first and second materials within the volume.

12. The method of claim 9 wherein mapping comprises: defining a plurality of elements based upon the points, each element including four points; defining a three-dimensionally curved surface within each element; calculating the surface area of each three-dimensionally curved surface to produce multiple area values; and adding the multiple area values to produce the estimate of the surface area of the interface between the first and second materials within the volume.

13. The method of claim 9 wherein calculating the at least one estimated physical characteristic comprises calculating an estimated volume of the first material based on the mapped surface.

14. A method of estimating a physical characteristic of a material, comprising: providing a flowing mixture of a first phase of the material and a second phase of the material in a conduit wherein the flowing mixture is controlled by at least one pump and at least one valve; disposing the flowing mixture moving in a flow direction within a three-dimensional sensor volume defined by a circumferential sensor wall and length dimensions of a sensor; disposing a plurality of capacitive sensors in a side-by-side relationship around and adjacent to the circumferential sensor wall, each sensor having a width and a length, and the length of each sensor being parallel to the flow direction of the flowing mixture; defining within the three-dimensional sensor volume a matrix of parallel voxels, each voxel having x, y and z dimensions with the z dimension being parallel to the flow direction; measuring capacitance within the three-dimensional sensor volume using the plurality of capacitive sensors and producing a tomogram in the form of a two-dimensional matrix containing multiple values with one value being associated with each voxel and corresponding to the electrical permittivity of the material within the voxel that is associated with the value, each value also corresponding to an amount of the first phase of the material in the associated voxel; calculating multiple hypothetical points having x, y and z coordinates in the three-dimensional sensor volume based on the multiple values of the tomogram, the z coordinate of each point being calculated from at least one value of the tomogram and the x and y coordinates of each point corresponding to at least the associated voxel of the at least one value; calculating at least one estimated physical characteristic of the material based on the multiple points and based on the assumption that the multiple points lie on a hypothetical interface between the first and second phases within the three-dimensional sensor volume; and controlling at least one of the pump or the valve to modify the flowing mixture based upon the estimated physical characteristic.

15. The method of claim 14 wherein the estimated physical characteristic is at least one of the following: a surface area of the interface between the first and second phases, a surface area of the sensor wall in contact with the first phase, a surface area of the sensor wall in contact with the second phase, a mass centroid of the material within the three-dimensional sensor volume, the fraction of the three-dimensional sensor volume occupied by the first phase, and the cross sectional area of the first phase within the three-dimensional sensor volume.

16. The method of claim 14 wherein a surface area of the sensor wall in contact with the first phase is estimated using only the outermost values in the tomogram.

17. The method of claim 14 wherein a surface area of the sensor wall in contact with the first phase is estimated using only the outermost points of the multiple points.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) The invention may best be understood by reference to various embodiments and variations of the invention, examples of which are described below in conjunction with the drawings in which:

(2) FIG. 1 is an illustration of various different types of multiphase flow in a tube;

(3) FIG. 2a is a somewhat diagrammatic isometric view of a tubular sensor for measuring capacitance of a flowing material and producing a tomogram of the material;

(4) FIG. 2b is a somewhat diagrammatic view of the capacitive sensor shown in FIG. 2a, but the sensor in FIG. 2b is unrolled and shown in a planar configuration for illustrating its construction;

(5) FIG. 3 is a tomogram produced by the sensor shown in FIG. 2a;

(6) FIG. 4a is an illustration of the tube containing a multiphase flow with a sensor disposed to sense the multiphase flow;

(7) FIG. 4b is an isometric drawing of the voxel shown in FIG. 4a;

(8) FIGS. 5a-5d are isometric views of voxels with each Fig. showing a different type of multiphase flow of material within the voxel;

(9) FIG. 6a is a tomogram produced by the sensor of FIG. 2a and corresponding to the illustration of FIG. 6b;

(10) FIG. 6b is an isometric illustration of multiple points in a tube that are mapped to a hypothetical surface corresponding to the interface between a a liquid phase and a gas phase flowing in the tube;

(11) FIG. 7 is a graphical illustration of a tomogram showing the pixels and the elements of the tomogram along with illustrations of points that are derived from the tomogram;

(12) FIG. 8 is an illustration of a non-uniform mesh defined by a plurality of points in three-dimensional space there calculated from a tomogram;

(13) FIG. 9 is a graphical illustration of elements and master elements in a mesh that is calculated from and defined by a plurality of points in three-dimensional space that are calculated from a tomogram;

(14) FIG. 10 is a graphical illustration of how each four-sided element of a mesh is broken into two triangular planar surfaces which are then used to calculate the surface area of a mapped surface;

(15) FIG. 11a is a graphical illustration of the outermost boundary pixels of the tomogram;

(16) FIG. 11b is an isometric illustration of the hypothetical boundary line to surface areas on the wall of the sensor;

(17) FIG. 12 is a graphical illustration of representative consecutive boundary pixels and the vectors used to find the angle between the pixels;

(18) FIG. 13 is an isometric illustration of a test surface use for estimating the error in using a trapezoidal calculation to determine the surface area of the sensor wall and contact with a particular material;

(19) FIG. 14 is a graphical isometric illustration of an element or voxel for purposes of describing how the volume of an element is calculated;

(20) FIG. 15 is a schematic diagram of a voxel showing two different phases within the voxel;

(21) FIGS. 16a and 16b are illustrations of flow showing three-dimensional and two-dimensional views of the cross-sectional areas of each material phase in the flow;

(22) FIG. 17 is an illustration of a sensor detecting the interface between to materials, salt and air, in the sensor;

(23) FIG. 18 is a three dimensional illustration of 4 different tomograms representing 4 different measurements of 4 different levels of salt within a sensor;

(24) FIG. 19 is a graph showing the correspondence of the theoretical solid fractions and the measured solid fractions of salt in the sensor of FIG. 17, and

(25) FIG. 20 illustrates an apparatus for implementing the methods discussed herein.

DETAILED DESCRIPTION

(26) Referring now to the drawings in which like reference characters refer to like or corresponding parts or elements throughout the several views, a technique is disclosed to estimate interfacial areas between two materials, such as a liquid and a gas. A technique is also disclosed for estimating interfacial areas between a material and a wall, such as between a liquid and wall and between a gas and a wall in two phase flow. The technique uses a standard 2D sensor in a fashion to infer 3D information about the liquid/vapor profile when the sensor length is much longer than the diameter. It will also allow the cross-sectional flow areas for the gas and liquid to be estimated as a function of the axial dimension of the sensor. The centroid of the mass in the sensor element can also be determined. Tomograms of the flow inside a sensor may, in some embodiments, be created by commercially available electric capacitance tomography (ECT) systems. The methods disclosed herein provide a quantitative interpretation of the tomogram providing estimates of 3D physical area information.

(27) Two phase flows consisting of a liquid and gas phase are common in many applications such as air conditioning, chemical and petroleum industries. Predicting pressure drop and heat transfer rates in these flows typically depends on knowledge of the void fraction which can be defined on an area or volumetric basis. In many cases the flow may be pulsating and chaotic which leads to difficulty in characterizing the interfacial areas between the tube wall and each phase and the interfacial area between each phase. The impact of the interfacial areas, A.sub.lw, A.sub.gw, A.sub.lg can be seen in the transient one dimensional momentum equations for separated flow given as: [1] (The bracketed numbers refer to references listed at the end of this Detailed Description.)

(28) $\begin{array}{cc}{}_l\left(\frac{{\stackrel{\_}{V}}_l}{t}+V_l\frac{{\stackrel{\_}{V}}_l}{z}\right)={}_l\cos \left(\right)-\frac{{}_{\mathrm{lw}}{A}_{\mathrm{lw}}}{\frac{D^2}{4}L}+\frac{{}_{\lg }{A}_{\lg }}{\frac{D^2}{4}L}-\frac{P}{z}\mathrm{liquid}\mathrm{phase}& \left(1a\right)\\ {}_g\left(\frac{{\stackrel{\_}{V}}_g}{t}+V_g\frac{{\stackrel{\_}{g}}_l}{z}\right)={}_g\cos \left(\right)-\frac{{}_{\mathrm{gw}}{A}_{\mathrm{gw}}}{\frac{D^2}{4}L}+\frac{{}_{\lg }{A}_{\lg }}{\frac{D^2}{4}L}-\frac{P}{z}\mathrm{gas}\mathrm{phase}& \left(1b\right)\end{array}$
(The numbers enclosed within parentheses at the end of equations are equation numbers.) As may be observed in the equations, the transient liquid one dimensional momentum decreases in response to an increase in the area of the liquid/wall interface and increases in response to an increase in the area of the liquid/gas interface. Also, the transient gas one dimensional momentum decreases in response to increases in both the area of the gas/wall interface and the area of the liquid/gas interface. Thus, to understand and predict these momentums, it is important to know or estimate the aforementioned interfacial areas, A.sub.lw, A.sub.gw, A.sub.lg, and such information can be used separately or in conjunction with void fraction information to better predict physical characteristics of the flowing material, such as pressure drop and heat transfer rates.

(29) In this discussion, the following nomenclature is used:

Nomenclature

(30) a=x.sub.2x.sub.1 the width of a pixel (m)

(31) A.sub.i area of the ith pixel (m.sup.2)

(32) A.sub.gw interfacial area between gas and wall (m.sup.2)

(33) A.sub.l cross sectional flow area of the liquid (m.sup.2)

(34) A.sub.lg interfacial area between liquid and gas (m.sup.2)

(35) A.sub.lw interfacial area between liquid and wall (m.sup.2)

(36) A.sub.g cross sectional flow area of the gas (m.sup.2)

(37) , B vectors

(38) b=y.sub.2y.sub.1 the height of a pixel (m)

(39) C.sub.i i.sup.th row in the connectivity matrix gives pixel number and the corner node numbers

(40) D diameter (m)

(41) ECT Electric capacitance tomography

(42) H(x) Heaviside step function given in eq. 27

(43) J.sub.i i.sup.th Jacobian given by eq. 7

(44) L sensor length (m)

(45) L.sub.liq length of a voxel that is occupied by liquid (m)

(46) m.sub.i mass of liquid and gas in the i.sup.th voxel (kg)

(47) n1.sub.i, n2.sub.i, n3.sub.i, n4.sub.i corner node numbers for the i.sup.th element

(48) N.sub.frame number of frames used in a temporal average

(49) NP number of active pixels used in a tomogram (e.g., 812 in the illustrated embodiment)

(50) NBP number of boundary elements (e.g., 88 in the illustrated embodiment)

(51) P pressure (Pa)

(52) R radius of tube (3.5 mm)

(53) S.sub.i area of the i.sup.th element on the liquid/gas interface

(54) t time (s)

(55) V mean velocity (m/s)

(56) x.sub.i x coordinate of the ith node (m)

(57) x.sub.ci, y.sub.ci, z.sub.ci coordinates of the centroid of the mass in the i.sup.th voxel

(58) x, y, z coordinates of the centroid of the mass in the sensor volume

(59) z axial coordinate (m)

(60) z.sub.i average value of the i.sup.th element of area of wall wetted by liquid or the average of the corner nodes for a liquid/vapor surface element

(61) z.sub.gi, z.sub.li centroidal axial location for the gas and liquid respectively in the i.sup.th voxel

(62) Subscripts

(63) g gas

(64) l liquid

(65) w wall

(66) Greek Variables

(67) specific weight (N/m.sup.3)

(68) x, y lengths of the sides of the square elements (m)

(69) .sub.i angle subtended by the ith area element of wall wetted by liquid (radians)

(70) relative permittivity

(71) .sub.* normalized relative permittivity ratio

(72) .sub.* spatial average of .sub.* over a tomogram

(73) <.sub.*(t)> temporal average of .sub.*

(74) angle (radians)

(75) vertical direction in master element

(76) horizontal direction in master element

(77) density (kg/m.sup.3)

(78) shear stress (Pa)

(79) , .sub.max angle and maximum value of the angle used in FIG. 13 (radians)

(80) .sub.ij .sub.j.sub.i;=x, y, z notation used in eq. 14a.

(81) .sub.i i.sup.th shape function used in the interpolation functions given by eq. 8

(82) Many techniques have been used to estimate the void fraction including optical, gamma ray attenuation, and techniques based on either electric resistance or capacitance. The present approach uses an electric capacitance tomographic (ECT) technique to estimate the liquid/vapor interface in flows that may have different physical characteristics as shown in FIG. 1. The flow may include bubbles of gas (phase 1) contained within a flow of liquid (phase 2) as shown by flow 20 of FIG. 1. Alternatively, flow 22 illustrates plugs of gas in a liquid; flow 24 shows slugs; flow 26 shows an intermittent flow of gas; flow 28 shows a stratified wavy flow of gas; and flow 30 shows an annular flow of gas in a liquid. The method described herein estimates the area interfaces of the phases regardless of the type of flow.

(83) Some embodiments of the invention may use an ECT sensor 32 from Industrial Tomographic Systems [2]. FIG. 2a is a schematic of the sensor 32, which consists of eight electrodes 34 on a flexible printed circuit board 36. In FIG. 2a the sensor 32 is shown in an isometric view as it appears in service wrapped around a tube 38. The measurement volume 40 (See FIG. 4a) is nominally the volume inside the electrodes 34 but is found to be slightly larger due to fringing effects at the ends of the electrodes 34. In FIG. 2b the circuit board 36 is shown unwrapped with the length of the electrodes nominally 51 mm. The tube diameter for the data used in this application is 7 mm so the length to diameter ratio is approximately 7.3. The ECT sensor is measuring capacitance and producing values corresponding to electrical permittivity of the material in the sensor volume 40, and a tomogram is produced in which each value (pixel) in the tomogram is a measurement of electrical capacitance corresponding to the electrical permittivity of a discrete volume (a voxel) within the sensor volume. The measured values are of course absolute value measurements of electrical permittivity that may be calibrated to accurately correspond to the electrical permittivity of the material within the voxels. Accurate absolute permittivity values may be used in the present invention but are not necessary. Instead, for example, normalized permittivity values may also be used. If gas and liquid phases of a material are present in the sensor, all permittivity values may be normalized against the permittivity of the gas. The gas permittivity and the liquid permittivity may each be determined empirically by filling the sensor with the gas and liquid separately and measuring permittivity. There is no need to calibrate those measurements to make them accurate in an absolute sense. Instead, the measurements are normalized against the measured gas permittivity. A normalized liquid permittivity is the liquid permittivity less the gas permittivity. A normalized measured permittivity when both gas and liquid are present is the measured permittivity minus the gas permittivity, and a normalized permittivity ratio is the normalized measured permittivity divided by the normalized liquid permittivity. The method uses the values of the normalized permittivity ratio, .sub.*, defined in eq. 2 at each pixel to develop the liquid/vapor interface. When a voxel in the sensor is full of gas, the normalized permittivity ratio is zero (0), and when voxel in the sensor is full of liquid, the normalized permittivity ratio is one (1). When the voxel is filled with half liquid and half gas, the normalized permittivity ratio is one half (0.5).

(84) $\begin{array}{cc}{\mathrm{.Math.}}_{*}=\frac{\mathrm{.Math.}-{\mathrm{.Math.}}_g}{{\mathrm{.Math.}}_l-{\mathrm{.Math.}}_g}& \left(2\right)\end{array}$

(85) .sub.g is the relative permittivity of the gas phase

(86) .sub.l is the relative permittivity of the liquid phase

(87) is the relative permittivity measured at a given pixel in the tomogram.

(88) Thus a value of .sub.*=0 corresponds to gas and .sub.*=1 corresponds to a liquid.

(89) A representative tomogram 42 of a two phase mixture of the refrigerant R134a is shown in FIG. 3. Tomogram 42 is hatched to represent the relative permittivity values in each pixel of tomogram 42. For example, the bottom pixels are hatched with solid vertical lines to indicate a relative permittivity of 1, meaning the lower portion of the sensor value was completely filled with liquid. The hatching and the corresponding relative permittivity are shown in the legend 43. The relative permittivity decreases in the higher areas of the tomogram, and the upper area (hatched with vertical dashes) has the lowest permittivity which is represented in this illustration as a 0.1 relative permittivity. In the past use of this tomogram, an average of the pixel values has been used as an estimate of the liquid fraction at a cross section as given in eq. 3.

(90) $\begin{array}{cc}{\stackrel{\_}{\mathrm{.Math.}}}_{*}\left(t\right)=\frac{\underset{i=1}{\overset{\mathrm{NP}}{.Math.}}{\mathrm{.Math.}}_{*,i}}{\mathrm{NP}}& \left(3\right)\end{array}$
NP=812 is the number of active pixels used in the tomogram
The spatial value is a function of time and the temporal average given as

(91) $\begin{array}{cc}.Math.{\stackrel{\_}{\mathrm{.Math.}}}_{*}\left(t\right).Math.=\frac{1}{N_{\mathrm{frame}}}\underset{j=1}{\overset{N_{\mathrm{frame}}}{.Math.}}{\stackrel{\_}{\mathrm{.Math.}}}_{*,j}& \left(4\right)\end{array}$

(92) Here N.sub.frame is the number of frames or tomograms that are used in the time average. The void fraction is then estimated as 1<.sub.*(t)> where the number of frames is large enough to ensure a stationary average.

(93) However, this information can be used in a new method to estimate the liquid profile and thus the interfacial areas as well as the centroids of the gas and liquid regions. It also can be used to calculate the estimated volumetric void fraction of the mixture in the sensor volume.

(94) Liquid/Vapor Interfacial Area

(95) To estimate the liquid profile in the sensor volume, the new method assumes that the pixel value represents the volume fraction of liquid in a rectangular voxel bounded by the pixel area times the length of the sensor as shown in FIGS. 4a and 4b, where FIG. 4a diagrammatically illustrates the sensor volume 40 within a tube 38. In FIG. 4a, three illustrative voxels 52, 54 and 56 are shown extending for the length 50 of the sensor volume 40. A gas slug 60 is shown flowing in the liquid 58 and the gas slug extends beyond both ends of the sensor volume 40.

(96) FIG. 4b shows a hypothetical voxel 62 that may exist in a particular sensor volume 40 at a hypothetical sample time. In voxel 62, the liquid 64 and the gas 66 are separated by a single vertical interface. It is recognized that the voxels within the sensor volume 40 could have other liquid distributions than that shown in FIG. 4b as illustrated in FIGS. 5a-d that diagrammatically illustrate for other voxels 68, 70, 72 and 74. In voxel 68, a drop of liquid 64 is disposed between two regions of gas 66, and in voxel 70 a drop of gas 66 is disposed between two regions of liquid 64. Voxel 72 represents a flow of liquid 64 along the bottom of the voxel 72 with a flow of gas 66 just above the liquid 64, and voxel 74 illustrates a flow in which one inclined interface is disposed between the gas 66 and the liquid 64. The distributions shown as 5a and 5b could occur if there are small drops or bubbles so the method will not be as accurate if the features are less than the sensor length. The distribution shown as 5c could occur in very smooth stratified flow but is unlikely to be seen in practice. The distribution shown as 5d could easily occur. This, however, will be practically the same as the uniform case shown in FIG. 4b because there are 812 pixels in the current sensor with a voxel length to height or width ratio of 232. The tapered section of the liquid profile in a typical flow will be very short. Thus, despite the many different theoretical profiles that might exist in various flows, for purposes of estimating interface areas as described below, the voxels may be deemed to have interfaces as shown in either FIG. 4b or FIG. 5d, and the interface areas may be accurately estimated using such hypothetical interfaces.

(97) To estimate surface areas, the values from a tomogram, such as tomogram 80 shown in FIG. 6a, are used to calculate points on a hypothetical interface between two phases of material in the sensor volume 40. Once the points are determined, a three dimensional curved surface is mapped through the points and the surface area of that three dimensional curved surface is calculated, which is the estimated surface area between the two phases.

(98) This procedure is discussed in more detail below in connection with FIG. 6a which illustrates a tomogram 80 generated by a sensor 32 that is sensing a gas and liquid flow, and FIG. 6b, which is a somewhat schematic illustration showing the points 82 calculated from the tomogram 80 using the method described above. The liquid vapor interface 84 within a sensor volume 40 for a horizontal tube 38 is shown in FIG. 6b based on the tomogram shown in FIG. 6a. The points in FIG. 6b have hashing symbols corresponding to the hash shading of the pixels in FIG. 6a. The volume below the points 82 is the liquid phase and the gas is represented by the volume above the points 82. In FIG. 6b the length has been normalized by the diameter. This is a qualitative indication so an approach to getting a quantitative value for the interfacial areas will be presented.

(99) The following discussion for an exemplary embodiment of the invention will use a square mesh, although the described methods may be applied to any non-uniform mesh (such as shown in FIG. 8) as well. FIG. 7 illustrates the pixel 90 locations based on a square mesh, where each square in the mesh is a pixel 90. It has a total of 3232=1024 pixels but only 812 which are totally inside the tube 38 are used. A new mesh is created from the centroids 92 of each pixel 90, and each square of the new mesh is an element 94.

(100) In summary, the tomogram represents a first square mesh 89 with each square 90 in the mesh representing a pixel. The first step of the disclosed method is to create a second square mesh where the centroids of the first square mesh form the corners of each square element 94 in the second mesh. The second mesh is not fully shown in FIG. 7 to prevent clutter in the illustration, but six of the square elements 94 of the second mesh are shown in FIG. 7, and a magnified view of one square element 94 is shown in the upper right corner of FIG. 7. In the discussion above the squares in the first mesh are referred to as pixels. The squares in the second mesh are called elements, and the corners of the elements are called nodes. This terminology is employed to help distinguish the first mesh from the second mesh. After the second mesh of elements 94 is created as described above, the second mesh of elements 94 is used to calculate the points 82 as shown in FIG. 6b as described below in detail.

(101) A connectivity matrix is created which lists the elements 94 and the four corner node numbers going clockwise around the element 94, C.sub.i=(i, n1.sub.i, n2.sub.i, n3.sub.i, n4.sub.i), where i is the element number and n1.sub.i, n2.sub.i, n3.sub.i, n4.sub.i are the node numbers for the element.

(102) A matrix corresponding to the x,y,z values of the nodes is also created, where the nodes are the corners of the elements 94, which are also the centroids of the pixels 90. The z value is the measured permittivity ratio from the tomogram times the length 50 of the sensor. The surface elements 94 are thus created and are then mapped to a master element 100 as shown in FIG. 9. The surface elements 94 are mapped to the master element 100 using a linear transformation given by:

(103) $\begin{array}{cc}=\frac{2\left(x-x_1\right)-a}{a};=\frac{2\left(y-y_1\right)-b}{b}& \left(5a,b\right)\end{array}$
where a=x.sub.2x.sub.1; b=y.sub.2y.sub.1 are the lengths of the sides of the pixels 90. The area of the i.sup.th element is then given as: [4]
S.sub.i=.sub.1.sup.1.sub.1.sup.1{square root over (J.sub.1.sup.2+J.sub.2.sup.2+J.sub.3.sup.2dd)}(6)
J.sub.i are the Jacobians and depend on the mapping function and the element geometry. The Jacobians are given by:

(104) $\begin{array}{cc}{J}_1=.Math.\begin{array}{cc}\frac{y}{}& \frac{y}{}\\ \frac{z}{}& \frac{z}{}\end{array}.Math.;J_2=.Math.\begin{array}{cc}\frac{z}{}& \frac{z}{}\\ \frac{x}{}& \frac{x}{}\end{array}.Math.;J_3=.Math.\begin{array}{cc}\frac{x}{}& \frac{x}{}\\ \frac{y}{}& \frac{y}{}\end{array}.Math.& \left(7\right)\end{array}$

(105) For the chosen geometry and nodes, first order interpolation functions are used to describe the coordinates in terms of the transformed variables and the nodal coordinates. Higher order functions can be described but additional nodes would be needed for each element. Thus,

(106) $\begin{array}{cc}x=\underset{i=1}{\overset{4}{.Math.}}{x}_i{}_i\left(,\right);y=\underset{i=1}{\overset{4}{.Math.}}{y}_i{}_i\left(,\right);z=\underset{i=1}{\overset{4}{.Math.}}{z}_i{}_i\left(,\right);{}_1\left(,\right)=\frac{1}{4}\left(1-\right)\left(1+\right);{}_2\left(,\right)=\frac{1}{4}\left(1+\right)\left(1+\right){}_3\left(,\right)=\frac{1}{4}\left(1+\right)\left(1-\right);{}_4\left(,\right)=\frac{1}{4}\left(1-\right)\left(1-\right)& \left(8\right)\end{array}$
The x.sub.i, y.sub.i, z.sub.i values are the coordinates for the i.sup.th node.

(107) The Jacobians can now be found from eq. 7. This will hold for the uniform mesh shown in FIG. 7 or a non-uniform mesh such as that shown in FIG. 8. All that is needed are the nodal locations.

(108) As illustrated in FIG. 9 (a graphical illustration of elements and a master element), for the uniform mesh with constant x and y the Jacobians are:

(109) $\begin{array}{cc}{J}_1=\frac{-y}{8}\left[\left(z_2-z_1\right)\left(1+\right)+\left(z_3-z_4\right)\left(1-\right)\right]& \left(9a\right)\\ J_2=\frac{-x}{8}\left[\left(z_2-z_3\right)\left(1+\right)+\left(z_1-z_4\right)\left(1-\right)\right]& \left(9b\right)\\ J_3=\frac{x*y}{4}& \left(9c\right)\end{array}$

(110) This can be shown to be equivalent to using an expression from Larson et al page 1032. [5] The expression for the surface is given as z=f(x,y) over a square region R given by x.sub.1xx.sub.2; y.sub.1yy.sub.2. The area is given as:

(111) $\begin{array}{cc}{S}_i={}_{y_1}^{y_2}{}_{x_1}^{x_2}\sqrt{1+{\left(\frac{f}{x}\right)}^2+{\left(\frac{f}{y}\right)}^2}\mathrm{dxdy}& \left(10\right)\end{array}$

(112) Here the same mapping to a master element 100 and linear interpolation functions are used and the interface written as z(,). The partial derivatives are evaluated using the chain rule and the area becomes:

(113) 0$\begin{array}{cc}S={}_{-1}^1{}_{-1}^1\sqrt{1+{\left(\frac{2}{a}\underset{i=1}{\overset{4}{.Math.}}{z}_i\frac{{}_i\left(,\right)}{}\right)}^2+{\left(\frac{2}{b}\underset{i=1}{\overset{4}{.Math.}}{z}_i\frac{{}_i\left(,\right)}{}\right)}^2}\left(\frac{\mathrm{ab}}{4}\right)dd& \left(11\right)\end{array}$

(114) Comparing eq. 11 with eq. 6 after eq. 7 and 8 have been substituted it can be seen to be the same for the special case of square elements. The area of a given element 94 is found using Gauss-Legendre quadrature. [6]

(115) $\begin{array}{cc}{S}_i={}_{-1}^1{}_{-1}^1\sqrt{J_1^2+J_2^2+J_3^2}dd\underset{i=1}{\overset{2}{.Math.}}\underset{j=1}{\overset{2}{.Math.}}f\left({}_i,{}_j\right)w_i{w}_j& \left(12\right)\\ f\left({}_i,{}_j\right)=\sqrt{J_1^2\left({}_i,{}_j\right)+J_2^2\left({}_i,{}_j\right)+J_3^2\left({}_i,{}_j\right)}{w}_i=w_j=1& \left(13a\right)\\ \left({}_i,{}_j\right)=\mathrm{.577350269189626}*\left(\begin{array}{c}\left(-1,1\right)\\ \left(1,1\right)\\ \left(1,-1\right)\\ \left(-1,-1\right)\end{array}\right)& \left(13b\right)\end{array}$

(116) The integration reduces to the sum of 4 integrand evaluations at the locations given in equation 13b. For the uniform mesh, only differences in z at the nodes need to be calculated for each element. The other terms are the same for all elements and only need to be calculated once. The total area is the sum of the areas of the elements 94.

(117) Alternative Method to Finding the Area

(118) The above described method of mapping a surface to the points 82 of FIG. 6b uses a precise curved three-dimensional surface and produces highly accurate results in terms of mapping a curved surface and calculating the area. However, since the objective is to estimate the surface area of the gas/liquid interface, a less precise form of mapping is normally sufficient. One alternative method to find the surface area is to subdivide all the square elements 94 into two triangles as shown in FIG. 10. Each triangle lies in its own plane so the cross product of vectors along two sides can be used to determine the area of each triangle. This creates two plane facets instead of the potentially curved surface that would be obtained with the previous shape functions. Here the surface area of the interface 84 of FIG. 6b is given as:

(119) $\begin{array}{cc}{S}_i=\frac{1}{2}\sqrt{{\left(y_{42}{z}_{13}-z_{42}{y}_{13}\right)}^2+{\left(z_{42}{x}_{13}-x_{42}{z}_{13}\right)}^2+{\left(x_{42}{y}_{13}-y_{42}{x}_{13}\right)}^2}& \left(14\right)\\ {}_{\mathrm{ij}}={}_j-{}_i;=x,y,z& \left(14a\right)\end{array}$
The differences in the x and y directions are constant for the uniform mesh and would only need to be evaluated once. For a non-uniform mesh they would need to be calculated for each element.
Liquid Wetted Area

(120) Another surface area that may be calculated using the pixel data from the tomogram 80 is the surface area of the interface between the liquid (or the gas) and the wall of the tube 38 within the sensor 32 (FIG. 2a). The area of the tube 38 within the sensor volume 40 that is wetted by liquid can be estimated by using the values (normalized permittivity ratios) of the boundary pixels 110 shown in the tomogram 112 of FIG. 11a. First a hypothetical wall boundary between the gas and the liquid is generated using only the values of the boundary pixels, and thus the hypothetical boundary would be a curved boundary line on the surface of the tube 38.

(121) FIG. 11b shows a vertical sensor volume 114 within a tube 38 with a hypothetical wall boundary between liquid and gas represented by the ellipse 116, which is a simplified example for illustration purposes. The perimeter 122 of the ellipse would constitute the boundary between the liquid and the gas along the wall of the tube 38. The liquid is below the ellipse 116 and the gas is above the ellipse 116, and the surface area 118 of the wall of the tube 38 below the ellipse would be the surface area of the liquid/wall interface, and the surface area 120 above the interface represents the gas/wall interface.

(122) The liquid/wall interface surface area 118 is found by integration around the tube perimeter and is demonstrated using the trapezoidal rule although other methods could be used.

(123) $\begin{array}{cc}{A}_{\mathrm{wl}}={}_0^2\mathrm{dA}={}_0^2\mathrm{zRd}\underset{i=1}{\overset{\mathrm{NBP}}{.Math.}}{\stackrel{\_}{z}}_{i}R{}_i& \left(15\right)\end{array}$
NBP is the number of boundary pixels 110. From the uniform

(124) $z_i=L{\mathrm{.Math.}}_{*,i}\mathrm{and}{\stackrel{\_}{z}}_i=\frac{L}{2}\left({\mathrm{.Math.}}_{*,i}+{\mathrm{.Math.}}_{*,i+1}\right).$
For the last segment the first boundary pixel is used for z.sub.i+1. This completes the circle around the tube.

(125) With the uniform mesh, the arc lengths corresponding to the segments defined by the boundary pixels 110 aren't the same. In some meshes .sub.i may be a constant but the following approach will find the appropriate value using the boundary pixel 110 locations. The pixels 110 are first sorted in a clockwise or counter clockwise fashion. The location of the centroid (e.g. centroids 122a and 122b) of each pixel represents a vector (e.g., vectors 124a and 124b) extending radially from the center of the tube 38 to the pixel as shown in FIG. 12. By taking the dot product of the vectors associated with adjacent pixels 110, the cosine of the angle between the vectors can be found as:

(126) $\begin{array}{cc}\mathrm{AB}\cos \left(\right)=\stackrel{\_}{A}.Math.\stackrel{\_}{B}->{}_i=\mathrm{acos}\left(\frac{\stackrel{\_}{A}.Math.\stackrel{\_}{B}}{\mathrm{AB}}\right)=\mathrm{acos}\left(\frac{x_i{x}_{i+1}+y_i{y}_{i+1}}{\sqrt{\left(x_i^2+y_i^2\right)\left(x_{i+1}^2+y_{i+1}^2\right)}}\right)& \left(16\right)\end{array}$
and the area of the wall of tube 38 wetted by liquid is approximated as:

(127) $\begin{array}{cc}{A}_{\mathrm{wl}}\frac{\mathrm{RL}}{2}\underset{i=1}{\overset{\mathrm{NBP}}{.Math.}}\left({\mathrm{.Math.}}_{*,i}+{\mathrm{.Math.}}_{*,i+1}\right){}_i& \left(17\right)\end{array}$
The area of the tube in contact with vapor (gas) is just the total tube area minus the liquid wetted area or

(128) $\begin{array}{cc}{A}_{\mathrm{wg}}\left[\mathrm{DL}-\frac{\mathrm{RL}}{2}\underset{i=1}{\overset{\mathrm{NBP}}{.Math.}}\left({\mathrm{.Math.}}_{*,i}+{\mathrm{.Math.}}_{*,i+1}\right){}_i\right]=\mathrm{DL}\left[-\frac{1}{4}\underset{i=1}{\overset{\mathrm{NBP}}{.Math.}}\left({\mathrm{.Math.}}_{*,i}+{\mathrm{.Math.}}_{*,i+1}\right){}_i\right]& \left(18\right)\end{array}$
In practice, A.sub.wg would calculated as A.sub.wg=DLA.sub.wl.

(129) The error in using this technique has two components. The first is associated with using the simple trapezoidal rule for the integration as opposed to a more complex formula. The second is approximating the values of .sub.*,i at the tube wall as the values given by the centroids of the boundary pixels 110. The error associated with using the trapezoid rule can be estimated by looking at a case where the liquid/vapor interface forms a plane 126 that intersects the sensor volume 128 as shown in FIG. 13. In this case the error between the exact solution and the trapezoidal rule was less than 0.35% for all values of /.sub.max1, which represents all possible angles of the interface 126. This is much smaller than the expected error in the values for .sub.* (normalized relative permittivity ratio). For purposes of estimation, this error is acceptable.

(130) Volumetric Void Fraction

(131) It is recognized that the information taken from a tomogram such as tomogram 42 of FIG. 3 is volumetric in nature due to the length of the electrodes 34 (FIG. 2b). The percentage of the measurement volume 40 occupied by liquid can also be estimated if there are no bubbles in the liquid or drops in the gas. The volume beneath the liquid/gas interface surface z=f(x,y) is given as
V=.sub.A.sup.1f(x,y)dxdy=.sub.A.sup.1f(x,y)J.sub.3dd(19)
The volume under a surface f(x.y) is illustrated in FIG. 14.
Using the same bilinear mapping for the elements 94 as that used to find the liquid/vapor interfacial area the volume under an element 94 is given as:

(132) $\begin{array}{cc}{V}_i={}_{-1}^1{}_{-1}^1{J}_3\underset{i=1}{\overset{4}{.Math.}}{z}_i{}_i\left(,\right)dd=\frac{x*y}{16}{}_{-1}^1{}_{-1}^1\left[z_1\left(1-\right)\left(1+\right)+z_2\left(1+\right)\left(1+\right)+z_3\left(1+\right)\left(1-\right)+z_4\left(1-\right)\left(1-\right)\right]dd& \left(20\right)\end{array}$
As with the area calculation, this can be written for a non-uniform mesh with higher order interpolation functions as well. Evaluating eq. 20 gives

(133) $\begin{array}{cc}{V}_i=\frac{x*y}{4}\left(z_1+z_2+z_3+z_4\right)=x*y*\stackrel{\_}{z}& \left(21\right)\end{array}$

(134) This is just the average value, z, at the nodes (n1, n2, n3, n4) times the area of the pixel 90. This is exact under the assumption of the bilinear variation of z over the element. This variation can be shown to be equivalent to z=a.sub.0+a.sub.1x+a.sub.2y+a.sub.3xy. The total volume under the liquid/vapor interface is the sum of the volumes under each element. The liquid volume fraction is the liquid volume divided by the sensor volume and the volumetric void fraction is one minus the volumetric liquid fraction. In general, the volumetric void fraction is not the same as .sub.*.

(135) Centroid of the Fluid Mixture in the Sensor Volume

(136) The centroid of the fluid mixture can also be found from the tomogram data. Referring to FIG. 7, x and y components of each centroid 92 (x.sub.ci, y.sub.ci) and the area of each pixel 90 are known. The mass of each voxel 62 is m.sub.i=.sub.iA.sub.iL. The mean density is given as
.sub.i=.sub.g+.sub.*,i(.sub.l.sub.g)(22)
The coordinates for the centroid of the mass in the sensor volume is given by

(137) 0$\begin{array}{cc}\stackrel{\_}{x}=\frac{\underset{i=1}{\overset{\mathrm{NP}}{.Math.}}{m}_i{x}_{\mathrm{ci}}}{\underset{i=1}{\overset{\mathrm{NP}}{.Math.}}{m}_i}\stackrel{\_}{y}=\frac{\underset{i=1}{\overset{\mathrm{NP}}{.Math.}}{m}_i{y}_{\mathrm{ci}}}{\underset{i=1}{\overset{\mathrm{NP}}{.Math.}}{m}_i}\stackrel{\_}{z}=\frac{\underset{i=1}{\overset{\mathrm{NP}}{.Math.}}{m}_i{z}_{\mathrm{ci}}}{\underset{i=1}{\overset{\mathrm{NP}}{.Math.}}{m}_i}& \left(23a,b,c\right)\end{array}$
To find z, z.sub.ci must be calculated. This can be found by referring to a schematic of a voxel as given in FIG. 15.

(138) $\begin{array}{cc}{z}_{\mathrm{ci}}=\frac{m_{\mathrm{li}}{\stackrel{\_}{z}}_{\mathrm{li}}+m_{\mathrm{gi}}{\stackrel{\_}{z}}_{\mathrm{gi}}}{m_i}=\frac{{}_l{\mathrm{.Math.}}_{*,i}{\stackrel{\_}{z}}_{\mathrm{li}}+{}_g\left(1-{\mathrm{.Math.}}_{*,i}\right){\stackrel{\_}{z}}_{\mathrm{gi}}}{{}_i}& \left(24\right)\\ {\stackrel{\_}{z}}_{\mathrm{li}}=\frac{L{\mathrm{.Math.}}_{*,i}}{2}{\stackrel{\_}{z}}_{\mathrm{gi}}=L{\mathrm{.Math.}}_{*,i}+\left(1-{\mathrm{.Math.}}_{*,i}\right)\frac{L}{2}=\frac{L}{2}\left(1+{\mathrm{.Math.}}_{*,i}\right)& \left(25a,b\right)\end{array}$
Substituting into 23c gives

(139) $\begin{array}{cc}\stackrel{\_}{z}=\frac{L}{2}\frac{\underset{i=1}{\overset{\mathrm{NP}}{.Math.}}\left[{}_g+{\left({\mathrm{.Math.}}_{*,i}\right)}^2\left({}_l-{}_g\right)\right]}{\underset{i=1}{\overset{\mathrm{NP}}{.Math.}}{}_i}& \left(26\right)\end{array}$
Cross Sectional Flow Areas

(140) The cross-sectional flow area 150 for the gas, A.sub.g, and the cross sectional flow area 152 for the liquid, A.sub.l, phases are shown in FIGS. 16a and 16b, and can also be estimated from the pixel values of the tomogram 80. The liquid cross sectional area in the sensor 32 can also be approximated if the uniform cross section distribution is used. It is simply the fraction of pixels with lengths calculated from the permittivity ratios greater than a particular value of z. i.e.

(141) $\begin{array}{cc}{A}_l\left(z\right)=\frac{D^2}{4*\mathrm{NP}}*\underset{i=1}{\overset{\mathrm{NP}}{.Math.}}H\left({\mathrm{.Math.}}_{*,i}-\frac{z}{L}\right)\mathrm{where}H\left(x\right)=\{\begin{array}{c}1\mathrm{if}x0\\ 0\mathrm{if}x<0\end{array}& \left(27\right)\\ A_g\left(z\right)=\frac{D^2}{4}-A_l\left(z\right)& \left(28\right)\end{array}$
Eq. 27 assumes the area of each pixel is the same. If they are different the expression can be expressed as:

(142) $\begin{array}{cc}{A}_l\left(z\right)=\underset{i=1}{\overset{\mathrm{NP}}{.Math.}}{A}_{i}H\left({\mathrm{.Math.}}_{*,i}-\frac{z}{L}\right)& \left(29\right)\end{array}$

(143) Here A.sub.i is the area of the i.sup.th pixel and would be calculated once for a given mesh using Gauss-Legendre quadrature. These areas can be can be retained in this discrete form or curve fit to provide a smoother approximation to the area. This also allows an approximation of the gradients in the axial direction of the flow areas either in a finite or continuous fashion. The cross sectional area of a particular material (e.g. liquid) in the sensor (such as sensor 32) will vary depending on the z position within the sensor. One method of estimating an average cross sectional area of a material in the flow would be to determine the cross sectional area at multiple z positions within a sensor. For example, the cross sectional area of the liquid could be calculated at every 0.1 inch along the length of the sensor and then those areas could be averaged to determine an average cross sectional area.

(144) The cross sectional area of a material at one z position in the sensor may also provide useful information. For example, the cross sectional area of a liquid in a gas/liquid flow may be calculated for the center of the sensor and such area may be calculated repeatedly over time. Each of the calculated areas may be compared to the others. If the cross sectional area of the liquid is fluctuating or oscillating over time, the frequency and amplitude of the oscillations in the cross sectional area would be a measure of the frequency and magnitude of waves in the flowing mixture within the sensor.

(145) Test Results

(146) The concept was tested using a vertical tube 160 with salt 162 and air 164 as two phase media in a vertical orientation as shown in FIG. 17. This test is using static materials in a sensor volume as opposed to flowing materials to simplify the test and to enable easy verification of the calculated information as compared to the actual information. The interface between the salt and the air can be visually or physically determined and compared to the interface that is calculated based on a tomogram. In this test, the actual position of the interface between salt and air was determined visually for four different levels of salt, and the visual measurements of the interface were very nearly the same as the interface calculated from a tomogram using the techniques described above. In this test, the interface 166 is horizontal and approximately flat, and the interface lies within the sensing volume of electrodes 168. The theoretical volume fraction of salt based on electrode length is

(147) ${\mathrm{.Math.}}_{*}=\frac{z-z_1}{L}.$
FIG. 18 shows the tomograms for four different levels of salt within the sensing volume of electrodes 168. FIG. 19 shows how the sensor model based on the concept z=L.sub.* compares with the experimental results. As can be seen, the sensor is sensitive to the presence of material approximately 5 mm or 0.714 diameters outside the electrodes. This is not unexpected as the electric field extends past the ends of the electrodes. The effect can be captured by using an apparent length of 60 mm for the electrodes and provides support that the proposed methods will be accurate for liquid vapor interfaces as long as the features are longer than the length of the electrodes. Thus the technique should provide a useful metric for the interfacial areas: A.sub.lg, A.sub.lw, A.sub.gw.
Exemplary Apparatus

(148) FIG. 20 illustrates an exemplary apparatus 170 for implementing the methods described above. A conduit 171 provides a supply of flowing multiple materials, such as water and steam, to a pump 172 which pressurizes the mixture. The output of pump 172 is supplied through a conduit 174 to an input of a mixing valve 176, with the flow of the mixture being indicated by arrow 184. The mixing valve 176 also receives an input of at least one flowing material from conduit 199, which for example may be water. The mixing valve 176 independently controls the flow from conduits 174 and 199 mixes the materials from conduits 174 and 199 and outputs the mixed materials through conduit 178 to the sensor 180. The flow into and out of the mixing valve 176 is indicated by the arrows 184, 199 and 186. The sensor 180 surrounds the conduit 178 and the flowing mixture continues uninterrupted through the sensor 180 and within the conduit 178 as indicated by flow arrow 188. The sensor 180 corresponds to the sensor 32 discussed above.

(149) The sensor 180 provides an output to a controller 200 which includes data processors and communication devices for implementing the methods. The controller 200 powers the sensor 180 and receives communications from the sensor 180 through lines 202, which are communication lines and power lines. The controller 200 is also connected to power and control pump 192, valve 198, mixing valve 176 and pump 172 through the lines 202. The pump 192 is connected to a supply conduit 190 and, in this example, is supplied with water. The output of pump 192 flows through conduit 194 to valve 198 as indicated by the flow arrow 196, and the valve 198 controls the flow through conduit 199 to the mixing valve 176.

(150) The sensor 180 measures capacitance and those capacitance measurements are provided to the controller 200 which calculates a tomogram as discussed above with respect to sensor 32. The controller 200 repetitively samples the sensor 180 and repetitively produces tomograms at a rate that is sufficient for a particular application, which will vary widely. In this application, the controller 200 is configured to produce tomograms at a rate of one sample per second. The controller 200 is also configured to calculate one or more of the hypothetical physical characteristics discussed above in less than one half a second. So, for example, the controller 200 may calculate a hypothetical surface area of the interface between the water and steam within the sensor 180. In addition, the sensor 180 may calculate additional hypothetical physical characteristics, such as the hypothetical area of a wall of the sensor 180 in contact with steam. Then, the controller 200 compares the hypothetical physical characteristics against predefined limits and transmits control commands when the hypothetical physical characteristics meet or exceed the predefined limits. So, for example if the hypothetical surface area between the wall of sensor 180 and steam exceeds its predefined limit, the controller 200 issues control commands to the pump 192 and the valve 198 causing a desired amount of flow through the conduit 199 and water is introduced through the mixing valve 176 into the conduit 178. The supply of water through the mixer 176 will decrease the amount of steam in the flowing mixture and will decrease the surface area of the sensor wall that is contacted by steam.

(151) The controller 200 may also be calculating the hypothetical surface area between the water and gas within the sensor 180. Also, it may be saving each such calculation and calculating a rate of change in the hypothetical surface area between the water and gas. When this rate of change exceeds a predefined limit, that circumstance in this particular embodiment can be predicting the formation of oscillations within the flowing mixture in the conduit 178. In this particular embodiment, such waves would constitute a dangerous or catastrophic event. Thus, the controller 200 in response to such condition issues commands to stop the pump 172 and 192. In addition, it will command the valve 198 and the mixing valve 176 to stop all flow through the conduit 178, and the apparatus 170 is shut down. Alternatively, when the rate of change exceeds a predefined limit, the controller 200 may be programmed to take corrective action. For example, the pump 192, valve 198 and mixing valve 176 may receive commands to introduce more water into the flow within conduit 178. By increasing the water, hopefully, the rate of change in the surface area between the water and steam will reverse or stabilize. The controller 200 will continue to monitor such rate of change and will allow the embodiment to continue functioning so long as the rate of change remains below the predefined limit. It will be understood that all of the various hypothetical physical characteristics discussed herein may be calculated by the controller 200 and compared against one or more predefined limits, and in each case corrective actions may be executed when any of the hypothetical physical characteristics exceed their limits, and one of those corrective actions could be a complete shutdown of the apparatus 170.

(152) Having described several embodiments and variations of the invention in the above Detailed Description, it will be understood that the invention is capable of numerous modifications, rearrangements and substitutions of parts without departing from the spirit of the invention as defined in the Claims.

REFERENCES

(153) [1] Wallis, G., One-Dimensional Two Phase Flow, 1969, McGraw-Hill Inc. [2] Industrial Tomography Systems plc, Sunlight House, 85 Quay Street, Manchester, M3 3JZ, UK [3] Kreitzer, P., Hanchak, M. and Byrd, L., Horizontal Two Phase Flow Regime Identification: Comparison of Pressure Signature, ECT and High Speed Visualization, presented at 2012 ASME IMECE, Houston, Tex. [4] Taylor, A. E., Mann, W. R., Advanced Calculus, 2.sup.nd ed., 1972, Xerox College Publishing [5] Larson, R. E., Hostetler, R. P., Edwards, B. H., Calculus with Analytical Geometry 6.sup.th ed., 1998, Houghton Mifflin Co. [6] Carnahan, B., Luther, H. A., Wilkes, J. O., Applied Numerical Methods, 1969, J. Wiley & Sons, Inc.