GB2223850A - Identifying flow conditions (eg distribution of different fluid phases in a pipe) - Google Patents

Abstract

Unknown flow conditions are identified in terms of their similarity to simulated conditions for which computed sensor output values have previously been stored. Variations from standard flow patterns (eg core, annular and stratified distribution of fluid phases) can thus be identified. An array of eight capacitive electrodes 1-8 may be used, giving twenty-eight capacitive output values that can be combined to form a vector representative of actual flow conditions for comparison with a "fingerprint" vector representing standard flow conditions. <IMAGE>

Description

FLOW CONDITION IDENTIFYING SYSTEM The present invention relates to a method and apparatus for identifying flow conditions within a pipe through which a fluid flows.

As reported by Huang, S.M., Green, R.G., Stott, A.L., and Beck, M.S. in wsroceedings of the 3rd International Conference on Multiphase Flow", Trle Hague, Netherlands, 18-20 May 1987, it has been proposed to use capacitance sensing techniques to provide a simple and economic means for implementing flow imagine systems. The proposal envisag te positioning of an array of electrodes, for example eight in total, around a pipe through which a multiphase flow passes.It was proposed to measure the capacitance between any two of the sensor electrodes and to reconstruct from this measured data an image of the component distribution within the pipe using an adaptation of back projection algorithms known from applied potential tomography developed for medical imaging purposes. It was suggested that stray-immune transducers would enable the sensitivity of the sensor to be focussed into a relatively small area of the pipe cross-section.

The implementation of the proposal outlined in the above report presented various problems. In particular in situations where the flow pattern can change rapidly, rapid data capture and processing is essential. Such situations are common in industrial processes. Furthermore, because of the differences between the distances separating various pairs of electrodes between which capacitance measurements must be made resulting from the distribution of the electrodes around the pipe, the measurements made with different electrode pairs have very different sensitivities. This considerably complicates the design of the measuring circuits required to measure the capacitance between the electrode pairs.

An operational system for obtaining useful data from a structure of the type referred to above is reported by Huang, S.M., Plaskowski, A.B., Xie, C.G.

and Beck, M.S. in "Capacitance based tomographic flow imaging system" Electronics Letters, 1988, 24, pp418-419. The described system enables an image of the flow conditions within a pipe to be produced on a visual display unit and thus is a powerful tool when a detailed analysis of the flow is required. There are circumstances however in which it is not necessary to display an image as such, but rather to be able to detect the occurrence of particular flow conditions, or to detect changes in previously stable flow conditions. It is an object of the present invention to provide a system with capabilities of this type.

According to the present invention, there is provided a method for identifying the flow conditions within a pipe through which a fluid flows, wherein an array of sensors is positioned around the pipe, each of the sensors providing an output representative of flow conditions in a region of the pipe physically adjacent the respective sensor, simulating predetermined flow conditions within the pipe and recording computed sensor outputs for the simulated conditions, monitoring the outputs of the sensors resulting from unknown flow conditions, and comparing the monitored outputs of the sensors with the recorded outputs of the sensors to identify the unknown flow conditions in terms of the similarity between the unknown flow conditions and the simulated conditions.

The present invention also provides an apparatus for identifying the flow conditions within a pipe through which a fluid flows comprising an array of sensors positioned around the pipe such that each of the sensors provides an output representative of flow conditions in a region of the pipe physically adjacent the respective sensor, means for recording computed sensor outputs corresponding to simulated flow conditions within the pipe, means for comparing the outputs of the sensors when the flow conditions are unknown with the stored outputs, and means for correlating the compared outputs to identify the unknown flow conditions in terms of the similarity between the unknown and simulated flow conditions.

Preferably, the sensors comprise pairs of electrodes each defining a respective capacitance, the electrodes being arranged around the pipe wall.

Each electrode forms a capacitance with each of the other electrodes, whereby an array of n electrodes defines n(n-l)/2 capacitances.

The expected capacitance between each pair of electrodes for simulated known conditions can be recorded. The resultant data can be considered as a vector having n(n-l)/2 elements. If unknown flow conditions through the pipe are the same as or similar to the previously simulated known conditions, the resultant vector will be the same as or similar to the recorded vector. The electrode array is such that different flow conditions produce markedly different vectors, and thus flow conditions can be readily identified by reference to prestored flow-representative vectors.

Where the type of flow conditions within the pipe remain constant, but relative concentrations of phases within the flow vary,-the measurement vectors which result are not constant. For example, where one phase occupies a cylindrical core region of the pipe, and another phase occupies an annular peripheral region of the pipe around the core, marked variations in the diameter of the core region will significantly vary the measurement vector. To compensate for such variations, the capacitances between each pair of electrodes can be computed for a range of concentrations to provide a set of vectors for each type of flow condition. Each vector can then be normalised and then the vectors can be averaged over the full concentration range.The result is a "fingerprint" vector with which any subsequent measurement vector can be compared to identify the type of flow condition to which the fingerprint vector is related. The fingerprint vectors are unique over wide concentration ranges.

An embodiment of the present invention will now be described, by way of example, with reference to the accompanying drawings, in which: Fig. 1 is a schematic sectional diagram through a pipe supporting an array of eight electrodes; Fig. 2 illustrates three capacitance measurement vectors obtained using the electrode array illustrated in Fig. 1 to monitor different flow conditions, and Fig. 3 graphically represents the correlation between measurement vectors and "fingerprint" vectors for three standard flow conditions over a range of flow concentrations.

Referring to Fig. 1, the illustrated arrangement comprises eight electrodes 1 to 8 sandwiched within the wall of the pipe having an inner surface 9 and an earthed outer surface 10. The mutually inclined axes 11 and 12 represent the coordinates of a finite element model of the eight electrode system. As described in the above mentioned paper published in Electronics Letters, a model of a multielectrode capacitance sensor structure of the type illustrated in Fig. 1 can be produced using finite element methods so as to calculate the capacitances of all the twenty-eight electrode pairs, that is the electrode pairs 1-2, 1-3, 1-4, 1-5, 1-6, 1-7, 1-8, 2-3, 2-4, --- 6-7, 6-8 and 7-8.

Using finite element methods the capacitances monitored between each of the electrode pairs can be computed providing the permittivity of the material within each element of the mesh shown in Fig. 1 inside the inner surface 9 is known. Thus 'standard' flow patterns can be simulated which correspond to flow patterns often found in practical applications, for example, core flow patterns in which a relatively narrow substantially cylindrical region along the axis of the pipe is occupied by a different material from the rest of the interior of the pipe, annular flow patterns in which a relatively narrow annular region adjacent the pipe wall is occupied by a different material from the rest of the interior of the pipe, and stratified flow patterns in which one side of the interior of the pipe is occupied by different material from the other side of the interior of the pipe.

For a given flow pattern to be simulated, the resultant twenty-eight capacitance changes Cij(f ) (i=1,2,...,7; j=i+l,...,8) can be computed over a range of concentrations i for a flow phase of constant permittivity t= . For example there may be a first phase of permittivity equal to 3.0 corresponding to pneumatically-conveyed sand and a second phase of permittivity 1.0 corresponding to air. The capacitance changes Cij(5)are the changes resulting from the replacement of air within the pipe by a material of different permittivity and can be considered as the static responses of the eight electrode capacitance flow measuring system, that is the instantaneous responses to a particular flow pattern.

The calculated data for the eight electrode capacitance transducer system can be considered as a vector C of twenty-eight elements ck (k=1,...,28).

Thus C = (cl, c2, c3, ..., c28) where ck follows the consistent order set out below: cl=C12, c2=C13,...,c7=C18, c8=C23,...,c28=C78.

Fig. 2 illustrates pseudo-waveforms which can be plotted by recording the variation of ck with k, the horizontal axis representing the element ck and the vertical axis the amplitude of each of these elements.

The upper waveform in Fig. 2 corresponds to core flow with a concentration ss of 0.393, the intermediate waveform of Fig. 2 corresponds to annular flow with concentration g equal to 0.204, then the lower waveform in Fig. 2 corresponds to one particular position/amount of stratified flow with a concentration X equal to 0.336. The waveform will vary with the relative location of the surface of the stratified flow and its depth. Although the pseudo waveforms illustrated in Fig. 2 each correspond to a particular concentration tests have indicated that the pseudo waveforms for each type of flow pattern are characteristic of that particular flow pattern over a wide range of concentrations.This means that it is possible to identify the pattern of a particular f-low by reference to the known pseudo waveforms examples of which are illustrated in Fig. 2.

Using by way of example three types of flow patterns represented by the pseudo waveforms of Fig.

2, a set of vectors C can be obtained for each of these three flow patterns at Mp different flow concentrations approximately covering the concentration range 0 to 1. Each vector can be identified as Cpq, where the subscript "p" denotes the type of flow pattern ( p =c" for core flow, tp"="aw for annular flow and "p"="s" for stratified flow) and "q" denotes the "q"th vector from the set of Mp vectors.Each vector C can be normalised to give

Each normalised vector can be identified as The "fingerprint " vector Cp for flow pattern "p" is defined as the average of the normalised vectors over the full concentration range, ie.

These fingerprint vectors of the standard flow patterns can be stored in a computer as a database, and the identification of unknown flow patterns can proceed by comparing any measured capacitance vector with each of these fingerprints.The closeness of the normalised measured capacitance vector to the fingerprint is measured by the relative distance between these two vectors, which is defined as follows: For two vectors X = (x1,....,xk) and Y = (Yl---,Yk), where Y is the reference vector (the fingerprint vector), then the relative distance h xy between them is

In terms of flow pattern identification, K mr is a measure of the match between the measured normalised capacitance vector and the capacitance fingerprint vector. If, using a particular fingerprint vector, the parameter #mr is small, then it can be concluded that the measured vector has resulted from a flow with a flow pattern very similar to that of the fingerprint. In this context "small" is defined as being less than a chosen upper limit hL.

In order to assess the value of > L, values A mr can be calculated for the three types of flow over the complete range of ss using the fingerprint vectors Cco, Cao, Cso and the calculated "measured" C' capacitance vectors Cpq (q=1,...,Mp). The results are illustrated in Fig. 3. This figure shows that the capacitance measurement vector of each flow pattern matches its own fingerprint vector best, as is indicated by the low values of & and aa and# compared with other & mr's (m#r). Note that core flows at high concentration (ss > 70%) are difficult to distinguish from the other flow patterns. An upper limits L can be set at 0.65.Based on this, a flow pattern identification system can be established which is valid for5470%.

After the type of flow pattern has been identified the twenty-eight capacitance measurements (ck,ckÇ gm) can be further used in conjunction with the appropriate calibrations Cij ( ss ) (i=1,...7, j=i+1,.,8) to give an estimate of the flow concentration.For example, assuming annular flow, an estimated concentration ss1 can be obtained from a measurement of cl and the calibrated C12(# ) curve for annular flow by means of a suitable interpolation function: B1 = fal(cl) A concentration vector g can be built up comprising 28 elements thus ss = (ss 1,---, ss28) = (fa1(c1),..., fa28(c28)) = fa(cm) (5a) Similarly for core and stratified flow patterns: ss = fc(Cm), ss = fS(cm) (5b) The "best" estimate of concentration, g mt is then obtained by

A problem with the interpolated concentration estimates is that some of the Cij( ) curves are not single-valued. It is therefore necessary to select the correct value. Sets of normalised capacitance vectors Cpqfl have already been calculated over a range ofX for the three standard flow patterns. A coarse estimate of g can be realised by calculating the relative distances between these vectors, for the appropriate flow pattern, and the normalised measured capacitance vector . The g giving the minimum vector distance can then be used as the coarse estimate.

Although only data for three standard flow patterns (core, annular and stratified) is discussed above, the same techniques could be extended to include more flow patterns.

Claims (6)

1. A method for identifying the flow conditions within a pipe through which a fluid flows, wherein an array of sensors is positioned around the pipe, each of the sensors providing an output representative of flow conditions in a region of the pipe physically adjacent the respective sensor, simulating predetermined flow conditions within the pipe and recording computed sensor outputs for the simulated conditions, monitoring the outputs of the sensors resulting from unknown flow conditions, and comparing the monitored outputs of the sensors with the recorded outputs of the sensors to identify the unknown flow conditions in terms of the similarity between the unknown flow conditions and the simulated conditions.

2. An apparatus for identifying the flow conditions within a pipe through which a fluid flows comprising an array of sensors positioned around the pipe such that each of the sensors provides an output representative of flow conditions in a region of the pipe physically adjacent the respective sensor, means for recording computed sensor outputs corresponding to simulated flow conditions within the pipe, means for comparing the outputs of the sensors when the flow conditions are unknown with the stored outputs, and means for correlating the compared outputs to identify the unknown flow conditions in terms of the similarity between the unknown and simulated flow conditions.

3. An apparatus according to Claim 2, wherein the sensors comprise pairs of electrodes each defining a respective capacitance, the electrodes being arranged around the pipe wall, each electrode forming a capacitance with each of the other electrodes, whereby an array of n electrodes defines n(n-1)/2 capacitances.

4. An apparatus according to Claim 3, wherein means are provided for computing the capacitances between each pair of electrodes for a range of concentrations for substantially constant flow conditions, whereby a set of vectors is provided for these flow conditions, and means are provided for comparing the resultant vector with subsequent measurement vectors to identify the type of flow condition to which the computed vector is related.

5. A method for identifying flow conditions substantially as hereinbefore described with reference to the accompanying drawings.

6. An apparatus for identifying flow conditions substantially as hereinbefore described with reference to the accompanying drawings.