Device and method for estimating the flow rate generated by a pump

Abstract

The invention relates to a method for estimating the estimated flow rate Q generated by a pump (3); said method comprising: - measuring a static suction pressure Pa at a point A located upstream of the pump (3), a static discharge pressure Pr at a point R located downstream of the pump (3), and a rotational speed N of the pump (3); and - estimating the estimated flow rate Q at least as a function of the variables Pa, Pr, and N. Figure to be published with the abstract: Fig. 1

Description

Title of the invention: Device and method for estimating the flow rate generated by a pump. Field of the invention

[0001] The invention relates to the field of fluidic installations and in particular to the pumps equipping such installations.

[0002] It relates more particularly to a device and a method for estimating a flow rate generated by a pump. Technological background

[0003] In the prior art, pumps are generally equipped with flow meters, which are physical sensors that allow the flow they generate to be measured.

[0004] Flow meters have several drawbacks. They can be prone to failure, leading to breakdowns and reduced pump performance. Furthermore, it is often difficult to predict or detect their malfunctions. Finally, although it is possible to ensure redundancy in flow measurement by using multiple flow meters, this increases costs.

[0005] Furthermore, for economic reasons, it is sometimes decided to use pumps without a flow meter, particularly when the cost of such a sensor exceeds that of the pump. This approach leads to suboptimal pump performance and energy consumption. Summary

[0006] Also, a problem that the present invention aims to solve is to provide a device and a method for estimating the flow rate generated by a pump allowing one or more of the aforementioned drawbacks to be overcome.

[0007] The present invention aims in particular to provide a device and a method for estimating the flow generated by a pump, allowing to replace at least temporarily a flow meter and / or to detect and / or compensate for its failures, in order to guarantee continuous control of the pump without interruption of the flow.

[0008] In order to solve this problem, and according to a first objective, a method for estimating an estimated flow rate Q generated by a pump having a rotor is proposed; said method comprising: - measure a static suction pressure Pa at a point A located upstream of the pump, a static discharge pressure Pr at a point R located downstream of the pump, and a rotational speed N of the pump rotor; and - estimated the flow rate Qestimated at least as a function of the variables Pa, Pr and N.

[0009] Thus, the value of the flow generated by the pump is estimated without using a signal delivered by a flow meter, which makes it possible to replace it temporarily, or even permanently, thus reducing the cost of the fluid installation while preserving the performance of the system and / or detecting a malfunction of such a flow meter.

[0010] According to embodiments, such a process may include one or more of the following characteristics.

[0011] According to one embodiment, the estimated flow rate Q is estimated using an estimator that uses a model based on a polynomial equation of the type: AP = f (Q, N ) ; with : AP: a variable representing the pressure difference Pr-Pa; Q: the flow rate; and N: the rotational speed of the pump rotor.

[0012] According to one embodiment, point A is located at a height Za and point R at a height Zr and: - AP = + Za - Zr; with p: the density of the fluid and g: the acceleration of gravity. The model can be simplified if Za = Zr.

[0013] According to one embodiment, the polynomial equation AP — f(Q, N) is as follows: AP = (A-a) x Q2 + B x Q + C with : A = Aref; B=Bref*l^ / \2 c=c re J^-) / « j \ re f / 8 _ ( 1 1 \ a “gx>T2»36002 g: the acceleration due to gravity (in m / s2); Da: an internal diameter of an upstream pipe at point A; Dr: an internal diameter of a downstream pipe at point R; Nref: a rotational speed of the pump rotor for which a polynomial equation HMTref = h (Q) establishes a relationship between a total manometric head HMTref and a flow rate Q at said rotational speed Nref; Aref, Bref, and Cref: the quadratic, linear, and constant coefficients of the polynomial equation HMTref = h(Q) = Aref Q² + Bref Q + Cref

[0014] According to one embodiment, the estimation process includes a preliminary model parameterization step comprising: - determine several pairs of values ​​corresponding to a total manometric head and a flow rate QmesUré for a plurality of opening levels of a valve located downstream of the pump while the rotor of said pump operates at a constant rotational speed taking the reference value Nref; - perform a polynomial interpolation in order to determine the coefficients Aref, Bref and Cref of the polynomial equation HMTref = h (Q) = Aref * Q2 + Bref xQ+Cref passing through the pairs of values ​​thus determined.

[0015] According to one embodiment, the estimator is an extended Kalman filter. This estimator has the advantage of being applicable to nonlinear systems, as in this specific case, and provides a real-time estimation of the flow rate.

[0016] According to one embodiment, the extended Kalman filter is based on: - a state vector taking the form: X(t) = APestil(t) where APestil(U): an estimate of AP at time t and Estimated Q(t) Estimated Q(t): an estimate of the flow rate Q at time t; and - a measurement vector taking the form: Y(t) = [ A Pmesureit)] with A Pmesure (t) = + Za - Zr, at time t.

[0017] According to a second object, a method for controlling a pump is proposed, comprising the step of controlling the pump as a function of the estimated flow rate Q estimated according to a aforementioned estimation method in order to control it to a setpoint value.

[0018] According to some embodiments, a control method may further include: - control the pump according to a flow rate QmeSuré measured using a flow meter; - compare the measured flow rate Q to the estimated flow rate Q; - detect when a difference between the measured flow rate (Qmeasured) and the estimated flow rate (Qestimated) exceeds a threshold; and - switch from pump control based on the measured flow rate Q to pump control based on the estimated flow rate Q in response to such detection.

[0019] According to a third object, a device for estimating an estimated flow rate Q generated by a pump is proposed, said estimating device comprising: - a connectivity module configured to receive a static suction pressure Pa measured upstream of the pump, a static discharge pressure Pr measured downstream of the pump, and a rotational speed N of the pump rotor; and - a calculation module configured to estimate the estimated flow rate Q at least as a function of the variables Pa, Pr and N.

[0020] According to embodiments, such an estimation device may include one or more of the following characteristics.

[0021] According to one embodiment, the calculation module includes an estimator using a model based on a polynomial equation of the type AP = f (Q, N) with: AP: a variable representing the pressure difference Pr-Pa; Q: the flow rate; and N: the rotational speed of the pump rotor.

[0022] According to one embodiment, the estimator is an extended Kalman filter.

[0023] According to a fourth object, a fluidic installation is proposed comprising a pump, an upstream pipe and a downstream pipe which are respectively arranged upstream and downstream of the pump, an upstream pressure sensor delivering the static suction pressure Pa, a downstream pressure sensor delivering the static discharge pressure Pr, a speed sensor delivering a signal representative of the rotational speed N of the pump rotor, and an estimation device of the aforementioned type. Brief description of the figures

[0024] Other features and advantages of the invention will become apparent from the following description of particular embodiments of the invention, given by way of example but not limitation, with reference to the accompanying drawings in which:

[0025] [Fig-1] is a schematic representation of a fluidic installation equipped with a pump and a virtual flow sensor.

[0026] [Fig.2] is a schematic representation of the functional architecture of the sensor virtual.

[0027] [Fig.3] is a flowchart representing the parameterization steps of the model used by the virtual sensor.

[0028] [Fig.4] represents a reference HMT curve, obtained experimentally, and plotted for a specific rotational speed of the pump rotor.

[0029] [Fig.5] is a schematic representation of the virtual sensor calculation module of Speed.

[0030] [Fig.6] is a flowchart representing the calculation steps implemented by the calculation module.

[0031] [Fig.7] is a schematic representation of the virtual sensor detailing its module connectivity. Description of the implementation methods

[0032] In relation to figures 1 to 7, we will describe below a virtual flow sensor 1, that is to say a device, implemented by computer, which exploits a mathematical model of a fluidic installation 2 including a pump 3 to estimate the estimated flow Q generated by the pump 3.

[0033] A fluidic installation 2 according to an exemplary embodiment is illustrated in [Fig. 1]. The installation includes a pump 3, of the rotodynamic type, also referred to as a centrifugal pump, which operates by converting mechanical energy into kinetic energy through a rotating rotor. The fluidic installation 2 includes an upstream pipe 4 and a downstream pipe 5 which are respectively arranged upstream and downstream of the pump 3. The upstream pipe 4 has an internal diameter, noted Da, while the downstream pipe 5 has an internal diameter, noted Dr.

[0034] The fluidic installation 2 also includes two pressure sensors, namely an upstream pressure sensor 8 and a downstream pressure sensor 9, which are connected to the control system 6. The upstream pressure sensor 8 is positioned at a point A on the upstream pipe 4, at a height Za, and delivers a signal representative of the suction pressure Pa. The downstream pressure sensor 9 is positioned at a point R on the downstream pipe 5, at a height Zr, and delivers a signal representative of the discharge pressure Pr. The downstream pipe 5 is also equipped with a variable opening valve 10.

[0035] Pump 3 and valve 10 are controlled by a control system 6. The control system 6 is configured, in particular, to control the pump frequency, i.e., to regulate the rotational speed N of the pump rotor 3, and to control the valve 10 in order to regulate the flow generated by the pump 3 to a setpoint value. To do this, the control system 6 is connected to a speed sensor 7 that provides a signal representative of the rotational speed N of the pump rotor 3.

[0036] In the illustrated embodiment, the fluidic system 2 also includes a flow meter 14, which, as will be detailed later, provides a measurement of the flow rate Qmeasured by the pump 3. This flow meter 14 can be used, in particular, when parameterizing the model defining the operation of the pump 3, as described below.As detailed further below, the flow meter 14 can either be retained in the operational phase, with the virtual flow sensor 1 then serving to detect and compensate for its malfunctions, or be removed, with the virtual flow sensor 1 then directly providing the estimated flow Q to the control system 6. According to another embodiment not shown, the model can be pre-established, which makes it possible to estimate the flow without going through a learning phase using a flow meter 14.

[0037] The virtual flow sensor 1 is connected to the speed sensor 7, the upstream pressure sensor 8 and the downstream pressure sensor 9, either via the control system 6, as illustrated in [Fig.1], or directly.

[0038] The functional architecture of the virtual flow sensor 1 is shown in the [Fig.2],

[0039] The virtual flow sensor 1 comprises: - a parameterization module for model 11 which aims to define the numerical values ​​of the model parameters defining the operation of pump 3; - a calculation module 12 which performs real-time estimations using input data from the aforementioned sensors 7, 8, 9, the model and an algorithm, such as that of an extended Kalman filter, to estimate the estimated flow rate Q generated by the pump 3; and - a connectivity module 13 which is particularly capable of providing connectivity with a human-machine interface as well as with the control system 6.

[0040] Initially, we will focus on describing the parameterization module of model 11 and the parameterization steps allowing us to specify the numerical values ​​of different parameters that influence the behavior of the model.

[0041] The model uses a Total Dynamic Head equation for pump 3, also referred to hereafter as the TDH equation. The Total Dynamic Head is the expression of a pressure converted into a liquid column height expressed in meters. The TDH equation establishes a relationship between the total dynamic head (in m) and the flow rate Q (in m³ / h) of pump 3 for a given rotational speed N of the pump 3 rotor.

[0042] The total manometric head is given by the following relation: HMT = hP + axtf (equation 1); where: - &p=^+Hg ; - 8 x (— - — - Q is the flow rate of pump 3 (in m3 / h); - Pa is the static suction pressure measured at point A (in Pa); - Pr is the static discharge pressure measured at point R (in Pa); - Hg is the geometric height between point A and point R, that is Za-Zr (in m); - Da is the internal diameter of the upstream pipe 4 (in m); - Dr is the internal diameter of the downstream pipe 5 (in m); - p is the density of the fluid (in kg / m3); and - g is the acceleration due to gravity (in m / s2).

[0043] In the case where the upstream pressure sensor 8 and the downstream pressure sensor 9 are placed at the same height, the equation can be simplified by removing the term Hg.

[0044] The parameterization steps according to one embodiment are represented on the flowchart of [Fig.3].

[0045] In a first step 3.1, the opening of the valve 10 is varied while maintaining the rotation speed N of the pump rotor 3 at a reference value, namely Nref.

[0046] For each degree of opening of valve 10: - The flow rate QmesUré is measured using the flow meter 14, as well as the suction pressure Pa and the average discharge pressure Prau from the upstream and downstream pressure sensors 8, 9; and - the total manometric head is determined by means of the aforementioned equation 1 as a function of the suction pressure Pa, the discharge pressure Pr, the volumetric flow rate QmeSuré of the internal diameters Da and Dr of the upstream and downstream pipes 4, 5 and, optionally, of the heights Za and Zr if they are not identical.

[0047] Thus, for each opening position of the valve 10, a pair of values ​​(HMT, Qmeasured) CSt Obtained.

[0048] Then, in a second step 3.2, a polynomial interpolation is carried out in order to identify a polynomial equation HMTref = h(Q) passing through the different pairs of coordinates. The polynomial equation HMTref = h(Q) is advantageously of the second degree and therefore takes the following form: HMT„f = Anl^+ B„fQ + C„f (Equation 2); with Aref, Bref and Cref: the quadratic, linear and constant coefficients of the equation of the curve HMTref = h (Q) corresponding to the reference rotation speed Nref.

[0049] By way of example, [Fig. 4] is a graphical representation of such a reference curve h(Q) of the pump 3 obtained for a rotor rotation speed pump 3 at 1000 revolutions per minute.

[0050] By applying the similarity law for rotodynamic pumps (step 3.3), it is observed that if the speed changes, the total head changes proportionally to the square of the speed ratio. In other words, at a rotational speed N, the total head HMT is related to HMTref corresponding to the rotational speed Nref by the following relation: HMT = HMTref x ^pEquatlon3)-

[0051] Thus, in step 3.4, the following pump operating equation is deduced: AP^ (Q, N) =(a - ajx^+BxQ + C (Operating equation) with : "AP = + Hg with Hg = Za-Zr (in m) (or, in simplified terms, AP = when the upstream pressure sensor 8 and the downstream pressure sensor 9 are positioned at the same height); A- Arefwith a = x (“r - —ï) ' B = Brief^^; N: the rotational speed of the pump rotor.

[0052] This operating equation is subsequently (Step 3.4) transformed into a state vector X(t). The state vector represents the unknown quantities to be estimated. It takes the following form: A Pestilé (01; where Question (t) - A Pestilest(t): an estimate of AP at time t; and - Qestitné (?) : an estimate of the flow rate Q at time t.

[0053] Furthermore, the measurement vector represents what is observed and takes the following form: Y(i) = [AP measure)] ; with - APmesure (t); Pr;?.®. . (or, in simplified terms, _ Pr^L. if Za=Zr), at at time t.

[0054] According to one embodiment, the parameterization can advantageously be implemented by means of a human-machine interface allowing a user: - on the one hand, to enter the static parameters, namely the internal diameters Da and Dr of the upstream pipe 4 and the downstream pipe 5 as well as where applicable the heights Za and Z, of the upstream pressure sensor 8 and the downstream pressure sensor 9; and - on the other hand, to connect the virtual flow sensor 1, via the connectivity module 13, to the control system 6 and to assign a dynamic variable of the control system 6 to each input parameter of the virtual flow sensor 1, namely: static suction pressure Pa, static discharge pressure Pr and rotation speed N of the pump rotor 3.

[0055] The human-machine interface may in particular take the form of a web page dedicated to the configuration of the virtual flow sensor 1.

[0056] In relation to Figures 5 and 6, the calculation steps implemented by the calculation module to estimate the estimated flow rate Q generated by pump 3 will now be described.

[0057] The calculation module 12 includes an estimator which estimates the state variable, namely the estimated flow rate Q generated by the pump 3, as a function of the following dynamic parameters: - the static suction pressure Pa; - the static discharge pressure Pr; - the rotational speed N of the pump rotor 3; the following static parameters: - the model M including in particular the parameters Aref, Bref, Cref and Nref of the polynomial equation HMTref = h (Q); and - the internal diameters Da and Dr of the upstream pipe 4 and the downstream pipe 5 as well as, where applicable, the heights Za and Zr of the upstream pressure sensor 8 and the downstream pressure sensor 9.

[0058] According to an advantageous embodiment, the estimator of the calculation module 12 is an extended Kalman filter, also known by the acronym EKF. This type of estimator has the advantage of being able to operate with nonlinear systems, as is the case here. Furthermore, it allows for real-time flow rate estimation, which facilitates its use for control and command functions.

[0059] The calculation module 12 implements the steps described below and represented schematically on the flowchart of [Fig.6].

[0060] Before the virtual flow sensor 1 is put into service, the calculation module 12 performs an initialization step 6.1. During this step, the calculation module 12 initializes, that is to say assigns a value to its static parameters and especially to the estimated flow value as well as to the internal gain of the extended Kalman filter.

[0061] Subsequently, at each instant t, the calculation module 12 implements an acquisition step - step 6.2 - which consists of reading the measurement vector, which contains APmesUré obtained at time t from the pressure values: Pa at the suction and Pr at the discharge.

[0062] Still in calculation module 12, the error E is then determined (step 6.3) by comparing measured AP and estimated AP.

[0063] The calculation module 12 then proceeds to an estimation step (step 6.4) by correcting Qestimated and APestimated. This estimation is based on the predicted values ​​at time t, while taking into account the error E, which was calculated during step 6.3, and weighting it by the gain of the Kalman filter at time t.

[0064] The estimated value of the estimated flow rate Qestimated is then delivered (Step 6.5).

[0065] Finally, during a prediction step (Step 6.6), the calculation module 12 prepares the calculation operations for time t+1 by making predictions on the estimated pressure AP(t+1) and estimated flow rate Q(t+1) values, based on the model. The gain of the extended Kalman filter is also updated, taking into account the covariance matrices of the measurement noise and the mathematical model of the pump 3.

[0066] In relation to [Fig.7], the connectivity module 13 will be described below.

[0067] The connectivity module 13 includes various applications that allow connect the virtual flow sensor 1 to different types of PI to Pn control systems, thus ensuring broad interoperability with various protocols of communication used in industrial networks (such as ADS, Profinet, OPC UA, IO-Link, etc.).

[0068] To achieve this interoperability, a software gateway 15 is used to support a large number of industrial PI, Pn protocols.

[0069] According to one embodiment, the CodeSys application is used to create this software gateway 15, thus enabling the computational module 12 to interface, via the TCP / IP protocol, with other industrial protocols PI, Pn. The TCP / IP protocol is advantageous because of its speed and ease of integration with the Python application, which is advantageously used to code the computational module 12.

[0070] The virtual flow sensor 1 can be implemented in various forms, either unitarily or in a distributed manner, using hardware and / or software components. Potential hardware components include specific ASIC integrated circuits, FPGA programmable logic arrays, and microprocessors. The software components can be developed in various programming languages, such as Python, C, C++, Java, or VHDL, although this list is not exhaustive. It should be noted that the coding of the computing module 12 is advantageously done in Python, a language that offers great ease of deployment and allows for performance optimization. Thanks to this approach, the system's response time is significantly reduced, approaching the millisecond range, which is crucial for real-time applications.

[0071] The virtual flow sensor 1 can have many applications.

[0072] According to one embodiment, the virtual flow sensor 1 provides the estimated flow rate Qmeasured to the control system 6, which is configured to control the pump 3 based on said estimated flow rate Qmeasured in order to control it to a setpoint value. The virtual flow sensor 1 thus makes it possible to temporarily replace a flow meter 14, particularly in the event of the latter's failure, in order to prevent any degradation in the performance of the pump 3. The virtual flow sensor 1 can also permanently replace a flow meter, thereby reducing the cost of the fluid system installation.

[0073] According to another complementary or alternative embodiment, the virtual flow sensor 1 delivers the estimated flow rate Qestimated to a diagnostic unit which compares the estimated flow rate Qestimated delivered by said virtual flow sensor 1 to a measured flow rate Qmeasured delivered by one or more flow meters 14. The diagnostic unit can further be configured to, in the event of detection of a discrepancy between the estimated flow rate Qestimated and the measured flow rate Qmeasured that exceeds a threshold: - issue an alert signal; and / or - switch to pump control mode 3 based on the estimated flow rate Qestimated, replacing the Qmeasured measurement provided by the flow meter 14.

[0074] Although the invention has been described in connection with several particular embodiments, it is clear that it is by no means limited to them and that it includes all technical equivalents of the means described as well as combinations thereof if these fall within the scope of the invention, as defined by the claims

[0075] The use of the verb "comprise", "comprendre" or "include" and its conjugated forms does not exclude the presence of other elements or other steps than those stated in a claim.

[0076] In the claims, any reference sign in parentheses shall not be interpreted as a limitation of the claim.

Claims

Demands

1. Method for estimating an estimated flow rate Q generated by a pump (3) having a rotor; said method comprising: - measuring a static suction pressure Pa at a point A located upstream of the pump (3), a static discharge pressure Pr at a point R located downstream of the pump (3) and a rotational speed N of the rotor of the pump (3); and - estimating the estimated flow rate Q at least as a function of the variables Pa, Pr and N.

2. Estimation method according to claim 1, wherein the estimated flow rate Q is estimated by means of an estimator which uses a model based on a polynomial equation of the type AP = f (Q, N}; with: AP: a variable representing the pressure difference Pr-Pa; Q: the flow rate; and N: the rotational speed of the pump rotor (3).

3. Estimation method according to claim 2, wherein point A is located at a height Za and point R at a height Z, and in which: A p = + Za - Zr with p: the density of the fluid and g: the acceleration due to gravity.

4. Estimation method according to claim 2 or 3, wherein the polynomial equation AP = f (Q, N) is as follows: AP = (A- "Jx^ + Bxg + C; with: A = Aref ; " - r "v -A... ; " - "ref x Nfef ' n =___8___ / J__LA ; gwZ^ôOO2 \ Dr4 Du / g : the acceleration due to gravity (in m / s2); Da : an internal diameter of an upstream pipe at point A; Dr : an internal diameter of a downstream pipe at point R; Nref : a rotational speed of the pump rotor (3) for which a polynomial equation HMTref= h (Q) establishes a relationship between a total head HMTref and a flow rate Q at said rotational speed; Aref, Bref, and Cref: the quadratic, linear, and constant coefficients of the polynomial equation HMTref = h(Q) = Aref + BrefQ + Cref-

5. Estimation method according to claim 4, comprising a preliminary model parameterization step including: - determining several pairs of values ​​corresponding to the total manometric head and the flow rate QmeSuré for a plurality of opening levels of a valve (10) disposed downstream of the pump (3) while the rotor of said pump (3) operates at a constant rotational speed taking the reference value Nref; - performing a polynomial interpolation so as to determine the coefficients Aref, Bref and Cref of the polynomial equation HMTref = h (Q) = Aref *Q2+ Bref XQ + Cref passing through the pairs of values ​​thus determined.

6. Estimation method according to any one of claims 2 to 5, wherein the estimator is an extended Kalman filter.

7. Estimation method according to claim 6, wherein the extended Kalman filter is based on: - a state vector taking the form: A Pestimated(t); where a Pestimated(f): an estimate * L Qestimated(Z) of AP at time t and Qestimated(t); an estimate of the flow rate Q at time t; and - a measurement vector taking the form: Y(t) = [ A Pmeasurement(t)]; with A Pmeasurement(t) = ?LÿL + Za - Zr at time t with p: the density of the fluid and g: the acceleration due to gravity.

8. Method of controlling a pump comprising the step of controlling the pump as a function of the estimated flow rate Q estimated according to an estimation method according to one of claims 1 to 7 in order to control it to a setpoint value.

9. A method for controlling a pump according to claim 8 further comprising: - controlling the pump based on a measured flow rate Q measured by means of a flow meter (14); - comparing the measured flow rate Q to the estimated flow rate Q; - detecting when a difference between the measured flow rate Q and the estimated flow rate Q exceeds a threshold; and

10.

11.

12.

13. - switch from pump control based on the measured flow rate Q to pump control based on the estimated flow rate Q in response to such detection. Device (1) for estimating an estimated flow rate Q generated by a pump (3), said estimating device comprising: - a connectivity module (13) configured to receive a static suction pressure Pa measured upstream of the pump (3), a static discharge pressure Pr measured downstream of the pump (3) and a rotational speed N of the pump rotor (3); and - a calculation module (12) configured to estimate the estimated flow rate Q at least as a function of the variables Pa, Pr and N. Estimation device (1) according to claim 10, wherein the calculation module (12) includes an estimator using a model based on a polynomial equation of the type kP = f(Q, N)', with: AP: a variable representing the pressure difference Pr-Pa; Q: the flow rate; and N: the rotational speed of the pump rotor (3). Estimating device (1) according to claim 11, wherein the estimator is an extended Kalman filter. Fluid installation comprising a pump (3), an upstream pipe (4) and a downstream pipe (5) which are respectively arranged upstream and downstream of the pump (3), an upstream pressure sensor (8) delivering the static suction pressure Pa, a downstream pressure sensor (9) delivering the static discharge pressure Pr, a speed sensor (7) delivering a signal representative of the rotational speed N of the pump rotor (3) and a device (1) according to any one of claims 10 to 12.

Images