Process for operation of a coriolis mass flow rate measurement device

Abstract

A process for operating a Coriolis mass flow rate measurement device which has at least one measurement tube, the measurement tube being excited into vibrations with a predetermined excitation frequency and a predetermined excitation phase. The response phase which is achieved thereby and the rate of change of the response phase are detected and the excitation frequency is changed by the frequency amount which arises by means of a predetermined function from the detected rate of change of the response phase. This makes it possible to maintain continuous measurement of the mass rate of flow even if two-phase flows occur.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 schematically shows excitation of the measurement tube via a controlled generator and

FIG. 2 schematically shows excitation of the measurement tube via direct feedback.

DETAILED DESCRIPTION OF THE INVENTION

Conventional mass flow rate measurement devices which work according to the Coriolis principle are characterized in operation with single-phase flows with respect to their accuracy and reliability. However, this reliability is generally absent in multiphase flows. A multiphase flow is defined as a flow which is composed of at least two phases with physical properties which are different from one another. These phases can consist of the same or of different substances. Phases are thus homogeneous and spatially limited regions of the flow. Examples are liquid-solid flows, gas-liquid flows, gas-solid flows, water-steam flows and water-air flows. In particular, multiphase flows also occur in processes such as filling, emptying, process start-up and switching of valves, and in swelling in the flow.

In applications with multiphase flows, considerable measurement errors typically occur. The major cause of this is the occurrence of asymmetrical filling of the measurement tube which leads to very rapid fluctuations of the resonant frequency of the measurement tube through which flow is taking place. Furthermore, the occurrence and disappearance of secondary flows in the measurement tube result in rapid attenuation, specifically when a secondary flow occurs, or rapid attenuation equalization, specifically when a secondary flow disappears. Fundamentally secondary flows are caused by different densities of the multiphase flows.

Rapid attenuation and attenuation equalization of the measurement tube by the transient flow and by the simultaneous rapid change of the resonant frequency often cause loss of the operating point in conventional, relatively slow control processes compared to the change of flow. Then, the maximum available power is no longer enough to maintain the vibrations of the measurement tube. The result is that Coriolis forces are no longer induced so that the mass flow can no longer be measured either. On the other hand, in the presence of a multiphase flow, energy is pumped into the vibrations of the measurement tube in order to maintain the vibrations of the measurement tube, and thus, the measurability of the mass flow, all the energy in the sudden exposure of the multiphase flow will be used to excite vibrations of the measurement tube at the new resonant frequency so that preventive shutoff is necessary for preventing deformations and damages of the measurement tube. For this reason, it is desirous to have a process for operating a Coriolis mass flow rate measurement device which makes it possible to maintain the operating point even when multiphase flows occur.

The following description of one preferred embodiment of the invention proceeds from a Coriolis mass flow rate measurement device which has a single measurement tube. The vibrations of this measurement tube can be described in a simplified manner by an oscillator for the driving mode with the following physical-mathematical model of the second order (equation 1):

$\begin{array}{cc}{G}_1\left(s\right)=\frac{k_1s}{s^2+2{\omega }_{01}d_1s+{\omega }_{01}^2}& \left(1\right)\end{array}$

This transfer function has three parameters, specifically the spring constant c₁, the vibrating mass m₁ and the attenuation coefficient d₁. The determination equation 2 reads as follows:

$\begin{array}{cc}{k}_1=\frac{1}{m_1},{\omega }_{01}=\sqrt{\frac{c_1}{m_1}}& \left(2\right)\end{array}$

The parameters of the measurement tube m₁ and d₁ are changed very quickly by the multiphase flow. Thus, according to the change of the vibrating mass m₁ and the attenuation coefficient d₁, a highly time-variant system is present. This means that the dynamics of control must keep step with the dynamics of change of the system since otherwise interruption of the measurement of process quantities is inevitable.

The dynamics of control is determined in addition to the mechanical structure of the Coriolis mass flow rate measurement device by the following:

the speed with which the location of the resonant frequency can be considered,

the stability and dynamic behavior of the Coriolis mass flow rate measurement device for given group delay times or dead times of the hardware components used,

efficient use of available manipulated variables and

the extent to which the control algorithms used can be adapted to the mass flow rate measurement device.

With consideration of these aspects, three control methods are determined below, specifically a method of forced excitation, a method with self-contained excitation and a method with hybrid excitation, therefore a combination of forced excitation with self-contained excitation. For reasons of clarity, the model of the measurement device for purposes of control is simplified to such an extent that it can be reproduced by the behavior of the driving mode by the transfer function (equation 1). This means that the other natural forms, such as the Coriolis mode, and couplings between the existing natural forms are not considered in the description of control. This is a good idea since, preferably, the driving mode is the vibration form for which the device is operated in resonance. Thus, this vibration form is determinative for control at the operating point.

In forced excitation, the driving of the measurement tube according to the preferred embodiment of the invention described here is supplied via a controlled generator, as is apparent from FIG. 1. Control of the generator takes place such that the phase between the excitation of the measurement tube which is proportional to the force and its response which is proportional to speed is zero. Moreover, the amplitude of the velocity signal is specified and kept constant.

In conventional applications, it has been assumed for phase control that the frequency of the output signal of the driving mode is equal to the frequency of the excitation signal. This assumption is dropped for this invention since it applies only under the assumption that the measurement tube is in a quasi-steady-state which is generally not present in a multiphase flow. For state transitions, the output signal is composed at least of superposition of the attenuated resonant frequency and induced excitation signal. Therefore, the following (equation 3) applies:

ν₁(t)=A sin(ωt+φ)+B e^−ω01Dtsin(ω₀₁t)   (3)

Here ω is the frequency of the induced excitation, and the parameters A and B are generally a function of the properties of the mass flow rate measurement device and of the fluid which flows through the mass flow rate measurement device. The output signal (equation 3) contains information about the location of the current resonant frequency. It is used in the conventional mass flow rate measurement devices which are operated self-contained and in which there is no forced excitation, but the excitation signal is derived from self-contained vibrations. Thus, in the self-contained excitation, the term A sin (ω+φ) in the output signal is eliminated.

In this case of forced excitation, different processes are used to accelerate phase control. It is common to these processes that the location of the current resonant frequency is considered in the control. This can be done using different processes, of which two different ones are described below.

According to a first preferred process the output signal is demodulated as in a conventional forced excitation. For this purpose, the output signal is multiplied by the known generated excitation signal using equations 4 & 5:

$\begin{array}{cc}\begin{array}{c}{v}_1\left(t\right)\sin \left(\omega t\right)=A\sin \left(\omega t+\varphi \right)\sin \left(\omega t\right)+\\ B{}^{-{\omega }_{01}\mathrm{Dt}}\sin \left({\omega }_{01}t\right)\sin \left(\omega t\right)\\ \frac{A}{2}\left[\cos \left(\varphi \right)+\underset{\underset{\mathrm{Filtered}\mathrm{out}}{}}{\cos \left(2\omega t+\varphi \right)}\right]+\\ \frac{B}{2}{}^{-{\omega }_{01}\mathrm{Dt}}\left[\cos \left({\omega }_{01}-\omega \right)t+\underset{\underset{\mathrm{Filtered}\mathrm{out}}{}}{\cos \left({\omega }_{01}+\omega \right)t}\right]\end{array}& \left(4\right)\\ \begin{array}{c}{v}_1\left(t\right)\cos \left(\omega t\right)=A\sin \left(\omega t+\varphi \right)\cos \left(\omega t\right)+\\ B{}^{-{\omega }_{01}\mathrm{Dt}}\sin \left({\omega }_{01}t\right)\cos \left(\omega t\right)\\ \frac{A}{2}\left[\sin \left(\varphi \right)+\underset{\underset{\mathrm{Filtered}\mathrm{out}}{}}{\sin \left(2\omega t+\varphi \right)}\right]+\\ \frac{B}{2}{}^{-{\omega }_{01}\mathrm{Dt}}\left[\sin \left({\omega }_{01}-\omega \right)t+\underset{\underset{\mathrm{Filtered}\mathrm{out}}{}}{\sin \left({\omega }_{01}+\omega \right)t}\right]\end{array}& \left(5\right)\end{array}$

The double frequency portions in these equations are filtered out using a lowpass filter. The signal portions with the difference of the forced driving frequency and the current resonant frequency are however further processed. The actually computed phase, specifically the output of the phase detector, is computed from the filtered signals in accordance with equation 6:

$\begin{array}{cc}{\phi }_{\mathrm{berachnet}}=\arctan \left[\frac{\frac{A}{2}\cos \left(\varphi \right)+\frac{B}{2}{}^{-{\omega }_{01}\mathrm{Dt}}\cos \left({\omega }_{01}-\omega \right)t}{\frac{A}{2}\sin \left(\varphi \right)+\frac{B}{2}{}^{-{\omega }_{01}\mathrm{Dt}}\sin \left({\omega }_{01}-\omega \right)t}\right]& \left(6\right)\end{array}$

This result can be summarized as follows: In the quasi-steady-state, the computed phase can agree with the actual phase shift by the measurement tube. In the transition from one quasi-steady-state to another quasi-steady-state, the phase migrates with a speed which is a function of the frequency difference between the excitation frequency and the resonant frequency. This important information about the spacing of the excitation frequency and the current resonant frequency is used to increase the speed of phase control. This accelerates the start-up of the new excitation frequency and correction of the phase shift to zero. This prevents the measurement tube vibration from failing.

According to a second preferred process, according to the rate of change of the phase a signal of the function of the difference frequency, of the induced driving frequency and of the current resonant frequency is generated and is supplied as information about the location of the current resonant frequency in the course of control.

The use of parametric excitation, due to the multiphase flow, makes it possible to approach the current resonance point specifically and quickly, before the response of the parametric excitation has decayed. This prevents loss of the optimum operating point and makes a restart and repeated time-consuming search for the working point superfluous. Thus, the rate of mass flow can be measured in the transitions between different quasi-steady-states.

If the measurement tube is excited at its resonant frequency, the excitation is proportional to the initial velocity of the driving mode. This means that the excitation—viewed quasi-steady-state—is used to cover the losses which are proportional to speed. These losses are dependent on the material properties and on the composition of the flow. In order to keep the desired amplitude of the velocity signal constant regardless of noise, amplitude control as described below is performed.

The amplitude is measured very quickly, specifically, within less than one half period. The measurement is taken according to the above described preferred embodiment of the invention using the process of absolute maxima seeking or using the gradient process. In the former method, the highest quantitative sampled value within a half period is established, preferably recursively. In the second process, the maximum value at the quantitative minimum of the gradient of the velocity signal is determined. A weighted combination of the determinations of the two processes can likewise be used.

Furthermore, the manipulated variable of amplitude control is changed such that the pulses of the flowing mass particles do not change suddenly. This reduces the flow disturbance for purposes of measurement and increases the efficiency of the driving power or braking power since a driving signal which is as cleanly sinusoidal as possible with a small harmonic portion is used. To achieve this, the amplitude of the driving signal is preferably changed at, but at least in the vicinity of, the passages of the operating signal through zero.

In a self-contained excitation, it is assumed that a self-contained vibration on the output signal has been established which is generally based on a wideband additive, or during operation on a multiplicative, i.e., parametric, excitation. The output signal based on the filter action of the mechanism with a band-pass property preferably contains signals of the frequency of the resonance point and can be described by the following equation 7:

ν₁(t)=Ae^−ω01Dtsin(ω₀₁t)   (7)

Parameters A and D are generally a function of the properties of the Coriolis mass flow rate measurement device.

The output signal according to equation 7 contains information about the location of the current resonant frequency ω₀₁. This signal is used in conventional mass flow rate measurement devices which are operated, self-contained, for excitation of the measurement tube. This means that there is no forced vibration, but that the excitation signal is derived preferably from self-contained vibrations, as can be taken, for example, from FIG. 2. In another preferred implementation, noise which is superimposed from self-contained vibrations are additionally suppressed with a filter with special properties, such as bandpass properties.

Thus in self-contained excitation a phase control circuit is eliminated. This is advantageous in that the location of the resonant frequency for forced excitation can be promptly supplied to the control. However, the disadvantage is that the resonant frequency must subsequently be exactly determined for purposes of measuring the density and the flow rate. The main disadvantage ultimately lies in that the location of the resonant frequency can be lost as a result of noise, for example, due to multiphase flows. This can occur when the attenuation rises quickly and all the energy stored in the oscillator is quickly dissipated such that the output signal decays almost aperiodically in the least favorable case. In this case, since there is no forced excitation for identification of the resonant frequency, the system lacks any information about the resonant frequency, and the Coriolis mass flow rate measurement device cannot set the operating point so that the flow rate cannot be measured. In conventional Coriolis mass flow rate measurement devices, then, restart with forced excitation during the starting phase is necessary; and this takes some time. This procedure with rapid changes of the flow conditions cannot be safely used in flows with transient phases. Two methods are used here to solve this problem:

The first process uses high-speed amplitude control, as described above. The amplitude controller can be various controller types, such as PI, PID-VZ1, etc. In this connection, preferably, a digital compensation controller is used which compensates for the attenuation losses of the driving mode. Preferably, the following control law 8 is used:

u _k =u _k−1 +p(ω_k−y_k)+y_k−2+y_k−  (8)

u being the output of the controller, p being the gain, ω the setpoint of the amplitude and y being the actual value of the amplitude.

The second process which can be used as an alternative is hybrid control which uses a combination of the two controls with forced excitation or with self-contained excitation. Thus, it is ensured that the advantages of the two processes can be jointly used. In one preferred implementation, depending on the frequency difference between the forced and the current resonant frequency, detected as described above in conjunction with phase control, part of the output signal is directly fed back.

The time behavior of the Coriolis mass flow rate measurement device is influenced by two parameters among others, specifically, the sample rate and the group delay times. i.e., the dead times in the hardware components. Studies of the behavior of the phase and amplitude controller show that the control behavior depends on the sampling rate, the phase control circuit being more sensitive than the amplitude control circuit. The higher the sampling frequency, the shorter the correction time. Starting with sampling rates of 8 kHz, the improvement flattens out so that at roughly 100 kHz a significant improvement can no longer be ascertained.

If the transfer function of the driving mode is examined simplified, i.e., without considering the coupling of the other modes, as a linear, second order system, and the dead times, for example, of the A/D converter and D/A converter, are considered, with T_d, the transfer function 9 is:

$\begin{array}{cc}{G}_1\left(s\right)=\frac{k_1s}{s^2+2{\omega }_{01}d_1s+{\omega }_{01}^2}{}^{-T_dt}& \left(9\right)\end{array}$

Without the dead time, the controlled system is stable for all proportional feedbacks. However, if dead times are allowed, the dead times influence the amplitude reserve and phase reserve and thus the dynamic behavior of the control circuit, so that, depending on the parameters of the transfer function, unstable controlled systems can arise. The stability of closed controlled systems is dependent on the parameters and the rate of their change. In a study of high speed parameter changes which always occur in two-phase flows, the set phase reserve and amplitude reserve can be reduced very quickly so that a tendency to vibration occurs.

In Coriolis mass flow rate measurement devices, there are two nonlinear control circuits which are very strongly coupled to one another, with dynamic time behavior which is not analytically present. Therefore, for initial orientation, the effect of the dead time on the two control circuits is experimentally studied. For this purpose, dead time elements with variable dead times were installed upstream from the D/A converter or downstream from the A/D converter. In the studies, the dead time was varied, and various unit step responses were recorded. As a result, it can be ascertained that, for a given dead time which has been caused, for example, by the D/A converter or the A/D converter, it can be prolonged and adapted preferably on the software side. Thus, in accordance with the invention, the stability and the dynamic behavior of the control circuits can be improved.

As described above, a careful examination of the group delay time or dead time is necessary in the control circuits for system stability. Especially in the case of parameter changes as occur, for example, in two-phase flows, it must be ensured that the dead times together with the variable system parameters lead to stable control circuits. Furthermore, it is important to use a dead time as small as possible so that prompt control—especially prompt correction of faults—can be implemented.

For robust stability, i.e., a large phase and amplitude reserve, it is important to choose and influence the dead time in a suitable manner. This can take place, for example, by the following measures:

selection of A/D and D/A converters with a group delay time as small as possible,

use of separate A/D and D/A converters for control, on the one hand, and measurement of flow properties, on the other, or use of A/D and D/A converters with separate filter stages or filters with internal taps for separate signals with different time constants for control, on the one hand, and measurement of flow properties, on the other,

use of a sampling rate as high as possible in order to obtain a dead time as small as possible at a given group delay time of digital FIR filters of A/D and D/A converters, and

use of buffering in software, digital filters, and computerized phase rotation in the software in order to specifically set a certain dead time which leads to stable, and moreover, fast system behavior.

The resulting dead time which arises from the dead time which is fixed by the hardware and the software-dictated dead time which is additionally introduced, can then be coupled permanently to the operating point. This means that the additional dead time is dynamically adapted to the vibration period. Thus, the resulting dead time can be set, for example, to a multiple of half the duration of the vibration period.

Claims

1-10. (canceled)

11. Process for operating a Coriolis mass flow rate measurement device which has at least one measurement tube, comprising the steps of: exciting the measurement tube into vibrations with a predetermined excitation frequency and a predetermined excitation phase,detecting a response phase and a rate of change of a response phase resulting from excitation of the measurement tube, andchanging the excitation frequency by a frequency amount which is a predetermined function of the detected rate of change of the response phase.

12. Process as claimed in claim 11, further comprising the step of setting a phase shift between the excitation phase and the response phase to a predetermined phase value.

13. Process as claimed in claim 12, wherein the predetermined phase value is zero.

14. Process for operating a Coriolis mass flow rate measurement device which has at least one measurement tube, comprising the steps of: exciting the measurement tube to vibrations by means of an excitation signal so as to produce a vibration signal,detecting the vibration signal produced and subjecting the vibration signal to A/D conversion, andusing the detected vibration signal to determine at least one parameter of a medium which is flowing through the measurement tube and at least one parameter of vibration excitation,wherein A/D conversion of the vibration signal for determining the parameter of the medium flowing through the measurement tube is performed independent of the A/D conversion of the vibration signal performed for establishing the parameter of the vibration excitation.

15. Process as claimed in claim 14, wherein a time constant of A/D conversion of the vibration signal for determining the parameter of the medium which is flowing through the measurement tube is different from a time constant of A/D conversion of the vibration signal for establishing the parameter of the vibration excitation.

16. Process as claimed in claim 15, wherein a cutoff frequency of A/D conversion of the vibration signal for establishing the parameter of the vibration excitation is greater than a cut-off frequency of A/D conversion of the vibration signal for determining the parameter of the medium flowing through the measurement tube.

17. Process as claimed in claim 16, wherein two A/D converters are used which are one of different from one another, A/D converters with separate filter stages and A/D converters with inner taps for separate signals.

18. Process as claimed in claim 14, wherein a cutoff frequency of A/D conversion of the vibration signal for establishing the parameter of the vibration excitation is greater than a cut-off frequency of A/D conversion of the vibration signal for determining the parameter of the medium flowing through the measurement tube.

19. Process as claimed in claim 14, wherein two A/D converters are used which are one of different from one another, A/D converters with separate filter stages and A/D converters with inner taps for separate signals.

20. Process for operating a Coriolis mass flow rate measurement device which has at least one measurement tube, comprising the steps of: exciting the measurement tube to vibrations by means of at least analog excitation signal which is periodic in time intervals,detecting an analog response signal produced by excitation of the measurement tube, andsubjected the analog response signal to A/D conversion by means of an A/ID converter so that a digital response signal is produced,using the digital response signal for determining a digital excitation signal, andsubjecting the digital excitation signal to D/A conversion so that an analog excitation signal is formed,wherein a predetermined additional dead time is added to respective hardware-specific dead time of one of the A/D converter and the D/A converter.

21. Process as claimed in claim 20, wherein at least one dead time is set to corresponds to an integral multiple of half the period of the excitation signal.

22. Process for operating a Coriolis mass flow rate measurement device which has at least one measurement tube, comprising the steps of: exciting the measurement tube to vibrations by means of a periodic measurement signal,detecting a vibration signal resulting from said exciting of the measurement tube, andsetting the amplitude of the vibration signal to a predetermined value as a manipulated variable by means of the amplitude of an excitation signal,wherein the amplitude of the excitation signal is set at zero passages of the excitation signal.