JPH02286899A - Stall predicting and preventing device for turbo machine - Google Patents

Abstract

PURPOSE:To avoid rushing into the revolutional stall condition by extracting continuous output signals pertaining to a change in the pressure or flow velocity sensed by a sensor, subjecting them to statistic processing one after another, and allowing a controller having an analyzer to perform time shift processing. CONSTITUTION:A controller 10 extracts a pressure sensing signal generated by a pressure sensor 7, which responds fully to high-frequency pressure variation resulting from passage of each rotary vane 5, and is equipped with a data input part 11 to convert the anlog signals as result from said extraction into digital signals and also with an analyzer 12 to perform statistic processing of changes in the fluid pressure one after another which are caused by rotation of the impeller 4. The data taken into the statistics are subjected to time shift processing according to an extract command, and the result is stored in a memory of the controller 10 as the latest information. Thus the difference between the statistic quantity kept with updating from time to time and the set threshold value will be lessened, and when it is sensed that the statistic quantity has exceeded the threshold value, the controller 10 emits a signal for output correction of the turbo-machine 1 concerned 1.

Description

[Detailed Description of the Invention] (Industrial Application Field) The present invention relates to a machine that is classified as a pump, blower, compressor, etc. for pumping fluid, and is divided into centrifugal type, axial flow type, mixed flow type, etc. related to machines collectively called turbomachinery.
The present invention relates to a system for predicting the occurrence of a rotational stall in advance in order to avoid a rotational stall condition that causes severe fluctuations in fluid flow rate and pressure in the turbomachine.

(Prior Art) First, regarding the technical background of the present invention, in turbomachinery, a stall phenomenon tends to occur when the flow rate of fluid supplied under pressure is reduced artificially or for some other reason. It should be understood that the stalled fluid in a portion of the wheel is approximately 3 times the rotational speed of the wheel.
The phenomenon of rotating at an angular velocity of 0% to 70% is called a rotating stall state.

This rotational stall causes surging, pulsation, and cavitation, and can also cause serious equipment problems such as vibration, noise, and fatigue failure of mechanical parts, which can cause great industrial damage. Become.

However, although there are conventional methods for detecting the turning stall condition, there is no known method for predicting the signs of the stall condition before it occurs.

Therefore, conventionally, by performing a test run of the equipment in advance,
The operating limit point at which the equipment enters a rotating stall state is set, the operating range of the equipment is defined, and the equipment is operated based on this.

Conventionally, in equipment equipped with turbomachinery, if a significant change in fluid behavior occurs in either the turbomachine or the attached pressurized fluid circuit due to long-term use or external causes, it will naturally occur. The operational limit point at which the vehicle enters the rotating stall state may change, and at some point the vehicle may suddenly enter the rotating stall state.

In addition, if the above-mentioned turning stall condition occurs, it is not easy to avoid it because the stall characteristics have hysteresis characteristics, and it may take a considerable amount of time to complete the avoidance measures, and serious trouble may occur during that time. There is a risk of triggering.

(Problem to be solved by the invention) When the turning stall phenomenon has progressed considerably,
It can be roughly detected by the degree of pulsation of the flow of pressurized fluid, vibration of the machine, noise generated, etc., but it is possible to detect a stall phenomenon that has developed in this way at an early stage and to detect a stall in a bad state. Even if a situation close to 41↑ is avoided, it takes a considerable amount of time to avoid this stall state, and there are many cases where it is impossible to take measures against the accident. This can be understood through the following experiment aimed at investigating the above-mentioned rotating stall phenomenon and the fluid action involved in the stall condition.

In order to detect the characteristics of the output waveform related to the above-mentioned rotating stall, we inserted hot-wire anemometers, pressure sensors, etc. at appropriate locations in the circuits of machines and equipment, and investigated the fluid action within one machine. Figure 2 shows the data, and the output waveform was recorded over time. The outputs in the first and third rows in the same figure are waveforms related to the frequency generated by the blade passage of the blade number x rotation speed that occurs as the blade rotates. 2 in the same figure
Rows and rows 4 are the outputs of rows 1 and 3, respectively, which are recorded by applying a low band filter of blade passing frequency x l/2 to remove the blade passing frequency component. The scale corresponds to one rotation of the impeller. Since the above-mentioned blade passing frequency is eliminated in the output waveform after applying this low-band filter,
As a phenomenon of rotating stall, disturbances in fluid pressure (or flow velocity) clearly appear as shown in the second and fourth rows of the figure. The small disturbance that can be seen in the second column of the figure indicates the initial state of the rotating stall phenomenon, and the disturbance in the fourth column indicates the fluid pressure that shifts to a waveform specific to a rotating stall as the stall phenomenon develops. This shows the phenomenon of

As is clear from the diagram of the experimental results above, the time required for the development process from the initial occurrence of the rotating stall phenomenon to the starting point of the stall condition is instantaneous, corresponding to only a few revolutions in terms of the amount of rotation of the impeller. be. Therefore, even if a stall phenomenon or stall condition is detected at a late stage after the initial point of the turning stall phenomenon described above, and this is used as a signal to implement stall avoidance measures or RV control, due to the operational characteristics of this machine, the control will not be effective. Furthermore, due to the inertia of the fluid flow, it is impossible to achieve the desired stall avoidance, and once this turbomachine enters a stall state, the hysteresis characteristic The stall condition cannot be resolved unless measures are taken to significantly increase the flow rate above the limit point at which the stall condition occurs.

As is clear from what has been said above, in order to reliably avoid the rotating stall state of the turbomachinery, it is necessary to predict the symptoms at least at an extremely early stage before entering the above-mentioned rotating stall state, and to promptly measures and controls must be taken to avoid the stall condition, while
The fluid flow in the turbomachine basically has complex turbulence, and it can be said that it is impossible to predict the stall phenomenon from the instantaneous output waveform in this turbulent fluid action.

In view of the above problems to be solved, the present invention predicts the upcoming transition to a rotating stall state when a sign of a rotating stall occurs during operation of a turbomachine, and promptly implements control related to the countermeasure. This was done for the purpose of avoiding entering a turning stall state.

(Means for Solving the Problems) This invention provides a pressure sensor (or hot wire current meter) installed in the casing of a blade wheel in order to investigate temporal changes in fluid action, which is one of the requirements for predicting a rotating stall state. Equipped on the inner surface of the The controller is equipped with a data input section that inputs the pressure-sensitive signal of the pressure sensor, and an analyzer that can sequentially calculate statistics as the pressure-sensitive signal is supplemented. The controller synchronizes the waveform of the pressure fluctuation (or flow velocity change) of the input pressure-sensitive signal with the rotation of the impeller, regularly extracts it, converts it into a digital quantity, and converts it into a digital quantity at a fixed point relative to the impeller. The controller has a function to calculate statistics such as the average value of the pressure-sensitive signal trace, the skewness, and the cross-correlation of the pressure-sensitive signal Δ between two points. It has a function to import the above-mentioned statistics, a function to recalculate the above-mentioned statistics, and a method for using time movement processing to gradually update the information.

In the above configuration, it can be said that the cross-correlation value is preferable as a statistic for finding signs of turning stall.
In order to configure such a system, the analyzer is required to have a high level of analytical ability, so even if it is possible to predict the above-mentioned symptoms by using statistics such as variance and skewness, which are technically easy to analyze, as information. Furthermore, if the amount of population data for calculating statistics is too large, the temporal sensitivity of the target prediction will become dull, and if it is too small, the accuracy of the prediction will tend to deteriorate. Considering this point, it can be said that the desirable amount of data is about 50 to 500.

The feature of this invention is that, with the above-described configuration, when the flow rate of the turbomachine reaches near the initial point of rotational stall, an abnormality occurs in the statistics of the pressure data in the turbomachine, which is a sign of the above-mentioned disturbance. However, the occurrence of a turning stall condition can be predicted by gradually monitoring these signs, that is, the above-mentioned statistics.

Another feature of the present invention is that, as a setting condition for the statistical value that is the standard of the detected pressure data, a turning stall is determined by exceeding a threshold value for a continuously changing phenomenon. A reference value is set for determining whether a symptom has occurred, and the signal transmitted as a result of determination based on the correlation value of statistics based on the threshold value is used as a control means to control the fluid pressure or By controlling the flow rate, the purpose is to prevent the occurrence of the above-mentioned rotating stall condition.

Another feature of the present invention is that a pressure sensor is embedded at a suitable location on the hub surface or casing surface of a turbomachine so as not to obstruct the flow of fluid, and the pressure-sensitive signal of the sensor is transmitted to a dedicated analyzer or a general-purpose computer such as a minicomputer or microcomputer. The information is periodically (regularly) and appropriately extracted according to the instructions of the analyzer (hereinafter simply referred to as the analyzer), and this information is AD converted and sequentially imported into the analyzer, and the latest information is used to gradually add, square, etc. It relates to a structure that calculates statistical quantities such as cross-correlation, variance, and skewness by performing addition, cross-product addition, cube addition, etc., and stores and displays these values. (See Figure 1.) Furthermore, as shown in the conceptual diagram of Figure 3, the present invention gradually updates the stored statistics according to the operation of the turbomachine (this procedure is called time movement processing), and updates the latest statistics. The present invention is characterized in that the comparison between the amount and the threshold value is used as a means to monitor the occurrence of a rotating stall phenomenon in the turbomachine, and is also used as a means for transmitting a control signal in the event of an emergency.

(Example) Hereinafter, the configuration of the present invention will be specifically described based on the example shown in FIG.

The turbomachine 1 has a blade wheel 4 which is supported by a hub 3 at the center of a casing 2 and rotates, and a suitable number of blades 5 are provided on this blade wheel. The aforementioned impeller is connected to a motor 6 and its operation is controlled. A pressure sensor 7 is provided on the wall surface of the casing 2, which can sufficiently respond to high-frequency pressure fluctuations caused by passage of the rotor blade. A controller IO is connected to this pressure sensor 7. This controller 1O includes a data input unit 11 that extracts the pressure-sensitive signal of the pressure sensor and converts the analog signal into a digital signal, and a data input unit 11 that gradually and statistically calculates changes in fluid pressure accompanying rotation of the impeller. It has an analyzer 12 for processing. The data incorporated into the statistics is subjected to time movement processing according to an extraction command, and is input and held as the latest information in the storage unit (memory) of the controller IO.

Then, when the difference between the gradually updated and held statistics and the set threshold value becomes smaller, and when it is detected that the statistics exceed the threshold value, the controller 10 is configured to transmit a signal for output correction.

(Operation of the Invention) The present invention implemented based on the above configuration operates as shown in the following record of experimental results.

Data extraction and statistics are gradually processed as time u11 moves and data is supplemented. Figure 3 shows its progress, with data indicated by horizontal lines.

Each of (2) and (2) conceptually represents the process of sequentially reading and calculating the statistics online, allowing you to understand the effects involved in the 1-hour movement process.

Figure 4 shows the results of investigating the time correlation of the pressure provided on the casing surface of the impeller with regard to statistics.

In Figure 4, the horizontal axis is synchronized with the rotation of the impeller, and the time taken by the two blades is divided into 50 equal parts, and the time is set, and the vertical axis is the point corresponding to the 50 equal parts of the horizontal axis. It is a drawing showing cross-correlation coefficients with data groups at other points based on a pressure data group.

In this figure, φ is the flow coefficient; φ=0.500 (upper figure) corresponds to normal operation, and φ=0.324 (lower figure) corresponds to a flow rate approaching the starting point of a rotating stall.

In the figure, =2.0 is the gap between the blade and the casing (unit: 1), and EZZT and ITB are symbols written to identify the insertion position of the pressure sensor and the data extraction method.

As can be seen in Figure 4 above, during normal operation, similar correlation values appear each time one blade passes, and in the relative flow field with respect to the blade, there is a strong correlation between the flow phenomena between adjacent blades. It can be seen that just before a rotating stall, the correlation between the flow between the adjacent blades becomes smaller.Using this phenomenon, data is extracted at intervals of fJJ] when the blade moves by one pitch, and By monitoring the cross-correlation values of the data group, it can be determined that the turning stall is approaching. Based on the same principle as above, in order to detect the rotating stall phenomenon in the day area user of a turbomachinery, the fluid action (pressure) at two points separated by one guide vane pitch is required.
It is useful to examine the spatial correlation of

The t55 diagram examines the cross-correlation of data groups before and after passing one blade pitch with respect to the flow coefficient, and it can be seen from this diagram that the correlation value decreases rapidly as the flow rate approaches the stall state. In addition, in Fig. 5, the flow coefficient is plotted on the horizontal axis,
It shows the cross-correlation coefficient between the pressure data group at the reference point and the pressure data group after χ has moved by one pitch. It should be noted that the numbers "D" and "II" in the figure indicate the types of impellers.

Figure 6 shows a histogram of the pressure collected at one point whose relative position with the blade is determined by periodic data extraction. It is a drawing showing the distribution of frequency numbers of groups as a histogram centered on the average value. This figure shows a histogram that is nearly symmetrical with respect to the average value during normal operation.
This shows that when approaching the starting point of the turning stall state, the value becomes asymmetrical with respect to the average value. The shape of such a histogram can be easily understood by taking skewness (cubic moment) as a statistical quantity.

Figure 7 shows the skewness of the flow rate coefficient, where the horizontal axis is the flow rate coefficient, the vertical axis is the skewness of the pressure data group at one point synchronized with the blade, and the symbol written at the top indicates the data. This is a symbol used to identify the extraction conditions of . It should be noted that if the variance value is used as a statistical quantity instead of the skewness, it can be seen that the variance value increases rapidly as the stall point approaches.

Figure 8 shows the cross-correlation between a group of pressure data at a reference point synchronized with the rotation of the impeller and a group of data after the blade has moved by one pitch, using time-shift processing. You can understand the changes.

(Effects of the Invention) Due to the functions and effects described above, the present invention has the following effects.

Since the flow of a turbomachine is generally highly turbulent, there is a tendency for abnormal pressure data to occur even during normal operation, but according to the present invention, this influence can be removed by statistical processing. As the turning stall approaches, the percentage of abnormal data mixed in tends to increase, so the statistics change gradually at first, and then suddenly change just before the stall occurs. It is possible to predict this occurrence by setting an appropriate threshold value depending on the use of the turbomachine, which has the effect of avoiding the turning stall.

[Brief explanation of drawings]

Fig. 151 is a simplified diagram showing an embodiment of the present invention, Fig. 2 is a drawing showing the output waveform when the turbomachine stalls in rotation, Fig. 3 is an explanatory diagram of time movement processing, and Fig. 4 is the time taken by the blade to pass. Fig. 5 is a diagram showing the cross-correlation coefficients for the pressure data group corresponding to the reference pressure data group and the pressure data group after moving by one pitch. , Figure 6 is a diagram showing the frequency distribution of the pressure data group as a histogram based on the average value, Figure 7 is a diagram showing the relationship between flow rate and skewness, and Figure 8 is a diagram showing the pressure data group at the reference point and the blade. A diagram showing how the cross-correlation with a data group after moving by one pitch changes with time and flow rate changes. l...Turbo machine 3...E-war hub 5...Blade 7...Pressure sensor 11...Data input section/casing/blade wheel/motor...Controller analyzer 2 ・ ・ 4 ・ ・ 6 ・ ・10 ・ 12... Figure Time data addition Figure Figure Figure

Claims (1)

[Claims] A turbomachine 1, a sensor 7 that is installed on the rotating region of the blade wheel of the turbomachine or on the casing surface or hub surface immediately before it and detects fluid action, and a continuous sensor 7 related to the pressure change or flow velocity change detected by the sensor A controller 10 that includes a data input section that extracts and inputs an output signal, and an analyzer that gradually statistically calculates and processes the input data and performs time movement processing to avoid rotational stall of a turbomachine. Turning stall prediction device for