AU2655900A - Method for controlling charging maneuvers in a hydroelectric installation with a surge tank - Google Patents

Abstract

The invention relates to a method for preventing unacceptable charging maneuvers in hydroelectric plants. Said method makes it possible to determine temporal progression of the level in a surge tank and to compare an existing state with a limit curve in a phase plane on the basis of an actual level (z) and a net inflow ( DELTA Q). Once the limit curve has been reached, the following charge increase is interrupted so that a given minimum or maximum value of the level in the surge tank cannot be exceeded.

Description

Sulzer Hydro AG, CH-8023 Z rich (Switzerland) Method for the controlling of load manoeuvres in a hydroelectric power plant with a surge tank The invention relates to a method for the controlling of load manoeuvres in a hydroelectric power plant with a high-pressure-side and/or a low pressure-side surge tank. Fig. 1 shows a hydroelectric power plant which comprises substantially a reservoir 1, an underwater shaft 2, a surge tank 3 with a section 4 which has a larger cross-section with the base area 5, a partly free-lying conduit 6 and a powerhouse 7. For topographical reasons it is often necessary for the water to be conducted over long distances to the powerhouse 6. The water is conducted from the reservoir 1 through the shaft 2, which is under pressure, and then through the conduit 6 to the powerhouse 7. In the stationary state a level 11 sets in at the reservoir which is designated as the upper water level. In the event of nominal power and stationary flow the pressure falls off as a result of the friction losses along the water conducting lines in accordance with line 12. In load manoeuvres both lower and higher pressures can arise. The upper and lower envelopes 13 and 14 of the possible pressure plots are entered as extreme pressure curves. The extreme pressure curves 15 and 16 result in a plant without a surge tank. If in a plant of this kind the turbines are switched off during operation, then the amount of water flowing in the entire supply line must be braked. In this situation a temporary pressure increase (pressure surge) Neuanm:nderungen:99115 1U -2 arises in the driving water path. On the other hand the acceleration of the water masses during start-up or a load increase on the turbines leads to a drop in pressure in the inflow line. In this a vacuum must not arise in any part of the line. In Fig. 1 this would be the case e.g. in the region of points D and E because the lowest energy heights which occur, which are characterised by line 16, lie substantially below the geodetic height of the line conducting the water. The large pressure variations in the driving water path are difficult to master and in addition substantially hinder the control of the turbines. These problems are largely solved when the surge tank 3 is built as close as possible to the powerhouse. The surge tank fulfils substantially the following tasks: 1. Protection of the shaft from pressure surges which result from the setting movements of the turbines. 2. Reduction of the pressure surge in the pressure line. 3. Improvement of the turbine control. 4. Water storage in a load take-up of the turbines. It is also possible to provide a surge tank in the low pressure part of the plant if a long run-off shaft is present. Surge tanks of this kind behave in a manner which is so to speak mirrored with respect to those in the high pressure part. The pressure rise is reduced during start-up and -3 the depression build-up is reduced when switching off. During operation the surge tank must take up at least the amount of water accumulating when switching off all turbines from full load without overflowing. In Figs. 2a and 2b the processes in a switching off of this kind are illustrated. Reference is made to Figs. 2a and 2b. In Fig. 2a the curve 21 shows the through-flow through the turbines, which is reduced to zero beginning at time point 100 s as a result of a switching off. The through-flow in the pressure shaft 2 is subjected to a delayed braking. The curve 22 shows the plot of the through-flow in the pressure shaft 2 after the switching off of the turbines. The difference 23 of the two through-flow amounts flows into the surge tank 3 and fills up the latter, as is shown by the curve 24 of Fig. 2b. The pressure difference between the surge tank level 24 and the upper water level 11 enables the deceleration of the flow in the pressure shaft 2. In the illustrated case the upswing of the level in the surge tank 3 is reduced by the section 4 above 1284 m in order to avoid an overflowing. In the event of an increase of the through-flow through the turbines, e.g. from idling to full load, processes similar to those in Figs. 2a and 2b occur, but with reversed sign however. An example is shown in Figs. 3a and 3b. In this example the lower limitation of the surge tank shaft (line 26) is not reached. A surge tank should not become empty when all turbines are loaded at the same time. This can by no means be fulfilled in all plants. cjr q, -4 The above mentioned overflowing or becoming empty of the surge tank is based on load variations which result from a stationary operating state of the hydraulic system. If more than one load change is taken into consideration, then the output conditions of a manoeuvre can be substantially less favourable and the fulfilment of the safety requirements becomes considerably more difficult. Every surge tank 3 forms an oscillatory system together with the pressure shaft 2 which is to be protected. Thus one recognises in Fig. 2b that at the time 280 s the water level in the surge tank 3 achieves a maximum value, with a simultaneous standstill of the water masses in the pressure shaft. The pressure difference between the surge tank level and the upper water level continues to act and effects a reversal of the flow in the pressure shaft 2 and the emptying of the surge tank 3 until the reversed flow also comes to a standstill, and so forth. The load variations, i.e. the variations of the volume flow Q of the turbines, thus act in this oscillatory system as disturbances which lead to oscillations of the level in the surge tank. The oscillations can be further exacerbated by an unfavourable time sequence of load changes. As a result, the surge tank can become empty or overflow in spite of correct dimensioning. Figs. 4a and 4b show an example. In this case, after a load release which is triggered at the time point 100 s, the turbines are again to be fully loaded at 450 s. In this case the level in the surge tank would sink to 1206 m above sea level and in the process drop below the lower limit 26 of the surge tank by 20 m. This means that the surge tank is completely emptied and a large amount of air z -5 enters into the driving water path, which can not be permitted. A number of measures for avoiding this problem are known from the prior art: - to provide a surge tank with a sufficient constructional volume. In the extreme case plants can be built which are adequate for any desired load sequence. Through the use of special constructional kinds of surge tanks, e.g. throttled surge tank, surge chamber or differential surge tank the constructional volume can be somewhat reduced. This solution is characterised by high construction costs; - restriction of the regulation stroke of the turbine depending on the number of turbines in operation; - sufficiently slow change of load, i.e. setting times on the order of magnitude of the oscillatory period of the level in the surge tank, e.g. several minutes. This solution is as a rule not suitable for controlled power plants; - the observance of waiting periods, e.g. between load release and a renewed loading of the turbines, or between start-up of the first, second, third turbine and further turbines. In this case waiting times of a similar order of magnitude likewise result in order to allow the oscillation in the surge tank to damp out. This rigid scheme proves to be disadvantageous because in some cases waiting times must be observed although there is no risk for a load manoeuvre; -6 - taking into account the oscillatory phase. A loading is permitted only in certain phases of the oscillation in the surge tank. Disadvantageous is the fact that this method does not fully exploit the dynamic possibilities of the plant; - taking the temporal gradient of the level in the surge tank into account. If the latter exceeds a specific threshold value, then the opening of the regulation members of the turbine are blocked. In this method as well, the dynamic possibilities are not completely exploited; - on-line pre-calculation of worst-case circumstances in a control computer and corresponding limitation of the desired load value. Disadvantageous is the very great cost and complexity. The object of the invention is to improve a method of the initially named kind. This object is satisfied in accordance with the invention by the features of claim 1. The advantage of the invention is to be seen in that it exploits the regulatory potential of the plant as far as possible at large load changes in that the permissible surge tank level is achieved in the shortest possible time; the desired load change is also thereby realised in the shortest possible time. The invention will be explained in the following with the help of the '71 -7 accompanying drawings. Shown are: Fig. 1 a schematic illustration of a hydroelectric power plant; Fig. 2a a diagram of the through-flow during a turbine load release in a plant with a high-pressure-side surge tank; Fig. 2b a diagram of the level in the high-pressure-side surge tank during a turbine load release in accordance with Fig. 2a; Fig. 3a a diagram of the through-flow during a turbine load take up in a plant with a high-pressure-side surge tank; Fig. 3b a diagram of the level in the high-pressure-side surge tank during a turbine load take-up in accordance with Fig. 3a; Fig. 4a a diagram of the through-flow during a turbine load take up after a load release in a plant with a high-pressure side surge tank; Fig. 4b a diagram of the level in the high-pressure-side surge tank during a load take-up after a load release in accordance with Fig. 4a; Figs. 5a - 5d diagrams of the improvement achieved in accordance with the object by the method in accordance with the -8 invention in a plant with a high-pressure-side surge tank; Fig. 6 a diagram of the load dependence of the boundary curves; Fig. 7 a diagram of a number of boundary curves for different load for maintaining a maximum level and Figs. 8a - 8d diagrams of the improvement achieved in accordance with the object by the method in accordance with the invention in a plant with a low-pressure-side surge tank. Reference is made to Figs. 5a to 5c, which illustrate diagrams of the load of a plant with a high-pressure-side surge tank starting from a stationary state. In Fig. 5a the plot of the opening Y of the regulation member of the turbine is shown as a function of time. At the time point Ti a desired value of the turbine load corresponding to 70 % opening is required. The curve 32 shows the plot of the turbine load, which rises in the manner of a ramp corresponding to the opening time of the turbines. If the turbine load (curve 33) is increased to the desired value without regard to the level plot (curve 34), then the permissible minimum level 26 is dropped below in the surge tank beginning at the time point T2 (Fig. 5c). At the time point T3 the load manoeuvre must be interrupted in order to approximately reach the minimum level in the surge tank at the time point T 4 , but not to drop below it (curve 24). The basic idea of the method consists in the definition of a criterion for the delaying or the interrupting of the load manoeuvre respectively. This V -yI'~ -9 criterion is tested for in a special representation form of the state of an oscillatory system, the phase plane (Fig. 5d). In the phase plane the behaviour of a state variation of the oscillatory system is illustrated two dimensionally. The coordinates of this representation are the level z in the surge tank, which is measured, and the net influx AQ to the surge tank, which is obtained from the derivative of the level with respect to time (dz/dt). In the phase plane there are special curves which describe the behaviour of the undisturbed system when the load is held constant. These curves can be constructed directly through numerical simulation of the system. That curve in Fig. 5d which passes through the point M with the desired minimum z-coordinate is the boundary curve 35. The vector in the phase plane of Fig. 5d which is formed by the level is compared at short time intervals with the boundary curve. As soon as the boundary curve is reached the turbine load is limited to the current value, e.g. 37.5 % opening (Fig. 5a). A relief however remains possible. It can be seen from Figs. 5c to 5d that the thus limited load manoeuvre satisfies the requirement of the minimum level. A safety margin is to be determined expediently and taking into account the precision of the data available. The limiting function is effective only at the left side of the phase plane, that is, when the level in the surge tank sinks. It can be recognised in Fig. 5c that after the z-coordinate has been reached the level remains approximately constant over a longer time interval. This is explained in that when the sign of AQ changes a load manoeuvre is permitted and a small increase of the load is permitted until the derivative of the level becomes negative again, whereupon the load -10 manoeuvre is suppressed. This time period with approximately constant level is concentrated in the phase plane at the point M. In the temporal behaviour of the through-flows (Fig. 5b) it is the section A, B, in which the two through-flow values of the shaft (22) and the turbines (21) agree, that is, the net influx AQ of the surge tank is zero. The boundary curve can change its position in dependence on different influences. The most important influences are the upper water level at the inlet to the shaft and the momentary load on the plant, represented by the openings of the turbines. In Fig. 6 the boundary curves are illustrated for five opening values at spacings of 20 % each under the assumption of the same upper water level. The lower crown is given by the chosen minimum point (M in Fig. 5d), whereas the upper crown is displaced to lower level values as a result of the frictionally caused pressure loss in the pressure shaft. In addition to the becoming empty of the surge tank there is the risk that the surge tank overflows during load manoeuvres. This can arise as a result of load releases. In order to prevent this a boundary curve is to be determined for each load prior to release. As Fig. 7 shows, the representation in the phase plane is again used for this, with the line 42 representing the level when the surge tank overflows and the line 43 representing the level when the surge tank becomes empty. During the operation, in the worst case, the load release can take place in the middle of the upswing of the surge tank, with the boundary curve which passes through the point S being reached. Through this definition a minimum point M is obtained for each turbine load which is capable of - 11 ensuring the maintaining of the maximum level S. For each minimum value a boundary curve can be determined for the down-swing of the surge tank. As in the case of the becoming empty, the upper water level also influences the boundary curve during the overflow. If both kinds of risk are present in a plant, then both criteria are continually monitored, with each criterion leading to a blocking of the turbine load independently of the other one. As already mentioned, a hydroelectric power plant can have a surge tank on the low pressure side. In plants of this kind there is the danger that the surge tank overflows when the turbines are loaded. In order to prevent this, the method as is provided against the becoming empty in high-pressure-side surge tanks can be used. In a manner analogous to the method which was described in connection with Figs. 5a to 5d the state is compared in the phase plane with a boundary curve or, when required, an array of boundary curves, here however in the case of the positive gradient, i.e. in the upswing phase. This is illustrated in Figs. 8a to 8d. If the boundary curve 35 is reached, then the blocking of the openings of all setting members takes place. In Fig. 8c the level changes occur with signs which are reversed with respect to Fig. 5c, and Fig. 8d is rotated by 180 degrees with respect to Fig. 5d since both axes are mirrored. The lower water level 41 takes the place of the upper water level, and the maximum level 42 takes the place of the minimum level. In the case of a low-pressure-side surge tank there arises the risk of -97 -12 becoming empty during the load release. This can be eliminated in analogy with the high-pressure-side surge tank. It must be ensured that a specific maximum level 42, which is again load dependent, is not exceeded at the beginning of the down-swing process. This is achieved through the use of load-dependent boundary curves during the down swing. In a plant with surge tanks on both sides of the powerhouse it is possible to secure against risks on both sides. The criteria for the blocking of the turbine opening are combined in the sense of an "or" link, that is, each criterion by itself leads to a blocking. It can be necessary or even advantageous to restrict the computational cost and complexity of the limitation device. This is possible in that the computational taking into account of individual influences is dispensed with when the boundary curve is used. In such cases the boundary curve has the character of an approximation. The numerical definition of the boundary curve must take this fact into account and contain an enlarged safety margin. In return, the simplicity of the method is preserved. This simplification can be meaningfully used when - certain parameters in the limitation device are not available as measurement values, e.g. the upper water level or the total load of the plant, - the oscillatory system is complex. With a higher order of the system,

ST

- 13 e.g. in the presence of a differential surge tank, a boundary surface would have to be used instead of a boundary curve, - the limitation is to be realised with a predetermined, limited computational capacity of the equipment. As can be seen in Fig. 6, the shape of the boundary curve varies considerably with the load. This is substantially a result of the static pressure fall-off in the shaft. Mostly a plurality of turbines operate at one shaft and surge tank. The load influence can be correctly represented only when information on the total through-flow is present in a device. In many cases the measurement signals for the through flow are too highly delayed or have a deficient resolution. It is therefore advantageous to calculate the through-flows. The calculation can be formed either in the device itself, in the turbine regulators or other devices as a result of other measurement signals, e.g. opening, pressure, speed of rotation. For the forming of sums from the individual through-flows a device which is on a higher level with respect to the group conduction plane is advantageously used. Both the current value of the level in the reservoir and the temporal behaviour of the level in the surge tank can be prepared with the help of an observer. The observer is a function in the region of the limitation device. It contains a simplified dynamic computational model of the most important plant components and continually calculates from the through-flow of the turbines the variations of the tube path pressure and of the level in the surge tank. The latter are compared with the measured values and the observer performs a correction as a result of a -7 zDD - 14 the differences, (e.g. value of the water level) so that the deviations are avoided. Each boundary curve is originally defined as the trajectory or phase portrait of an oscillatory process about the stationary surge tank level which depends on the level of the reservoir bordering on the shaft. Since this water level varies in the course of time, the boundary curves are displaced in dependence on the level of the reservoir. The value of the level must be available as information in the limitation device in order to be able to take this influence into account. In plants with a plurality of turbines which operate at the same shaft it is advantageous to provide the limitation function not in the turbine regulator but rather in a higher level device, e.g. in a plant regulator, power plant computer or joint control. This has the advantage that a simplified regulator structure can be used in the regulators on the machine level and an unnecessarily large number of signal inputs is avoided. The blocking function can be qualitatively improved in some applications. For this namely two possibilities are selectively available: - in order to avoid the responding in a plurality of stages (displaced opening), two boundary curves which extend in parallel are used which control the switching on and off of the blocking function in the sense of a hysteresis. At the outer boundary curve the blocking becomes effective; at the inner one it is removed again.

- 15 - in applications with coarsely approximated boundary curve it is advantageous to use two boundary curves at a larger spacing through which the blocking function is not switched on and off, but rather the permissible speed of the load increase is continuously varied between zero and the permissible maximum value. At the outer boundary curve the permissible loading speed becomes zero (the blocking is thus reached); at the inner one it achieves a maximum value which corresponds to the shortest provided opening time of the setting member. The security against failure of the limitation device requires the monitoring of the input signals. When a signal drops out the limitation function is no longer ensured by the boundary curves. In this case the limitation function is automatically displaced into the gradient limiter, which is placed ahead of the position regulation of the setting member of each turbine. The maximum permissible opening speed, which is determined in this element, is reduced in the event of a failure of the limitation device to a value which is low enough in order to avoid critical pendulum movements in the surge tank at any desired load manoeuvres.

YXI

Claims (13)

1. Method for the controlling of load manoeuvres in a hydroelectric power plant with a high-pressure-side and/or a low-pressure-side surge tank, characterised in that a level is determined in the surge tank; in that on the basis of the temporal behaviour of this level the state variables: actual level (z) and net influx (AQ) are determined for the surge tank and are compared in a phase plane with a boundary curve which corresponds to the state plot at fixed load and which extends approximately through the point of the maximum or minimum permissible level; and in that the load increase is interrupted when the boundary curve is reached.

2. Method in accordance with claim 1, characterised in that the desired state is monitored at time intervals.

3. Method in accordance with any one of the claims 1 to 3, characterised in that the determination of the level in the surge tank takes place through measurement.

4. Method in accordance with claim 1, characterised in that a minimum level and/or a maximum level is determined in order to prevent a becoming empty or an overflowing of the surge tank respectively.

5. Method in accordance with any one of the preceding claims, characterised in that a minimum or a maximum level is maintained in order that in the event of a switching off a - 17 maximum or minimum level threshold value is not exceeded in the succeeding swinging back of the level in the surge tank.

6. Method in accordance with any one of the preceding claims, characterised in that the boundary curve is determined in dependence on the load.

7. Method in accordance with any one of the preceding claims, characterised in that one or more parameters are used in the determination of the boundary curve.

8. Method in accordance with claim 7, characterised in that in the determination of the boundary curve, parameters which are selected from the group containing: level in the reservoir, total through-flow of the turbines, level in the differential surge tank are used so that the latter become a boundary condition of a higher order.

9. Method in accordance with any one of the preceding claims, characterised in that the total through-flow through the turbines is determined.

10. Method in accordance with any one of the preceding claims, characterised in that the determination of parameters and/or the state takes place in the phase plane by an observer in the sense of regulation technology.

11. Method in accordance with any one of the preceding claims, characterised in that a second boundary curve is determined - 18 which extends parallel to the first boundary curve in order to control a load variation in the sense of a hysteresis.

12. Method in accordance with any one of the preceding claims, characterised in that instead of the blocking of the load increase a gradual decrease in the permissible speed of the load increase is used which is controlled by the position of the state in the phase plane.

13. Method in accordance with any one of the preceding claims, characterised in that when the information on the level drops out the permissible speed of the load increase is automatically limited to an amount which is harmless in any situation.