JPH06241003A - Starting method for steam turbine - Google Patents

Abstract

PURPOSE:To optimize life consumption in sequence by estimating a turbine operation parameter so as to optimize an evaluation function, and controlling a turbine controller in advance according to the estimated value. CONSTITUTION:Based on the signals of temperature detection parts 8, 9, a high pressure rotor thermal stress computation part 10 and a medium pressure rotor stress computation part 11 compute rotor radial temperature distribution and thermal stress generated at a surface and a bore to output the computed values to a model estimation part 1. The model estimation part 1 receives the rotor radial temperature distribution, the thermal stress generated at the surface and the bore, and present turbine state values of a main steam pressure detection part 5, a main steam temperature detection part 6, generator output from a generator 14 and so on. A prediction section Sn+1 is determined as its initial value to give a speed increasing rate Vtn+1 in the case of operating only for the next one section to a valve opening controller 2. It is thus possible to always keep an optimum operating condition taking into consideration life consumption.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for starting a plant consisting of a boiler and a steam turbine, and particularly for starting a steam turbine of a plant such as an intermediate load operating plant for power generation, where rapid and frequent start / stop operations are required. Regarding the method.

[0002]

2. Description of the Related Art When a steam turbine is started, the steam temperature fluctuates significantly due to acceleration and load changes. As a result, a metal temperature distribution is generated inside the turbine rotor, which is a wall pressure member, and a large thermal stress is applied to the rotor surface and bore. Occurs. It is known that when the thermal stress exceeds a certain value, the life of the rotor is consumed. Further, the life consumption per one start is quantitatively grasped by the life consumption curve of FIG. 3 according to the peak size and the number of times of the thermal stress at that time.

In recent years, rapid and frequent start-stop operation has been required for intermediate load operation plants for power generation. Therefore, at the time of this rapid start-up, there is a need for a start-up method that suppresses the thermal stress within the limit value and also makes the rotor life consumption appropriate for one start-up. An example of a conventional method for this problem will be shown.

At the time of starting the steam turbine, it is the rotor portion after the first stage of the high pressure turbine that is in the most severe thermal stress condition. Therefore, at the time of conventional activation, the automatic activation operation is performed using a mismatch chart as shown in FIG. This chart shows the high pressure
Assuming the post-stage steam temperature, the speed increase rate, heat soak time, initial load amount, from the amount of mismatch with the casing inner surface metal temperature after the first high pressure (assumed to be equivalent to the turbine rotor surface metal temperature) measured online at the same time, The operation parameters such as the load change rate are selected from several patterns.

[0005]

However, in the turbine start-up method as described above, since the mismatch temperature between the steam and the turbine metal at the time of start-up can be dealt with by only a finite number of patterns of start-up methods, the turbine thermal stress can be reduced. However, optimum driving cannot always be performed.

The life consumption can only be predicted to some extent with respect to the life consumption with respect to a predetermined starting pattern, and the life consumption is positively considered when a disturbance such as a sudden change in steam condition occurs on the boiler side. Control correspondence is not performed.

Therefore, an object of the present invention is to sequentially optimize steam turbine start-up operating parameters in accordance with a steam turbine start-up dynamic characteristic model, so that the start-up time and heat can be reduced regardless of the turbine state before start-up and disturbance during operation. It is an object of the present invention to provide a steam turbine starting method that sequentially optimizes stress and life consumption.

[0008]

A steam turbine starting method of the present invention comprises a dynamic characteristic model at the time of starting a steam turbine,
According to the model, the starting time and the thermal stress generated in the rotor,
To predict the turbine operation parameter so as to optimize the evaluation function constituted by the rotor life consumption up to a certain section in the future every certain time period, and to pre-control the turbine controller according to the predicted value. Is characterized by.

[0009]

In the turbine control method as described above, a feedback effect is generated by repeating the predictive calculation one after another, and a deviation from the optimum value due to the difference between the model and the actual machine and the influence of disturbance such as a rapid main steam pressure fluctuation caused by the boiler. Can be made smaller.

[0010]

EXAMPLE An example of the present invention will be described below. In the present invention, by using the model predictive control technique, which is one of the process control methods, the turbine start-up operation parameter is optimized within the sequential prediction section, and the turbine operation is performed.

A model capable of calculating a starting limit value such as a condition inside the turbine, in particular, a thermal stress generated in the high and medium pressure turbine rotor and a life consumption thereof due to the condition is prepared from the turbine inlet steam condition and the steam turbine operating parameter.

The evaluation function for the starting limit value and the starting time is created, and the function is used to perform the optimization calculation up to the prediction section for each certain time period. The turbine is predicted by predicting the operation pattern from that time to the future. Start up. FIG. 1 is a configuration diagram of an embodiment of the present invention, and FIG. 2 is a characteristic diagram showing how the rate of speed increase changes by optimization calculation.

From the signals of the temperature detecting units 8 and 9, the high pressure rotor thermal stress calculating unit 10 and the medium pressure rotor thermal stress calculating unit 11
The temperature distribution in the radial direction of the rotor and the thermal stress generated on the surface and the bore are calculated, and the calculated values are output to the model predicting unit 1.

In the model predicting unit 1, the temperature distribution in the radial direction of the rotor, the thermal stress generated on the surface and the bore, the main steam pressure detecting unit 5, the main steam temperature detecting unit 6, and the generator 14
The current state value of the turbine such as the generator output from
New optimizes the prediction interval S _{n + 1} of these as an initial value, a speed increasing ratio V _{tn + 1} of the signal to drive the fastest rate V _{tn + 1} to determine the valve opening controller 2 only the following one section give.

The model used in the model predicting section 1 is a general physical model formulated by mass balance and energy balance, as shown in FIG. Due to the nature of the model, L, T and the thermal stresses S _HB , S _HS , S _IB , and S _IS constituting the evaluation function J _c, which are the limiting values, are not explicit functions but nonlinear. . Therefore, the algorithm for the optimization calculation is to solve a non-linear optimization problem, which is a direct search method that uses only the function value described above and does not require a derivative. This time, as an example, the Rosenbrock algorithm, which is a kind of direct search method as shown in FIG. 5, was used. This method sequentially searches the evaluation function in the direction of the coordinate axis. If it succeeds, the step width in that direction is tripled, and if it fails, it is multiplied by 0.5 in the opposite direction, and at least 1 in all directions. If the number of times succeeds and then fails at least once in all directions, the coordinate axes are rotated, and the search is sequentially performed again in the rotated direction. Further, since the conditional optimization problem is solved, the judgment of "success?" In FIG. 5 is limited by whether the thermal stress limit value is exceeded. Further, after the end of the speed increase, until the turbine target load, an evaluation function in which the operating parameter is changed to the load change rate may be used.

[0016]

Therefore, according to the present invention, by successively performing the model predictive optimizing operation as described above, a feedback effect is generated, and the deviation from the optimum value due to the difference between the model and the actual machine and the influence of disturbance can be reduced, and the life is always maintained. It is possible to maintain the optimum operating condition in consideration of consumption.

Further, by varying the weight in the evaluation function, it is possible to set the start / end time-oriented mode and the life consumption-oriented mode online steplessly at the discretion of the plant operator.

[Brief description of drawings]

FIG. 1 is a configuration diagram showing an embodiment of the present invention.

FIG. 2 is a characteristic diagram showing how the speed increase rate changes by optimization calculation.

[Figure 3] Characteristic diagram of thermal stress and life consumption rate

FIG. 4 is a block diagram of a model predicting unit of the present invention.

FIG. 5 is an explanatory diagram of a Rosenbrock algorithm used for optimization calculation.

FIG. 6 is an explanatory diagram of a conventional mismatch chart.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Model prediction part, 2 ... Valve opening controller, 3 ... Main steam stop valve, 4 ... Control valve, 5 ... Main steam pressure detection part, 6 ... Main steam temperature detection part, 7 ... High pressure first stage rear rotor metal (Casing inner surface metal) temperature detection unit, 8 ... Medium pressure first stage rotor metal (casing inner surface metal) temperature detection unit, 9 ... High pressure rotor thermal stress calculation unit, 10 ... Medium pressure rotor thermal stress calculation unit,
11 ... High-pressure turbine, 12 ... Medium-pressure turbine, 13 ... Generator.

Claims (1)

[Claims] 1. A dynamic characteristic model at the time of starting the steam turbine is provided, and an evaluation function constituted by the starting time, the thermal stress generated in the rotor, and the rotor life consumption according to the model is used every certain time period until a certain section in the future. A method for starting a steam turbine, comprising predicting a turbine operation parameter and optimally controlling the turbine control device according to the predicted value so as to optimize during the period.