CN104748807A - Online power station main steam flow calculation method based on flow correction - Google Patents

Abstract

The invention relates to an online power station main steam flow calculation method based on flow correction. According to the method, an online main steam flow calculation model is corrected through a mass flow balance equation, the model finishes flow calibration on the basis of condensate water flow measured by a spray nozzle, under the different load working conditions, iteration method verifying calculation is utilized to obtain the main steam flow value, the Friuli Greig formula is corrected through the verified main steam flow measuring value, and finally an online main steam flow calculation model is built and obtained. By means of the method, the change tendency of the main steam flow can be reflected quickly, and meanwhile the high accuracy of the flow can be guaranteed. Considering the situation that sensor drift distortion occurs in a field adjusting level pressure measuring point, whether the adjusting level pressure value is distorted or not is judged according to the association coefficient, accordingly an adjusting level pressure soft measurement replacement solution is proposed, and the stability of an online main steam flow calculation module is guaranteed. The online main steam flow calculation strategy can be referred to by a power station and be used for conducting economic analysis.

Description

A kind of power station main steam flow on-line calculation method based on flux modification

Technical field

The present invention relates to a kind of power station main steam flow on-line calculation method, the soft-sensing model having comprehensive pivot analysis and support vector machine concurrently belongs to machine learning modeling field.

Background technology

In process industrial, there are some cannot directly measure or measure the variable having very large time delay, needs to be estimated it by soft-measuring technique Modling model.In the monitoring of fired power generating unit on-line performance with Optimal Management System, need to use main steam flow in the calculating of the economic target such as heat consumption rate, coa consumption rate, the accuracy that main steam flow calculates will directly have influence on the reliability of unit performance calculating and running optimizatin.The method of current calculating unit main steam flow mainly contains the direct method of measurement, the indirect method of measurement and artificial intelligence model method.

The direct method of measurement refers to directly to be measured by installing orifice plate constant pitch stream device, but the method can cause very large restriction loss.The ultimate principle of the indirect method of measurement calculates main steam flow based on pressure after governing stage, and the method is subject to the impact of flow passage component fouling, can reduce the accuracy that steam flow calculates result.Main steam flow model based on neural network will be that machine learning method is applied in hard measurement, but it has the poor shortcoming of adaptability, robustness.

Main steam flow measurement model, needs the sensitivity and the accuracy that consider model.Therefore, the present invention obtains main steam flow computation model in conjunction with flux balance equations and Fu Liugeer formula.Because although Fu Liugeer formula can reflect the variation tendency of main steam flow when being in variable working condition accurately; Based on the main steam flow computing method that condensing water flow is checked, the main steam flow value obtained under its quiescent conditions is more accurate.According to the feature of these two kinds of main steam flow computation models, the main steam flow that the present invention's condensing water flow obtains after checking, to revise Fu Liugeer formula, makes the main steam flow accuracy calculated during the variable working condition on a large scale all be improved with response promptness.Further, consider from robustness and angle, when first stage pressure measuring point breaks down, judge that fault occurs and adopts hard measurement substitute revision of option in time, guarantee that model is normal.

Summary of the invention

Goal of the invention: in order to overcome the deficiencies in the prior art, the invention provides a kind of power station main steam flow on-line calculation method based on flux modification.

Technical scheme: for solving the problems of the technologies described above, a kind of power station main steam flow on-line calculation method based on flux modification of the present invention, described step comprises as follows:

(1) field data exports data-interface to via the network switch;

(2) batch capture condensing water flow D _ns, feedwater flow D _gs, middle pressure gate bar steam loss D _kf, high-pressure door bar steam loss D _af, reheating spray flow D _zr, overheated spray flow D _gr, unit load P _load, pressure P after level ₁₀, temperature T after level ₁₀, and high adding at different levels import and export heat regenerative system parameter, wherein unstable data removed by filtrator, set up sample database;

(3) by the sample database in step (2), check module by the laggard inbound traffics of load classification, obtain flow G ₁₀;

(3.1) the condensing water flow D under each load section is obtained via Data Input Interface _ns, feedwater flow D _gs, middle pressure gate bar steam loss D _kf, high-pressure door bar steam loss D _af, reheating spray flow D _zr, overheated spray flow D _grand the high import and export parameter added at different levels, calculate extraction flow;

Described step (3.1) concrete steps are: calculate extraction flow

$d_1=\frac{D_{\mathrm{fw}}(h_{w1}-h_{w2})}{(h_1-h_{s1})}$

$d_2=\frac{D_{\mathrm{fw}}(h_{w2}-h_{w3})-d_1(h_{s1}-h_{s2})}{(h_2-h_{s2})}$

$\begin{array}{c}{d}_3=\frac{D_{\mathrm{fw}}(h_{w3}-h_{w4})-(d_1+d_2)(h_{s2}-h_{s3})}{h_3-h_{s3}}\\ -\frac{D_{\mathrm{af}}(h_{\mathrm{gin}}-h_{s3})+D_{\mathrm{kf}}(h_{\mathrm{zin}}-h_{s3})}{h_3-h_{s3}}\end{array}$

$\begin{array}{c}{d}_4=\frac{(D_{\mathrm{fw}}+D_{\mathrm{zrps}}+D_{\mathrm{grps}})(h_{w4}-h_{w5})}{h_4-h_{w5}}\\ -\frac{(d_1+d_2+d_3)(h_{s3}-h_{w5})}{h_4-h_{w5}}\end{array}$

D _kf＝k ₁*D _fw+k ₂

D _af＝k ₃*D _fw+k ₄

H _wi, h _wi+1, h _i, h _siwhat represent that this section draw gas respectively goes out saliva enthalpy, import water enthalpy, and draw gas enthalpy, hydrophobic enthalpy, and enthalpy is obtained by water vapour computing formula by pressure and temperature;

D _kffor middle pressure gate bar air leakage; D _affor high-pressure door bar air leakage; Coefficient k ₁, k ₂, k ₃, k ₄by thermal test gained; D _zrpsfor reheating spray flow; D _grpsfor overheated spray flow, if cross hot water spray to export extraction, then D from No. 1 high adding _grpsget 0.

(3.2) according to mass conservation formula D _{fw_js}=D _ns+ d ₁+ d ₂+ d ₃+ d ₄+ D _kf+ D _af-D _zr-D _gr, with extraction flow d ₁..., d ₄computing formula simultaneous, calculates feedwater flow initial value and is set to D _gs;

(3.3) | D _{fw_js}-D _gs| during > difference in flow limit value, D _gs=D _{fw_js}, and enter step (3.2), iterative computation, until | D _{fw_js}-D _gs| < difference in flow limit value, exports now feedwater flow, and then calculates main steam flow G ₁₀, t/h.

(4) under each load section, the main steam flow expression formula that after governing stage, pressure and temperature represents: in formula, p ₁₀for with G ₁₀force value after corresponding governing stage, unit is Mpa; T ₁₀for temperature value after the governing stage under declared working condition, unit is K, sets up unit load P _loadwith coefficient k _m, funtcional relationship, k _m=f (P _load);

(5) by real time data unit load P _load, first stage pressure P ₀with temperature T after level ₀, according to formula obtain and calculate main steam flow G;

(6) judge to calculate the correlativity between main steam flow and load: in formula, X, Y represent main steam flow and load two data samples respectively, ρ _xYfor the related coefficient between X, Y; DX, DY are respectively the variance of variable X and Y; E (X), E (Y), E (XY) are the average of X, Y, XY respectively,

If correlation coefficient ρ _xYduring >0.8, be judged to be to calculate main steam flow credible, can be used as the main steam flow source in performance calculating; Otherwise, be judged to be first stage pressure sensor fault, enter step (7) first stage pressure hard measurement correction module;

(7) first stage pressure sensor first stage pressure P is under normal circumstances obtained ₀and extraction pressure P at different levels _{1 ...,}p ₈, after this type of sample pivot analysis, as the input of support vector machine, the output valve after Modling model is first stage pressure hard measurement value, replaces first stage pressure sensor values, and then enter step (5) by this result;

(7.1) steam turbine first stage pressure P is affected via Data Input Interface acquisition ₀each parameter: heat regenerative system extraction pressure P at different levels _j(j=1,2 ... 8), it can be used as the input parameter of pivot analysis, be designated as X _old(n × m), sampling number is measured in n representative, and m represents measurement attribute number;

(7.2) according to following formula, standardization input data, obtain X (n, m)

$x(i,j)=\frac{x_{\mathrm{old}}(i,j)-\mathrm{average}\left(x_{\mathrm{old}}(:,j)\right)}{\mathrm{std}\left(x_{\mathrm{old}}(:,j)\right)}$

In formula: i=1,2 ..., n, j=1,2..., m, average (x _old(:, j)) represents the average of a jth variable down-sampling point, std (x _old(:, j)) represents the standard deviation of a jth variable down-sampling point;

(7.3) the covariance matrix COV (X) of X (n, m) is calculated, and eigenvalue λ _iwith proper vector p _i

$\mathrm{COC}\left(X\right)=\frac{X^{T}X}{m-1}$

COV(X)p _i＝λ _ip _i

Carry out SVD decomposition to COV (X), and carry out pivot analysis to X, choose k pivot, k is constant, determines that the process of score matrix T so far pivot analysis terminates;

(7.4) by k pivot obtaining and first stage pressure sampled value P ₀respectively as the input and output of support vector machine, wherein n/2 group data are as training data, and n/2 group is as test data in addition, first by inputoutput data standardization;

In high-dimensional feature space, construct optimum linearity decision function y (x)=sgn [w ψ (x)+b], take following formula objective function:

$\min \underset{w,b,\mathrm{\&xi;}}{J}(w,\mathrm{\&xi;})=\frac{1}{2}{w}^{T}w+C\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{\mathrm{\&xi;}}_i$

In formula, constraint condition w is weight factor, and C is penalty parameter, and b is deviate, C ∈ R ⁺punishment parameter, ξ _i=[ξ ₁..., ξ _n] ^t, that a Nonlinear Mapping can x _ibe mapped to the feature space of higher-dimension (even infinite dimension) from the input space thus realize non-linear regression the input space and be converted into linear regression in high-dimensional feature space, constrained optimization can be converted into unconstrained optimization by structure Lagrange function, according to KKT condition the optimization problem solved finally can be converted into and solve linear equation:

$\left[\begin{array}{cc}0& I_v^T\\ I_v& \mathrm{\&Omega;}+c^{-1}I\end{array}\right]\left[\begin{array}{c}b\\ a\end{array}\right]=\left[\begin{array}{c}0\\ y\end{array}\right]$

Wherein, data y=[y is exported ₁..., y _n] ^t; Unit array I _v=[1 ..., 1] ^t; Lagrange multiplier a=[a ₁..., a _n] ^t; Ω={ Ω _ij| i, j=1 ... n}; k () is kernel function, can select Radial basis kernel function K (x herein _i, x _j)=exp [-|| x _i-x _j|| ²/ (2 σ ²)];

(7.5) calculated value of support vector machine part of detecting is hard measurement income value, enters step (5), replaces first stage pressure sensor values by this result.

So far, the main part that main steam flow calculates completes.The through-flow characteristic considering unit operationally between longer or midway change through large light maintenance, step 1-4 can according to on-site actual situations, regular or irregular repeat calculation, and the actual through-flow performance of primary steam computing module and unit is conformed to.

Beneficial effect: the present invention in terms of existing technologies, has the following advantages:

(1) the online computation model of main steam flow of mass rate balance equation correction Fu Liugeer formula, carrys out iteration with condensing water flow and checks feedwater flow, thus calculate the higher main steam flow of degree of accuracy to revise Fu Liugeer formula.Not only accuracy is high for the Fu Liugeer formula that final correction obtains, and dynamic response is timely, can reproduce main steam flow value comparatively truly.

(2) in conjunction with PCA and support vector machine to the modeling of Steam Turhine Adjustment stage pressure, the deviation that can occur when calibrating (base measuring) pressure sensor is made mistakes preferably.

(3) along with the change of the through-flow characteristic of unit, the parameter upgraded in time in computation model, ensures that model calculation value conforms to actual main steam flow.

(4) for power plant's monitoring information system Premium Features module (condition monitoring and fault diagnosis etc.) provides can reference model.

Accompanying drawing explanation

Fig. 1 is process flow diagram of the present invention.

Fig. 2 is power of the assembling unit curve map in the present embodiment.

Fig. 3 is the change curve of two kinds of main steam flow calculated values in the present embodiment.

Fig. 4 is the related coefficient graph of a relation in the present embodiment between different main steam flow calculated value and load.

Embodiment

Below in conjunction with accompanying drawing, the present invention is further described.

For certain power station 600MW overcritical resuperheat condensing-type unit, to gather in SIS system the data of 10 on April 10th, 10 o'clock 1 April 01 in 2014, acquisition interval 1 minute.Framework of the present invention mainly contains nucleus module such as input data prediction, flow capacity checking module, main steam flow computing module, first stage pressure hard measurement etc., detailed process as shown in Figure 1:

1) field data exports data-interface to via the network switch;

2) condensing water flow (D was gathered with 1 minute time interval _ns), feedwater flow (D _gs), middle pressure gate bar steam loss (D _kf), high-pressure door bar steam loss (D _af), reheating spray flow (D _zr), overheated spray flow (D _gr), unit load (P _load), pressure (P after level ₁₀), temperature (T after level ₁₀), and the high import and export parameter added at different levels, filtrator is removed wherein after unstable data, remains 9372 groups of samples.

3) sample data step 2 obtained, checks module by load section laggard inbound traffics of classifying, obtains flow G ₁₀.

4) load (P is set up _load) and coefficient k _m funtcional relationship, k _m=f (P _load), prepare for calculating main steam flow in real time.

5) access real-time unit load (P _load), first stage pressure (P ₀) and level after temperature (P ₀), according to formula obtain and calculate main steam flow G.

6) judge to calculate the correlativity between main steam flow and load: if correlation coefficient ρ _xYduring >0.8, be judged to be to calculate main steam flow credible, can be used as the main steam flow source in performance calculating; Otherwise, be judged to be first stage pressure sensor fault, enter step first stage pressure hard measurement correction module.

7) first stage pressure sensor first stage pressure (P is under normal circumstances obtained ₀) and extraction pressure (P at different levels _{1 ...,}p ₈), after this type of sample pivot analysis, as the input of support vector machine, the output valve after Modling model is first stage pressure hard measurement value.Replace first stage pressure sensor values by this result, and then enter step 5.

8) main part that main steam flow calculates completes.The through-flow characteristic considering unit operationally between longer or midway change through large light maintenance, step 1-4 can according to on-site actual situations, regular or irregular repeat calculation flow, and the actual through-flow performance of primary steam computing module and unit is conformed to.

Calculated examples: suppose the standard value 50t/h under feedwater flow value off-design operating mode, obtain the feedwater flow value after checking, and by it compared with exact value, its result of calculation is as shown in table 1 below by the condensing water flow iterative computation under design conditions.

Table 1 feedwater flow checks the error analysis of model

As can be seen from Table 1, relative error between the feedwater flow obtained by condensing water flow calculation and check and exact value is no more than 1%, show, when feedwater flow measuring point error is larger in scene, the condensing water flow iterative computation less according to measuring error feedwater flow value relatively more accurately can be obtained.

After the governing stage that the main steam flow obtained by condensing water flow calculation and check, unit are collected, force value and exhaust temperature of HP value, obtain coefficient k _mvalue is 2.45196, thus can this unit main steam flow timely monitor model.

Gather unit carrys out verification model dynamic perfromance at the operational parameter value in certain varying load stage, the change curve of unit power under this varying load state as shown in Figure 2, adopt the main steam flow computing method that the present invention proposes, calculate the main steam flow value in this varying load stage, in the following Fig. 3 of its change curve (D1 is the curve of top in figure) shown in D1.Under this varying load state, spray water the main steam flow curve of cyclical fluctuations that is added and obtains with unit feedwater flow and overheated desuperheat as shown in D2 in Fig. 3 (curve that D2 is Figure below).

Adopt the algorithm of matrix times window, calculate two kinds of correlation coefficient value calculated between main steam flow and load respectively.Getting time window span is 50, and every 50 data namely from first sample data, as a matrix, obtain 431 sample matrix.Correlation coefficient value between the main steam flow that the computing method proposed by this invention obtain and the sample matrix of load represents with variable C1, correlation coefficient value between the main steam flow that feedwater flow measuring point calculates and the sample matrix of load variable C2 represents, is illustrated in figure 4 the correlation curve of C1 and C2.

As seen from Figure 4, C1 value perseverance is greater than C2, and when sudden load change, C2 amplitude of variation comparatively C1 is more obvious, and this shows in load fluctuation process, adopts the retardance of the computation model of feedwater flow measured value calculating main steam flow larger.It is better compared to directly calculating main steam flow value dynamic perfromance with feedwater flow measuring point that this shows to adopt the present invention's main steam flow used computation model.

The above is only the preferred embodiment of the present invention; be noted that for those skilled in the art; under the premise without departing from the principles of the invention, can also make some improvements and modifications, these improvements and modifications also should be considered as protection scope of the present invention.

Claims (4)

1., based on a power station main steam flow on-line calculation method for flux modification, it is characterized in that: described step comprises as follows:

(1) field data exports data-interface to via the network switch;

(2) batch capture condensing water flow D _ns, feedwater flow D _gs, middle pressure gate bar steam loss D _kf, high-pressure door bar steam loss D _af, reheating spray flow D _zr, overheated spray flow D _gr, unit load P _load, pressure P after level ₁₀, temperature T after level ₁₀, and high adding at different levels import and export heat regenerative system parameter, wherein unstable data removed by filtrator, set up sample database;

(3) by the sample database in step (2), check module by the laggard inbound traffics of load classification, obtain flow G ₁₀;

(4) under each load section, the main steam flow expression formula that after governing stage, pressure and temperature represents: in formula, p ₁₀for with G ₁₀force value after corresponding governing stage, unit is Mpa; T ₁₀for temperature value after the governing stage under declared working condition, unit is K, sets up unit load P _loadwith coefficient k _m, funtcional relationship, k _m=f (P _load);

(5) by real time data unit load P _load, first stage pressure P ₀with temperature T after level ₀, according to formula $G=f\left(P_{\mathrm{load}}\right)\&times;\frac{P_0}{\sqrt{T_0}},$ Obtain and calculate main steam flow G;

(6) judge to calculate the correlativity between main steam flow and load: in formula, X, Y represent main steam flow and load two data samples respectively, ρ _xYfor the related coefficient between X, Y; DX, DY are respectively the variance of variable X and Y; E (X), E (Y), E (XY) are the average of X, Y, XY respectively,

If correlation coefficient ρ _xYduring >0.8, be judged to be to calculate main steam flow credible, can be used as the main steam flow source in performance calculating; Otherwise, be judged to be first stage pressure sensor fault, enter step (7) first stage pressure hard measurement correction module;

(7) first stage pressure sensor first stage pressure P is under normal circumstances obtained ₀and extraction pressure P at different levels _{1 ...,}p ₈, after this type of sample pivot analysis, as the input of support vector machine, the output valve after Modling model is first stage pressure hard measurement value, replaces first stage pressure sensor values, and then enter step (5) by this result.

2. the power station main steam flow on-line calculation method based on flux modification according to claim 1, is characterized in that: described step (3) concrete steps are:

(3.1) the condensing water flow D under each load section is obtained via Data Input Interface _ns, feedwater flow D _gs, middle pressure gate bar steam loss D _kf, high-pressure door bar steam loss D _af, reheating spray flow D _zr, overheated spray flow D _grand the high import and export parameter added at different levels, calculate extraction flow;

(3.2) according to mass conservation formula D _{fw_js}=D _ns+ d ₁+ d ₂+ d ₃+ d ₄+ D _kf+ D _af-D _zr-D _gr, with extraction flow d ₁..., d ₄computing formula simultaneous, calculates feedwater flow initial value and is set to D _gs;

(3.3) | D _{fw_js}-D _gs| during > difference in flow limit value, D _gs=D _{fw_js}, and enter step (3.2), iterative computation, until | D _{fw_js}-D _gs| < difference in flow limit value, exports now feedwater flow, and then calculates main steam flow G ₁₀, t/h.

3. the power station main steam flow on-line calculation method based on flux modification according to claim 1, is characterized in that: described step (7) concrete steps are:

(7.1) steam turbine first stage pressure P is affected via Data Input Interface acquisition ₀each parameter: heat regenerative system extraction pressure P at different levels _j(j=1,2 ... 8), it can be used as the input parameter of pivot analysis, be designated as X _old(n × m), sampling number is measured in n representative, and n represents measurement attribute number;

(7.2) according to following formula, standardization input data, obtain X (n, m)

$x(i,j)=\frac{x_{\mathrm{old}}(i,j)-\mathrm{average}\left(x_{\mathrm{old}}(:,j)\right)}{\mathrm{std}\left(x_{\mathrm{old}}(:,j)\right)}$

In formula: i=1,2 ..., n, j=1,2..., m, average (x _old(:, j)) represents the average of a jth variable down-sampling point, std (x _old(:, j)) represents the standard deviation of a jth variable down-sampling point;

(7.3) the covariance matrix COV (X) of X (n, m) is calculated, and eigenvalue λ _iwith proper vector p _i

$\mathrm{COV}\left(X\right)=\frac{X^{T}X}{m-1}$

COV(X)p _i＝λ _ip _i

Carry out SVD decomposition to COV (X), and carry out pivot analysis to X, choose k pivot, k is constant, determines that the process of score matrix T so far pivot analysis terminates;

(7.4) by k pivot obtaining and first stage pressure sampled value P ₀respectively as the input and output of support vector machine, wherein n/2 group data are as training data, and n/2 group is as test data in addition, first by inputoutput data standardization;

In high-dimensional feature space, construct optimum linearity decision function y (x)=sgn [w ψ (x)+b], take following formula objective function:

$\min \underset{w,b,\mathrm{\&xi;}}{J}(w,\mathrm{\&xi;})=\frac{1}{2}{w}^{T}w+C\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{\mathrm{\&xi;}}_i$

In formula, constraint condition w is weight factor, and C is penalty parameter, and b is deviate, C ∈ R ⁺punishment parameter, ξ _i=[ξ ₁..., ξ _n] ^t, that a Nonlinear Mapping can x _ibe mapped to the feature space of higher-dimension from the input space thus realize non-linear regression the input space and be converted into linear regression in high-dimensional feature space, constrained optimization can be converted into unconstrained optimization by structure Lagrange function, according to KKT condition the optimization problem solved finally can be converted into and solve linear equation:

$\left[\begin{array}{cc}0& I_v^T\\ I_v& \mathrm{\&Omega;}+c^{-1}I\end{array}\right]\left[\begin{array}{c}b\\ a\end{array}\right]=\left[\begin{array}{c}0\\ y\end{array}\right]$

Wherein, y=[y ₁..., y _n] ^t; I _v=[1 ..., 1] ^t; A=[a ₁..., a _n] ^t; Ω={ Ω _ij| i, j=1 ... n}; k () is kernel function, can select Radial basis kernel function K (x herein _i, x _j)=exp [-|| x _i-x _j|| ²/ (2 σ ²)];

(7.5) calculated value of support vector machine part of detecting is hard measurement income value, enters step (5), replaces first stage pressure sensor values by this result.

4. the power station main steam flow on-line calculation method based on flux modification according to claim 2, is characterized in that: described step (3.1) concrete steps are: calculate extraction flow

$d_1=\frac{D_{\mathrm{fw}}(h_{w1}-h_{w2})}{(h_1-h_{s1})}$

$d_2=\frac{D_{\mathrm{fw}}(h_{w2}-h_{w3})-d_1(h_{s1}-h_{s2})}{(h_2-h_{s2})}$

$\begin{array}{c}{d}_3=\frac{D_{\mathrm{fw}}(h_{w3}-h_{w4})-(d_1+d_2)(h_{s2}-h_{s3})}{h_3-h_{s3}}\\ -\frac{D_{\mathrm{af}}(h_{\mathrm{gin}}-h_{s3})+D_{\mathrm{kf}}(h_{\mathrm{zin}}-h_{s3})}{h_3-h_{s3}}\end{array}$

$\begin{array}{c}{d}_4=\frac{(D_{\mathrm{fw}}+D_{\mathrm{zrps}}+D_{\mathrm{grps}})(h_{w4}-h_{w5})}{h_4-h_{w5}}\\ -\frac{(d_1+d_2+d_3)(h_{s3}-h_{w5})}{h_4-h_{w5}}\end{array}$

D _kf＝k ₁*D _fw+k ₂

D _af＝k ₃*D _fw+k ₄

H _wi, h _wi+1, h _i, h _siwhat represent that this section draw gas respectively goes out saliva enthalpy, import water enthalpy, and draw gas enthalpy, hydrophobic enthalpy, and enthalpy is obtained by water vapour computing formula by pressure and temperature;

D _kffor middle pressure gate bar air leakage; D _affor high-pressure door bar air leakage; Coefficient k ₁, k ₂, k ₃, k ₄by thermal test gained; D _zrpsfor reheating spray flow; D _grpsfor overheated spray flow, if cross hot water spray to export extraction, then D from No. 1 high adding _grpsget 0.