CN106156518A - A kind of method building fuel skeleton model of oxidation - Google Patents

Abstract

A kind of method building fuel skeleton model of oxidation of the present invention, belongs to combustor structure applied technical field.The method includes the first step, sets up C₀–C₁Submodel；Second step, sets up lump fuel molecule submodel, including setting up C₂–C₃Submodel and set up C₄–C_nLumped model, covers the low temperature condition range to high temperature；3rd step, determines functional group's submodel；4th step, uses genetic algorithm optimization C₄–C_nSubmodel.The fuel skeleton model of oxidation built can in broad condition range the delay period of Accurate Prediction different fuel, key component concentration, heat liberation rate, heat release rate, flame propagation velocity and flame-out extensibility；The final scale of model is less, can directly and the burning of multidimensional CFD model coupled simulation real engine and discharge behavior；Use genetic algorithm that reaction rate is optimized, the structure time of final mask can be shortened；Only need to build C₄–C_nSubmodel and functional group's submodel, can apply this method to build the skeleton model of oxidation of different fuel.

Description

A kind of method building fuel skeleton model of oxidation

Technical field

The invention belongs to combustor structure applied technical field, relate to a kind of side building fuel skeleton model of oxidation Method.

Background technology

Along with global energy crisis and the aggravation of environmental pollution, following electromotor faces raising fuel economy and fall The double challenge of low emission, chemical dynamic model being combined with multidimensional CFD model is to optimize having of present engine performance Effect means.Although detailed chemical dynamic model can provide the accurate information of fuel combustion, but owing to it is in large scale, nothing Method directly couples with multidimensional CFD model.For solving this problem, researcher proposes serial of methods to reduce chemical kinetics mould The scale of type.But, the model obtained by these methods is only applicable to specific operation, and range of application is less.Therefore, in burning Numerical simulation field, needs a kind of new method badly and builds effective chemical dynamic model so that it is can be at broad condition range The oxidation characteristic of fuel in the multiple reactor of interior Accurate Prediction, can maintain again less scale simultaneously, can directly and multidimensional CFD mould Type is coupled.

Summary of the invention

For solving problems of the prior art, the invention provides a kind of side building fuel skeleton oxidation mechanism Method.

The concrete technical scheme of the present invention is:

The first step, sets up C₀–C₁Submodel.

Use different C₀–C₁Model prediction H₂、CO、CH₂The flame propagation velocity of O, delay period and fuel element concentration, The minimum model of Φ value is chosen as C with method of least square₀–C₁Submodel, as shown in formula (1):

$\mathrm{\Φ}=min\underset{r=1}{\overset{n}{\Σ}}{\left(\frac{\[{\mathrm{\η}}_r-{\mathrm{\η}}_r^{obs}\]}{{\mathrm{\σ}}_r^{obs}}\right)}^2---\left(1\right)$

In formula, η_rRepresent the mechanism predictive value to r target,Represent the experiment value of r target,R The standard deviation of the experiment value of target.

The C that final selected Curran seminar builds₀–C₁Model.

Second step, sets up lump fuel molecule submodel.

(1) C is set up₂–C₃Submodel

The detailed model of oxidation of desired fuel is carried out sensitivity analysis, identifies leading C₂–C₃Concentration of component develops, stagnant combustion Phase and the C of flame propagation velocity₂–C₃Reaction, forms C₂–C₃Submodel, this C₂–C₃Submodel includes 8 components and 22 reactions, Wherein 8 components are C₃H₇、C₃H₆、C₃H₅、C₃H₄、C₂H₆。C₂H₅、C₂H₄、C₂H₃。

(2) C is set up₄–C_nLumped model

Detailed model of oxidation to desired fuel, carries out path analysis with low temperature to the delay period of worst hot case for target:

1) desired fuel RH forms relevant fuel base R through dehydrogenation reaction.

2) at worst cold case, fuel base R first with O₂Reaction generates RO₂, RO₂QOOH, QOOH is generated through isomerization reaction Continue and O₂Reaction generates O₂QOOH, O₂QOOH discharges an OH base and generates C_nKET, C_nKET then passes through series chain branching reaction Generate less component, and be 2 reactions by the reaction lump of these chain components.

At worst hot case, fuel base R generates C through a series of β-cracking reaction₀–C₃Component and molecule of functional group, in order to Reduce the scale of final mask, be 1 reaction by these continuous print β-cracking reaction lump.

Its dominant response path is as follows:

RH+H=R+H₂ (R1)

RH+O₂=R+HO₂ (R2)

RH+OH=R+H₂O (R3)

RH+HO₂=R+H₂O₂ (R4)

Worst cold case:

R+O₂=RO₂ (R5)

RO₂=QOOH (R6)

QOOH+O₂=O₂QOOH (R7)

O₂QOOH=C_nKET+OH (R8)

C_nKET=> OH+ aldehyde+ketone+little molecule (R9)

Aldehyde/ketone+O₂=> less component (R10)

Worst hot case:

R=> C₀–C₃Component+molecule of functional group (R11)

3) affected by desired fuel molecular structure, above-mentioned course of reaction can be generated substantial amounts of isomers, in order to Reduce the scale of final mechanism, be a lumped component M by these isomers lumps.Wherein, the speed of lumped reaction is normal Number k ', is tried to achieve by formula (2):

$k^{\′}=\frac{\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{k}_i\left[M_i\right]}{M}---\left(2\right)$

In formula, k_iFor the reaction rate of i-th isomers, [M_i] it is the proportion of i-th isomers, M is lump Component, the isomers number that n is comprised by component M.

3rd step, determines functional group's submodel.

Analyzing the chemical kinetics characteristic of dissimilar desired fuel, sensitivity analysis based on second step and path are divided Analysis, build reflection desired fuel oxidation characteristic functional group's submodel (as the OH base effect in Aalcohols fuel functional group reactions, Saturated fatty acid methyl ester causes CO₂Generate functional group reactions earlier).

4th step, uses genetic algorithm optimization C₄–C_nSubmodel.

After determining response path, use genetic algorithm that the skeleton model of oxidation obtained is optimized.Optimizing Cheng Zhong, if skeleton model of oxidation can delay period in Accurate Prediction shock tube, then this model also is able to Accurate Prediction speed Delay period in press；If the predictive value of target components concentration coincide with experiment value in jet mixing reactor (JSR) Very well, then related component predictive value in other reactors and experiment value also coincide preferably.It addition, C₀–C₁Drilling of component Change, flame propagation velocity and flame-out extensibility are by C₀–C₁Submodel controls, it is not necessary to its further optimization.And C₂–C₃Submodel It is taken from detailed chemical dynamic model, without being optimized.Therefore, the present invention is with the delay period in shock tube and JSR In fuel, O₂、C₂H₂And C₂H₄Concentration be target, by genetic algorithm optimization C₄–C_nThe speed constant of reaction, its step is such as Under:

1) by N_kExperiment value under individual different operating mode is as desired value.

2) N is randomly generated according to initial mechanism_RIndividual primary response mechanism is as initial population.

3) use CHEMKIN program bag simulate each initial mechanism in step 1) in N_kDuring ignition lag under individual operating mode Between.

4) calculating simulation value and the relative error of experiment value, as fitness function.

5) according to fitness function, carry out selecting, intersect, make a variation, merge and non-dominated ranking and crowding calculate, and Produce new population, the newest N_RIndividual reaction mechanism.

6) by newly generated N_RIndividual reaction mechanism brings step 3 into) in, circulation is carried out, until producing the n-th generation population, and output The rate constants k ' of optimum lumped reaction.

The invention has the beneficial effects as follows: the fuel skeleton model of oxidation of structure can in broad condition range Accurate Prediction The delay period of different fuel, key component concentration, heat liberation rate, heat release rate, flame propagation velocity and flame-out extensibility；The scale that model is final Less, can directly and the burning of multidimensional CFD model coupled simulation real engine and discharge behavior；Use genetic algorithm to reaction Speed is optimized, and can shorten the structure time of final mask；Only need to build C₄–C_nSubmodel and functional group's submodel, Application this method builds the skeleton model of oxidation of different fuel.

Accompanying drawing explanation

Fig. 1 is the flow chart using the method to build fuel skeleton model of oxidation.

Fig. 2 is the predominating path figure of the fuel skeleton model of oxidation using the method to build.

Fig. 3 is flow chart based on genetic algorithm optimization speed constant.

Detailed description of the invention

The detailed description of the invention of the present invention is described in detail below in conjunction with technical scheme and accompanying drawing.

The present invention is used to build methyl caprate skeleton model of oxidation, as shown in Figure 1:

The first step, sets up C₀–C₁Submodel.

Use different C₀–C₁Model prediction H₂、CO、CH₂The flame propagation velocity of O, delay period and fuel element concentration, The minimum model of Φ value is chosen as C with method of least square₀–C₁Submodel, as shown in formula (1):

$\mathrm{\Φ}=min\underset{r=1}{\overset{n}{\Σ}}{\left(\frac{\[{\mathrm{\η}}_r-{\mathrm{\η}}_r^{obs}\]}{{\mathrm{\σ}}_r^{obs}}\right)}^2---\left(1\right)$

In formula, η_rRepresent the mechanism predictive value to r target,Represent the experiment value of r target,R The standard deviation of the experiment value of target.

Finally choose the C that Curran seminar builds₀–C₁Model is as substrate.

Second step, sets up lump fuel molecule submodel.

(1) C is set up₂–C₃Submodel

LLNL (Lao Lunsi National Laboratory) is built methyl caprate model of oxidation carry out path analysis and sensitivity Analyze, identify leading methyl caprate delay period, flame propagation velocity and C₂–C₃These reactions are carried out whole by the reaction of concentration of component Close, form C₂–C₃Submodel, final C₂–C₃Submodel comprises 8 component (C₃H₇、C₃H₆、C₃H₅、C₃H₄、C₂H₆。C₂H₅、C₂H₄、 C₂H₃) and 22 reactions.

(2) C is set up₄–C_nLumped model, as shown in Figure 2.

First with low temperature to the delay period of worst hot case methyl caprate as target, capric acid first is dominated by path analysis identification The dominant response path of ester delay period, the dominant response path according to path analysis information acquisition is as follows:

MD+H=MDJ+H₂ (R1)

MD+O₂=MDJ+HO₂ (R2)

MD+OH=MDJ+H₂O (R3)

MD+HO₂=MDJ+H₂O₂ (R4)

MDJ+O₂=MDJO₂ (R5)

MDJO₂=MDJOOH (R6)

MDJOOH+O₂=MDJOOHO₂ (R7)

MDJOOHO₂=MDKET (R8)

MDKET=> C₆H₁₃CHO+CH₂CO+OH+CH₃OCO (R9)

C₆H₁₃CHO+O₂=> C₃H₇+C₃H₆+CO+HO₂ (R10)

MDJ=> C₃H₇+2C₂H₄+MP2D (R11)

Wherein, the rate constants k ' of lumped reaction, formula (2) try to achieve:

$k^{\′}=\frac{\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{k}_i\left[M_i\right]}{M}---\left(2\right)$

In formula, k_iFor the reaction rate of i-th isomers, [M_i] it is the proportion of i-th isomers, M is lump Component, the isomers number that n is comprised by component M.

3rd step, determines functional group's submodel.

Analyze the oxidation characteristic of methyl caprate, based on sensitivity analysis and path analysis information, build reaction methyl caprate Unique oxidation characteristic is (such as CO₂Generate relatively early) functional group's submodel, its functional group's submodel is as follows:

MP2D+H=MP3J (R12)

MP3J+H=MP2DMJ+H₂ (R13)

MP3J+OH=MP2DMJ+H₂O (R14)

MP2DMJ=C₂H₃CO+CH₂O (R15)

C₂H₃CO+M=C₂H₃+CO+M (R16)

MP3J=C₂H₄+CH₃OCO (R17)

CH₃OCO=CO+CH₃O (R18)

CH₃OCO=CO₂+CH₃ (R19)

(4) genetic algorithm optimization C is used₄–C_nThe speed constant of reaction in submodel, as shown in Figure 3.

The methyl caprate skeleton model of oxidation built based on said method comprises 43 components and 151 reactions, in broadness Condition range T=500 1700K, p=0.1 5.0MPa andIn, this model to delay period, key component concentration, The predictive value of flame propagation velocity and flame-out extensibility and experiment value coincide good, its predict the outcome with experimental result the most such as Shown in table 1-4.It addition, this scale of model is less, the burning predicting real engine directly can be coupled with multidimensional CFD model And emission performance.

The comparison of table 1 delay period

The comparison of MD concentration in table 2JSR

The comparison of table 3 flame propagation velocity

Table 4 stops working the comparison of extensibility

Claims (1)

1. the method building fuel skeleton model of oxidation, it is characterised in that comprise the steps:

The first step, sets up C₀–C₁Submodel；

Use different C₀–C₁Model prediction H₂、CO、CH₂The flame propagation velocity of O, delay period and fuel element concentration, by minimum Square law chooses the minimum model of Φ value as C₀–C₁Submodel, as shown in formula (1):

$\mathrm{\Φ}=min\underset{r=1}{\overset{n}{\Σ}}{\left(\frac{\[{\mathrm{\η}}_r-{\mathrm{\η}}_r^{obs}\]}{{\mathrm{\σ}}_r^{obs}}\right)}^2---\left(1\right)$

In formula, η_rRepresent the mechanism predictive value to r target,Represent the experiment value of r target,The r target The standard deviation of experiment value；

Second step, sets up lump fuel molecule submodel；

(1) C is set up₂–C₃Submodel

The detailed model of oxidation of desired fuel is carried out sensitivity analysis, identifies leading C₂–C₃Concentration of component develop, delay period and The C of flame propagation velocity₂–C₃Reaction, forms C₂–C₃Submodel, this C₂–C₃Submodel includes 8 components and 22 reactions, wherein 8 components are C₃H₇、C₃H₆、C₃H₅、C₃H₄、C₂H₆。C₂H₅、C₂H₄、C₂H₃；

(2) C is set up₄–C_nLumped model

Detailed model of oxidation to desired fuel, carries out path analysis with low temperature to the delay period of worst hot case for target:

1) desired fuel RH forms relevant fuel base R through dehydrogenation reaction；

2) at worst cold case, fuel base R first with O₂Reaction generates RO₂, RO₂Generate QOOH, QOOH through isomerization reaction to continue With O₂Reaction generates O₂QOOH, O₂QOOH discharges an OH base and generates C_nKET, C_nIt is raw that KET then passes through the reaction of a series of chain component The component of Cheng Geng little, and be 2 reactions by these component lumps；

At worst hot case, fuel base R generates C through a series of β-cracking reaction₀–C₃Component and molecule of functional group, by these even Continuous β-cracking reaction lump is 1 reaction；

3) by above-mentioned reaction produce isomers lump be a lumped component M；The rate constants k ' of lumped reaction, by Formula (2) is tried to achieve:

$k^{\′}=\frac{\underset{i=1}{\overset{n}{\Σ}}{k}_i\[M_i\]}{M}---\left(2\right)$

In formula, k_iFor the reaction rate of i-th isomers, [M_i] it is the proportion of i-th isomers, M is lump group Point, the isomers number that n is comprised by component M；

3rd step, determines functional group's submodel；

Analyze the chemical kinetics characteristic of dissimilar desired fuel, sensitivity analysis based on second step and path analysis, structure Build functional group's submodel of reflection desired fuel oxidation characteristic；

4th step, uses genetic algorithm optimization C₄–C_nSubmodel；

With the fuel in the delay period in shock tube and jet mixing reactor, O₂、C₂H₂、C₂H₄Concentration be target, use lose Propagation algorithm optimizes C₄–C_nSubmodel, its step is as follows:

1) by N_kExperiment value under individual different operating mode is as desired value；

2) N is randomly generated according to initial mechanism_RIndividual primary response mechanism is as initial population；

3) use CHEMKIN program bag simulate each initial mechanism in step 1) in N_kIgnition delay time under individual operating mode；

4) calculating simulation value and the relative error of experiment value, as fitness function；

5) according to fitness function, carry out selecting, intersect, make a variation, merge and non-dominated ranking and crowding calculate, and produce New population, the newest N_RIndividual reaction mechanism；

6) by newly generated N_RIndividual reaction mechanism brings step 3 into) in, circulation is carried out, until producing the n-th generation population, output optimum The rate constants k ' of lumped reaction.