http://www.paper.edu.cn

Thermal Properties and Combustion Modeling of Dimethyl Ether Xibin Wang，Deming Jiang，Longbao Zhou （School of Energy and Power Engineering, Xi’an Jiaotong University, Xi’an, 710049,China） E-mail：[email protected]

Abstract Based on dimethyl ether(DME) engine experiments, a turbulent combustion model was developed to study the combustion characteristics of DME engine. It consisted of two sections. First, the correlation of DME thermal properties to temperature are required for numerical simulation of DME spray and combustion process, but cannot be obtained instantly through measurement. So molecule theory was applied here to estimate the correlation of the DME thermal properties to temperature, including latent enthalpy, surface tension, thermal conductivity and diffusion coefficient. And other DME thermal properties measured by experiments, including vapor pressure and viscosity coefficient, are also collected here. Then the combustion model was developed. The model took into account the influence of turbulence on combustion by coherent flamelet model(CFM) in addition to the original chemical kinetic model, so closer to the experiments and can be applied to the qualitative analysis of DME engine combustion. Numerical simulation indicated that, in engine condition DME spray has short tip penetration than diesel oil, with almost no wall impingement occurring. The upper edge of spray near the nozzle ignite first for the appropriate fuel concentration and higher temperature. In diffusion combustion the combustion is slow and lasts long for the slow air-fuel mixture formation. The results of numerical simulation are in good agreement with experiments. Keywords：dimethyl ether(DME); thermal properties, molecule theory, CFM model, combustion characteristics state have not yet been measured thoroughly by experiments. And in engine use, the DME fuel is first pressurized to be liquidated and then is injected in liquid state into cylinder, then the spray penetrate and the fuel droplets begin to breakup, evaporate, mix with air to form combustible mixture and begin combustion process. To simulate such a complicated process, the liquid thermal properties are needed. Now the researchers have begun the research work of measuring the thermal properties, and some properties have been measured and is available now. But these data are still not enough since numerical simulation requires a lot of thermal properties, especially its correlation to temperature. So other techniques should be applied to solve such problems. Here, apart from the method of looking for the measurement data of DME thermal properties, the author also estimated DME thermal properties with molecule theory.

Since the low cost of production of dimethyl ether(DME) became reality in 1995, the research of DME as alternative diesel fuel has become more and more concerned. Compared with diesel oil, diesel engine fueled with neat DME shows low emissions and noise and good efficiency, no soot emission in all the engine operations. But mechanism of its excellent performance is still confusing, and many researchers have done a lot of work in this field. In recent years, with the development of CFD technology, numerical simulation is becoming one important method to study the in-cylinder spray and combustion process of internal combustion engine. Like diesel fuel in diesel engine, DME fuel is also injected into the chamber by fuel injection system, and intake air swirl is also applied to improve the fuel-air mixture formation. Then liquid fuel droplet will breakup and evaporate and mix with air after injection and there is strong turbulence in the cylinder also. So the breakup process and the influence of turbulence on fuel-air mixing and combustion should be taken into account. Base on these considerations, the author developed a turbulent combustion model with KIVA 3 program to study the combustion process of DME engine.

1.1 Available DME properties The general properties of DME is listed in table 1. 【 】 【 】 Table 1 general properties of DME 1 2 type DME formula CH3OCH3 molecule weight 46 oxygen content-mass% 34.8 stoichiometric air fuel ratio 9.1 lower heating value(kJ/kg) 28800 cetane number >55 liquid density@15℃(g/cm3) 0.668 -24.9 boiling point℃ viscosity@20℃（μPa·s） 0.131

1 Thermal properties of DME Before the establishment of DME combustion model, the thermal properties should be obtained first. But unlike diesel oil, DME is in gas state in normal pressure and temperature, and the research of DME as alternative diesel fuel has been developed since a few years ago. And the thermal property of DME in liquid

-1-

http://www.paper.edu.cn

vapor pressure@25℃(MPa) latent heat @20℃(kJ/kg) critical pressure(MPa) critical temperature ℃ ignition temperature℃ ignition limit (vol% air) liquid thermal conductivity @20℃(W/m.K)

expressed as follows,

0.51 410 5.22 127 235 3.4~18 0.131

pr = p / pc Tr = T / Tc V r = V / Vc

Table 2 Measurement data of DME vapor pressure T/K Pvap/kPa T/K Pvap/kPa 538.4666 562.4703 594.4856 628.7713 667.8232 701.1717 737.0115 776.4311 818.3639 866.8156 912.1426 959.3064 1009.149 1059.903

321.3673 323.3765 325.3924 327.3773 329.3943 331.398 333.4165 335.4334 337.4293 339.4523 343.4569 345.4494 347.4643

1112.926 1167.85 1224.444 1282.35 1343.844 1407.069 1473.077 1540.901 1608.674 1682.727 1834.244 1913.415 1996.017

227.218 232.277 237.471 243.152 248.297 253.138 258.153 263.141 268.154 273.157 278.156 283.151

258.928 240.191 221.92 209.52 198.576 189.18 178.413 169.869 162.432 154.939 148.215 142.287

288.152 293.145 298.157 303.162 308.148 312.534 317.518 322.498 327.501 332.493 337.467 342.448

137.681 131.523 127.443 122.282 116.252 111.857 108.217 103.205 99.423 96.688 93.419 90.759

log10 µ = −5.7282 + 631.031 / T + （3） 1.453 × 10 − 2 T − 1.8225 × 10 −5 T −5 (mPa ⋅ s)

The average error of this correlation is about 0.52%，and the maximum error is 1.82%。

ln ( p r ) = −1 （1） − 6.57349τ + 0.61057τ 1.5 − 0.17752τ 2.5 Tr where, τ = 1 − Tr ，Tr and pr are reduced temperature and reduced vapor pressure individually. Besides, Vr is

)

reduced specific volume, all the three parameters

【 】

viscosity 4 The temperature range of DME liquid viscosity measurement is 227.218K~342.448K. The measurement data are listed in Table 3. Table 3 Measurement data of DME liquid viscosity T/K μ(mPa·s) T/K μ(mPa·s)

Then a correlation of viscosity estimation was proposed

Then a correlation of Wagner type was proposed and may be applied to temperature range of 293～ 400K:

(

（3）

（4） The subscript character c indicates value at critical point. The average error of this correlation is about 0.023%，and the maximum error is 0.096%。

During the spray and combustion process, the in-cylinder temperature varies from 300K to over 2000K. So for numerical simulation of spray and combustion process, many thermal properties over a wide range of temperature, including viscosity, vapor pressure, latent heat, surface tension, heat transfer coefficient and diffusion coefficient and so on, are required and cannot be considered as fixed values, therefore the correlations of thermal properties to temperature are required. The correlations of DME vapor pressure and viscosity to temperature have already been measured in Xi'an Jiaotong University and are listed as follows. 【 】 vapor pressure 3 The temperature range of DME vapor pressure measurement is 293.7762K~347.4643K. The measurement data are listed in Table 2.

293.7762 295.3016 297.2731 299.2863 301.4904 303.2872 305.1383 307.1023 309.1083 311.3343 313.339 315.3346 317.3668 319.3339

（2）

are

-2-

1.2 Estimation of DME thermal properties with 【 】 molecule theory 5 Since no measurement data are available for other DME thermal properties, including latent heat, surface tension, heat conduction coefficient, diffusion coefficient and so on. Other techniques should be applied to estimate the other DME thermal properties. 1.2.1 The concept of molecule theory In the all-too-frequent situation in which no experimental value is at hand, the value must be estimated or predicted. Estimation may be base on theory, on correlations of experimental values, or on a combination of both. A theoretical correlation may serve adequately in specific cases, such as numerical simulation. An ideal system for the estimation of physical

http://www.paper.edu.cn

properties would (1)provide reliable properties at any temperature, pressure and composition (for mixtures), (2)require a minimum of input data, (3)choose the least-error route, (4) indicate the probable error, (4) minimize computation time. Few of the available methods approach this ideal, but some serve remarkable well. In some case molecules theory is one of the few methods. In molecule theory, it is thought that the physical properties of every substance depend directly on the nature of the molecules of the substance, that is, molecule structure, size and motion speed. Estimation of fluid thermal properties should take into account the following parts, (1) corresponding theory or the law of corresponding states, i.e. the fluid thermal properties are related to the critical properties. So the reduced property is commonly expressed as a fraction of the critical property: pr, Tr and Vr , as mention above in equation 2,3 and 4. (2) Nonpolar and polar molecules. Nonspherical and weakly polar molecules fit rather well, but the deviation should be accounted for by the consideration of acentric factor. (3) Molecule structure. All macroscopic properties are related to molecule structure, i.e. atoms, atomic groups bond types etc. Here the unknown thermal properties of DME are estimated with molecule theory, including latent heat, surface tension, liquid heat conduction coefficient and 【 】 diffusion coefficient 6 。 . 1.2.2 latent heat

+

(J/mol)

10

19395.97

421.6515

20

18755.57

407.7297

30

18076.9

392.9762

40

17353.35

377.2467

50

16576.18

360.3518

60

15733.57

342.034

70

14808.75

321.9294

80

13776.74

299.4943

90

12597.3

273.8543

100

11198.16

243.4382

110

9423.177

204.8517

120

6774.035

147.2616

127

0

0

where, ρ Lb =

4

⎛ 1 − Tr ⎜⎜ ⎝ 1 − Tbr

⎞ ⎟⎟ ⎠

4n

(7)

ρ b / M （g /(cm3·mol)），Tbr is boiling

temperature，M is mole weight. For DME, Tbr =-24.9℃， ρb =0.72956 g/cm3 【10】，then

ρ Lb =0.72956/46=1.586e-2（g/cm3·mol） Tbr=(273.1-24.9)/400.1=0.62025 [P] is the parachor, and depends on the molecule structure. Since the parachor value is CH3-(55.5),-O-(20.0), then the parachor of DME is [P]=55.5+55.5+20=131 For ethers, the exponent n is suggested as 0.29. Then the correlation of DME surface tension to temperature can be obtained as equation (8).

For DME,ω＝0.2，R=8.3143 J/mol.K，Tc=400 K, then the above correlation may be changed as equation (6) 0.354

434.8543

σ = ([P ]ρ Lb )

∆H v 0.354 0.456 = 7.08(1 − Tr ) + 10.95ω (1 − Tr ) RTc

4

20003.3

The measurement values of latent heat at -20℃ and 20℃ are 460kJ/mol and 410kJ/mol，and the errors are -0.121% and -0.554 % individually. 1.2.3 surface tension Surface tension is predicted by the correlation presented by Goldhammer

The latent heat of vaporization is the difference between the enthalpy of the saturated vapor and that of the saturated liquid at the same temperature. Pitzer have shown that tha latent heat can be related to T, Tc and acentric factorω.For 0.6＜Tr≤1, Pitzer presented the following correlation of latent heat

∆H v = 2.3546 × 10 (1 − Tr ) 0.456 7.283 × 10 3 (1 − Tr )

0

4n

⎛ 1 − Tr ⎞ ⎛ 1 − Tr ⎟⎟ = 18.63⎜⎜ σ = ([P ]ρ Lb ) ⎜⎜ ⎝ 1 − Tbr ⎠ ⎝ 1 − Tbr 1.16 = 57.29(1 − Tr ) (dyn / cm) 4

(6)

The calculated data of DME latent heat are listed in Table (4) Table 4 Calculated data of DME latent heat ΔHv T/℃ J/mol kJ/kg -33 21819.36 474.334

= 57.29 ×10−3 (1 − Tr )

1.16

⎞ ⎟⎟ ⎠

1.16

(8)

（N/m）

The calculated data of DME surface tension are listed in Table 5. Table 5 Calculated data of DME surface tension ℃ σ(10-3N/m) ℃ σ(10-3N/m)

-30

21664.46

470.9666

-20

21134.48

459.4453

-20

17.93814

60

7.210012

-10

20581.68

447.4279

-15

17.23232

65

6.589669

-3-

http://www.paper.edu.cn

-10

16.53047

70

5.977286

-5

15.8327

75

5.373449

0

15.13915

80

4.778843

5

14.44996

85

4.19429

10

13.76527

90

3.620789

15

13.08525

95

3.05959

20

12.41008

100

2.512298

25

11.73993

105

1.981069

30

11.07502

110

1.46896

35

10.41557

115

0.980707

40

9.761837

120

0.524811

45

9.11409

125

0.122711

50

8.472635

127

0

55

7.837814

20 30 40 50 60 70 80 90 100 110 120 127

From Table (6), the calculated value of DME thermal conductivity at 20℃ is 0.138W/m. K，and the measured value is 0.131 W/m. K，error is about 5.3%. 1.2.5 diffusion coefficient The diffusion coefficient of DME in air is estimated with correlation (12) presented by Fuller, Schettler 和 Giddings.

Since no measured data of DME surface tension are available now, the error cannot be predicted. But in general case the error rarely exceed 5~10% at low pressures, even lower for nonpolar molecule like DME. So the calculated surface data are adequate when no experimental data are available. 1.2.4 liquid thermal conductivity Sato method was applied here to estimate the liquid heat conduction coefficient at any temperature. First the thermal conductivity λ Lb at normal boiling point Tb was calculated with equation (9)

λ Lb =

1.11 (W/m. K) M 1/ 2

D AB =

λL =

1.11 × 10 M 1/ 2

（9）

3 + 20(1 − Tr ) (W/m. K)（10） 3 + 20(1 − Tbr ) 2 / 3

After the DME parameters are applied, equation (10) may be expressed as equation (11)

)

λ L = 1.2213 × 10 −5 3 + 20(1 − Tr )2 / 3 (W/m.K)（11）

]

(cm 2 / s)

8.21 × 10 −6 T 1.75 = (cm 2 / s ) （13） 10 p

Like surface tension, no measured data are available, and error prediction is improbable now. But Nain 和 Ferron discovered that the error of this equation is relatively large only for compound of polar molecules, but for compounds of nonpolar molecules like DME and air, estimations are in good agreement with measurements and can meet the demand of numerical simulation.

Table 6 Calculated data of DME thermal conductivity

-33 -30 -20 -10 0 10

+ (Σv )

1/ 3 2 B

1/ 2

MB=28.96 After all these parameters are applied, equation (12)may be expressed as equation (13)

DDME − air

The calculated data of DME thermal conductivity are listed in Table 6.

℃

[

(10 p) (Σv )

1/ 3 A

⎡M A + M B ⎤ ⎢ ⎥ ⎣ M AM B ⎦

( Σv ) A = 51.77 ( cm3 / mol ) , MA=46.08 for air, (Σv )B = 19.7(cm 3 / mol ) ，

2/3

(

0.00143T 1.75

（12） where ,T and p are temperature(K) and pressure(MPa), MA and MB are molecule weights of two component. (Σv ) A and (Σv )B are atomic diffusion volumes of two components, depending on the molecule structure. From the table of atomic diffusion volume of atom and structures, then by simple calculation, the atomic diffusion volume may be obtained. For DME,

where, M is mole weight. Then the thermal conductivity at any temperature can be calculated with equation (10) −3

0.138047121 0.131626135 0.124980187 0.11807398 0.110861206 0.103278861 0.095237028 0.086598637 0.077133359 0.066386631 0.053103749 0.03663928

thermal conductivity(W/m.K) 0.169245025 0.167582229 0.161961258 0.156211253 0.150319481 0.144270829

2 combustion model Here, the combustion model consists of spray model, premixed combustion model and diffusion combustion model. Since the ignition point of DME fuel is very low and DME combustion is in a stage of high

-4-

http://www.paper.edu.cn

temperature kinetics at temperature above 740K, and the in-cylinder temperature during the injection process is well above 740K, so the self ignition model is not applied right now. 【 】 2. 1 spray model 2 Here, the Wave breakup model was applied to predict the liquid droplet breakup process. In this model, The radius of child drop is

B0 Λ ⎧ ⎪ ⎪ ⎪ ⎧ 3π a 2 Λ ) 0.33 r=⎨ ) ⎪( ⎪ min ⎪ 2Ω ⎨ 2 ⎪ ⎪ ( 3a Λ )0.33 ⎪ ⎪⎩ 4 ⎩⎪

very low, the fuel may have already evaporated at the transition. So the initial flame area is supposed to be 1% of the surface area of the cell at the transition. Numerical simulation indicated that the model is relatively insensitive to the initial value of flame area. 2.4 Diffusion combustion model 【 】 The coherent flamelet model(CFM) 9 proposed by Mark P. Musculus and C. J. Rutland was applied here to simulate the diffusion combustion. The CFM model treats combustion as a group of laminar flamelet which are imbedded in a turbulent flow field. The cell in diffusion burn stage is divided into two sides by the thin flame sheet, oxygen on one side while fuel on the other side. Both the fuel and oxygen are transported to the flame. Since the reaction rate is very fast compared with the mixing rate in diffusion stage, the local reaction rate may be obtained from the fuel and air transportation. So the overall reaction rate is found by integrating the local reaction rate per unit area over all flame area and is given by equation (19) ρ&i = ρ i ,∞V D ,i Σ （19）

B0 Λ ≤ a

(14)

B0 Λ > a, once

where, Ω and Λ are the maximum wave growth rate and the corresponding wavelength. a is drop radius. B0=0.61, model constant. The breakup time τ is obtained by equation (15) 0.5 ⎧ ⎡ ρ1 a 3 ⎤ ⎪1.72 B1 ⎢ We2 > 6.0 ⎥ ⎪ ⎣ 2σ ⎦ （15） τ =⎨ a ⎪B 0.5 We2 / Re 02.5 > 0.5 ⎪⎩ 1 U ( ρ 1 / ρ 2 )

where ρ i ,∞ is the reactant species density far from the flame front,Σ is the local flame area density, and VD,i is the volume rate of consumption of that reactant per unit flame area, it may be expressed as equation(20) for fuel,

VD, f = ρD

where,，We2 and Re2 are Weber number and Reynold number of gas. ρ1 and ρ2 are density of liquid and gas. σ is surface tension. B1=1.73, model constant.

/ ρ f ,∞ （20）

% ∂ % %%) = ∇ • ηt ∇ S + ( S ) + ∇ • ( Su Sc ρ ∂t （21） d Σ% d Σ% ρ p −ρ d dt dt where ~ means the Favre(mass) averaged quantities, ηt

ϖ&prem = A[Fuel ][O2 ]3 exp(− E A / RT ) （17）

The constants for the model are A=7.2e20, the activation temperature EA/R=1.193e4(K). 2.3 Transition to the diffusion burn The transition to the diffusion burnn stage occurs when the kinetics become very fast compared to the rate of mixing, and the fuel no longer has time to mix with air before combustion. A critical Damkoler number Da is applied to describe the transition to diffusion burn stage.

A exp(− E A / RT ) ε /k

dx

where Yf is the mass fraction of the fuel far away from the flame, D is the bulk diffusion coefficient. From equation (19), we can know that the main object of diffusion model is to obtain the local flame area density Σ. For convenience, a new variable is defined for transport such that S = ρΣ . With some manipulation using this substitution, the transport 【 】 equation becomes 9

2.2 High temperature premixed combustion model In this model, at the early stage of combustion the combustion is not controlled by turbulence, but by chemical kinetics for the premixed fuel-air mixture. The high temperature premixed combustion is expressed with a global reaction (16), CH 3 OCH 3 + 3O 2 → 2CO 2 + 3H 2 O （16） and the inverse time scale is given by (17)

Da ≡

dY f

and Sc are turbulent absolute viscosity and Schmidt number. The terms with subscript p and d are production term and destruction term of flame area, can be obtained from the local flame area, turbulence and mixture parameters.

（18）

3 Mesh generation

where, k and ε are local turbulent energy and dissipation rate individually. Once the critical Da is reached locally, a transition to the diffusion burn occurs. The flame area must be initialized during the transition. Since the critical temperature of DME is

The experiment and simulation are both for TY1100 diesel engine, at a speed of 1800r/min. Since the numerical simulation is very time-consuming, considering that the DME engine uses nozzle of 5 holes, and the holes are nearly located uniformly around the

-5-

http://www.paper.edu.cn

chamber, the mesh was generated just for 1/5 of the cylinder to save computation time. The generated mesh is shown as figure 1.

agreement with the experiment. 4.2 Analysis of combustion process of DME engine using CFM modeling From the above analysis, we learn that the simulation using CFM model can predict the combustion process. In the following is the qualitative analysis of spray and combustion process of DME engine using CFM model. 4.2.1 Spray penetration and fuel density distribution of DME engine Fig 3 is the simulated spray and fuel density distribution of DME engine, (a) is at TDC and (b) at 10 ℃A ATDC. The contour lines indicated the difference of fuel density, h means highest density area and l means lowest one. Fig (3) indicates that at the early stage of injection period, DME spray has longer penetration for the lower temperature and pressure. But once the combustion begins, the in-cylinder temperature and pressure rise rapidly, makes the spray penetration decrease sharply. Simulation also indicated that DME spray penetration is shorter than that of diesel oil, almost

Fig.1 Mesh generation of TY1100 engine

4 Research of combustion process of DME engine with numerical simulation

7

With the development of computer technology and computational fluid dynamics, numerical simulation has become one powerful tool for research of combustion process in internal combustion engine. Here qualitative analysis was done for the combustion process of DME engine. 4.1 Comparison between kinetic model and CFM model for DME engine In the numerical simulation of combustion of IC engine, kinetic model is often used, i.e. a global reaction is used to calculate the heat release during the combustion period. But in reality the combustion process is very complicated, has hundreds of kinetic reaction. Besides, strong turbulence plays an important role in the combustion process. Fig. 2 shows the comparison between kinetic model, used in KIVA 3 program, and CFM model used in this paper. In Fig. 2, the peak pressure of original model is smaller than that of CFM model and experiment, and the former pressure curve is far below the latter ones. This leads to lower in-cylinder temperature and heat release rate. Change of model constants doesn't make change to this situation. On the other hand, CFM model have taken into account the influence of turbulence on mixing and combustion at diffusion burn stage and been in close

experiment original model CFM model

6

P/MPa

5 4 3 2 1 0 -60

-30

0

30

60

90

120

deg(CA)

Fig.2

Pressure comparison among CFM model, original model and measurement

no wall impingement phenomena occurs. At 10 ℃ A ATDC, nearly the end of injection period, the evaporated DME fuel distributed only small area in the middle of combustion chamber. 4.2.2 Self-ignition of DME engine

-6-

http://www.paper.edu.cn

flame began to propagate rapidly. In DME engine condition, for its very low critical temperature and very high vapor pressure, so the spray penetration is rather short and almost no wall impingement occurs. The injected DME fuel soon evaporated after the end of injection duration. Most of the evaporated fuel distributed in the middle region of the chamber and formed very dense mixture, and the mixture in the region near the wall was very rare. As the combustion went on, the flame spread out from the ignition point. Because the mixture in the middle region was too dense, oxygen was soon exhausted and the diffusion combustion stage began. Distinct boundary was formed between the dense region and the rare region. The flame was maintained by the transportation of oxygen and fuel from both side of the boundary, and the diffusion combustion speed was slow, and diffusion combustion last longer than ordinary diesel engine.

Fig 3

Spray and fuel density distribution for DME engine Fig 4 is the in-cylinder temperature distribution of DME engine at TDC. The self ignition occurred just before TDC, so fig. 4 indicated the self ignition phenomena of DME engine. By comparison fig. 3 and fig. 4, we can see that the site of self ignition is located near the nozzle hole at the upper edge of the spray. Such result agrees with the experiment result in reference [7].The reason lies in that, because of the upward movement of air flow led by piston movement, a large amount of the evaporated fuel lies in the upper region of spray, so this region can first reach the appropriate air-fuel ratio and temperature then self ignition occurs first. For region too close top the hole with too dense fuel and low temperature for the evaporation of fuel, and other regions with too little fuel, all these have negative influence on reaction speed.

Such a situation can be explained by the following reasons: the critical temperature of DME is very low, and the DME spray evaporates quickly in engine condition. On heavy load condition, the injected fuel quantity is large, it is difficult to obtain fuel- air mixture of good quality all around the chamber for the inadequate spray penetration. So the mixture in the middle region is too dense and that near the wall is too rare. The above simulation indicated that more research work should be done to acquire the optimum match of chamber, injection system and air flow movement.

Fig. 4 Temperature distribution of DME engine at TDC 4.2.3 Combustion characteristics of DME engine Fig.5 shows the temperature distribution(a) and equivalent ratio distribution(b) at 20 ℃ A ATDC. Comparison of (a) and (b) indicates that the high temperature area in (a)coincides with the boundary between the dense area and rare area in (b). After the ignition, since there is already large amount of mixture with appropriate equivalent ratio,

Fig.5 Temperature(a) and equivalent ratio(b) distribution of DME engine at 20℃A ATDC

5 Conclusion From the above calculation and analysis, we can reach the following conclusions.

-7-

http://www.paper.edu.cn

1. Experimental data of DME thermal properties, including vapor pressure and viscosity coefficient, are collected. 2. Molecule theory was applied to estimate the correlation of the DME thermal properties to temperature, including latent enthalpy, surface tension, thermal conductivity and diffusion coefficient. The error level of estimation is satisfactory. 3 CFM model considered the influence of turbulence on combustion, can demonstrate more precisely the combustion characteristics than kinetic model. 4 Simulation indicated that the spray penetration of DME engine is shorter than diesel fuel, almost no wall impingement. DME fuel distributed only in the middle region of the chamber. 5 Simulation indicated that the upper edge of spray near the nozzle ignite first for the appropriate fuel concentration and higher temperature. In diffusion combustion the combustion is slow and lasts long for the slow air-fuel mixture formation. Acknowledgment This work was supported by (1) the Research Fund of the Doctoral Program of Higher Education of China (RFDP No:20020698044) (2)The state key project of fundamental research plan titled as “New generation of Engine Combusiton Principle and Approach to Application of Alternative Fuels.” Grant No:2001CB20908

-8-

Reference: 【1】Jiang Deming. Combustion and emission of internal combustion engine. Xi'an:Xi'an Jiaotong University Press.2001. 【2】Sorenson S.C. Dimethyl Ether in Diesel Engines: Progress and Perspectives. Transaction of ASME, vol.123,pp. 652~658, July 2001. 【3】Wu Jiangtao, Liu Zhigang, Pan Jiang, et al. Research of DME saturated vapor pressure Xi'an: Journal of Xi'an Jiaotong University.2003 【4】Wu Jiangtao, Liu Zhigang, Bi Shengshan,et al. Viscosity of saturated liquid dimethyl ether from 227 to 343K. Journal of Chemical & Engineering data [J] (accepted).2003. 【5】Reid R.C. Prausnitz J.M. and Sherwoods T. K. The Properties of gases and liquids[M].New York:McGraw-Hill Book Company. 1977. 【6】Vargaftik N.B. Tables on the thermophysical properties of liquids and gases : in normal and dissociated states[M]. Washington : Hemisphere Pub. Corp.19 【7】Wang Hewu. Experimental and theoretical studies on performance and combustion characteristics of direct injection diesel engine fueled with dimethyl ether. Xi'an:：Xi'an Jiaotong University doctoral thesis. 2001(in Chinese) 【8】Wang Xibin, Jiang Deming, Wang Hewu. The Spray Model and Experimental Research of Dimethyl Ether. Xi'an:Journal of Xi'an Jiaotong University.2002(9) 【9】Mark P. Musculus and Christopher J. Rutland. Coherent flamelet modeling of diesel engine combustion . US:Combust. Sci. and Tech. pp.295~337,1995. 【10】Yao Mingfa, Xu sidu, Xu Junfeng, et al. An investigation on the combustion characteristics of dimethyl ether(DME). Tianjin: Journal of CSICE. 2001(3).