IC Engine is almost in a mature stage. Only problem is now to find ways to reduce emissions. National Laboratories and your mentioned Superuniversities are putting their heads together to reduce emissions in IC engines, Gas Turbines and other industries.

Accelerated Strategic Computing Initiative (ASCI)

Academic Strategic Alliances Program (ASAP)

For proposal questions contact: Lynn E. Rippe, alliances@llnl.gov

For technical questions contact: Dick Watson, Lawrence Livermore National Laboratory Ann Hayes, Los Alamos National Laboratory Charles Hartwig, Sandia National Laboratory

UCRL-MI-125901 Modified on 9/29/99.

Technical Note John,

Star-CD also has unstrucuted FV capabilities!

...............................Duane

Usally, I read Detroit Free Press. If you are interested then you can go to the library and find out. You can also go to specially Diesel Engine Companies directories and find out the persons to contact. If you look at Detroit Free Press of yesterday, you can find out about the 1.9 billion dollar lawsuit against GM in California. GM is claiming that the lawyers are after Automotive Engineers. My believe is that lawyers are after anybody where they can find faults.

John,

what do you mean by:

"The code should be in perfect condition to solve the user's problem when it arrives at the user's site. Otherwise, it should be returned."

I hope you don't mean that the disk is scratched or the maual has a coffee stain on it? (Pardon my sarcasm) We are not talking about buying a pair of shoes! How can a user anticipate apriori all of the needs for a CFD code and then evaluate the adequacy of a code for these needs at the time it is delivered. The process of CFD code development and CFD use is an iterative and growing one. The most important factor in the CFD industry is the commercial code developer's ABILITY AND WILLINGNESS to meet the customer's needs. That is an ongoing process NOT a package that arrives on the doorstep!

Regards,

Duane Baker

As a forewarning, I am an employee of Gamma Technologies, which writes and markets GT-Power, a competing engine simulation code to WAVE. I will try not to make this an advertisement.

I feel the need to point out that these codes are made to simulate the airflow and combustion in an engine mostly to simulate the performance of an engine, i.e., power, torque, fuel usage and acoustic characteristics. They can also be used for muffler matching, warm-up studies, performance when attached to specific vehicle, etc. Emission predictions are only part of the package. It is also well known to us and others that there are limitations to the emission predictions right now. KIVA may be a better code to use for this at this point in time.

Above all, I don't think any code can be used to design a product without experimental validation. The purpose of simulations is to reduce the cost and time spent on product development by building a model that can be validated and then optimizing the design on the computer, and then building a final design and verifying the simulations again. This will reduce the number of prototype which need to be built, hence reducing the cost and time spent in development. This is true of FEA, CFD or any specialized code. Simulations are not meant to eliminate the need for experimental testing. Maybe they can be sometime in the future when these codes have been perfected, but they can never eliminate typographical errors, misreading of blueprints, etc.

KIVA may be a better code to use for this at this point in time. I am impressed that at least somebody is admitting the truth. Yes, I agree with you that we still need to do experimental testing. Thank you T.J. Wanat.

(1). I think, you are right. (2). Whether we say it explicitly or implicitly, a product has to go through the validation, verification phases before it can be called a product. Otherwise, we can only call it a prototype at best. (3). Whether that product is a hardware product, or a software product, there are always standard procedures required to ensure the integrity of the product. (4). So, I think, the responsibility is on the developer's side to properly validate and verify the product for the intended range of applications. This common sense guideline is applicable to both the performance codes or 3-D CFD codes. (5). Experimental testing is one way to achieve this goal.

(1). You have raised the core issue of CFD software. (2). It is a much bigger issue and should be discussed separately. (3). I think, as a product, there is no difference between a pair of shoes , a computer game, or a CFD software. (4). What you mentioned was CFD consulting service.

You maybe a nice person but when you are coming to engineering technical issues, there is no compromise. Phillip Colella of Mechanical Engineering, UC Berkeley is working to develop software for IC engine performance in USA. You can also search for IC engine at University of Wisconsin, Madison, Engine Research Center and find out what software they are developing. I am a novice in this field. Getting angry with me will not solve problems. I am also attaching some information about KIVA below (KIVA is used in Sweden for developing Volvo engines.)--

MODELING OF SPRAY FORMATION,

IGNITION AND COMBUSTION IN INTERNAL

COMBUSTION ENGINES

Valerie Golovitchev, Assoc. Professor Niklas Nordin, PhD student

Jerzy Chomiak, Professor

This report presents the results of the third stage of the project with the aim of

developing a research CFD code for numerical simulations of reacting

multi-phase flows in piston engines, including soot and NOx formation. The

main goals of the project have been formulated in our previous annual report

1997.

The KIVA3 and KIVA-3V versions of the KIVA code are used for simulation,

with modified sub-models of fuel atomization, droplet evaporation, turbulence,

ignition, combustion, NOx and soot formation based on the RNG k-e

turbulence model. Such an approach allows for the treatment of

compressibility effects, detailed chemistry and turbulence/chemistry

interaction accounting for the reactant segregation, micro-mixing and sub-grid

scale reaction rates.

The neat dimethyl ether, DME, has been the fuel of choice due to two

reasons: there is substantial interest in the use of DME as a potential Diesel

fuel, the combustion chemistry of DME in air is relatively simple, albeit

almost unknown, and simulation can be done using a detailed chemical

mechanism. In addition to previous model verifications, the spray atomization

and evaporation models were tested once again using the constant volume

vessel experimental data obtained at the University of Hiroshima by Nishida

et al. The accuracy of the models at Diesel-like conditions is evaluated by a

comparison of predicted and experimental data on liquid and vapor phase

penetrations and ignition development. Then, the computational model was

applied to an axisymmetric bowl-in-piston engine geometry with a peak in the

center of the bowl to evaluate the performance of the turbocharged Volvo

AH10A245 DI Diesel engine fueled by DME. The results of numerical

simulations allowed to preliminary answer the question: "Is neat DME really

Diesel fuel of promise?"

The models of fluid dynamics, turbulence, chemical reactions, droplet

evaporation, collision and dispersion and so on, were used to reproduce the

whole picture of turbulent spray combustion. Model validation and

modifications are the subject of current work. As a part of the validation

procedure, the KIVA computer results were compared with those produced

with the help of the commercial AVL FIRE code.As a result of this work, the

FIRE code fuel library has been updated using our recommendations and

software.

Recently, the reduced, but still comprehensive (44 species, 195 reactions

including NOx chemistry), chemical mechanism for a generic aliphatic

hydrocarbon fuel has been developed and applied to n-heptane combustion.

The model has been supplemented by soot formation and oxidation processes.

Brief description of the soot model is given below.

Table 1. Large processes and rate constants of the soot model

No.

Formation subprocess

Reaction rate const.

Ai (mol,cc,s)

Ei (cal/mol)

1.

Fuel (n-Heptane) consumption

Detailed chemistry

2.

Precursor radical formation

k1 = A1 exp(-E1/RT)

7.00*10e11

1.20*10e5 3.

Acetylene formation

Detailed chemistry

4.

Precursor radical oxidation

k3 = A3 exp(-E3/RT)

1.00*10e12

4.00*10e4 5.

Acetylene oxidation

Detailed chemistry

6.

Soot particle inception

k5 = A5 exp(-E5/RT)

1.00*10e10

5.00*10e4 7.

Soot particle growth

k6 = A6 exp(-E6/RT)

4.20*10e4

1.20*10e4 8.

Soot particle oxidation

Nagle and Strickland-Constable

Constants:

kA = AA exp(-EA/RT)

2.00*10e1

3.00*10e4

kB = AB exp(-EB/RT)

4.46*10e-3

1.52*10e4

kT = AT exp(-ET/RT)

1.51*10e5

9.70*10e4

kZ = AZ exp(-EZ/RT)

2.13*10e1

4.10*10e3

Reasonable agreement between computational and experimental data

for DME ignition at high pressures in a constant volume bomb have

been achieved. The conditions of effective performance of DI Diesel

engines fueled with neat DME are formulated.

The general structure of the soot model formation and oxidation

developed looks similar to that developed by Fusco et al. (see

Proceedings of COMODIA 94, Yokohama, Japan, pp. 571-576

(1994)), but this does not belittle its merits because of all model

components are different. The main processes of our model are shown

schematically in Figure 9 and Table 1. The pre-soot chemistry is

represented by the detailed oxidation mechanism that includes

formation of main sooting agent acetylene, C2H2, and some of its

higher derivatives. Such an approach is reasonable for the combustion

of both low and aliphatic hydrocarbons. Soot inception has been

simulated via conversion of soot precursors into incipient soot

particles in the "quasi-equilibrium" processes CH4=2H2 +C(s),

C2H2=H2 +2C(s).

The notation C(s) + H2 can be formally attributed to incipient soot as

the latter contains a significant amount of hydrogen. This is, strictly

speaking, not correct, but has, however, been adopted in the absence

of a more acceptable alternative.

Then, the Nagle-Strickland-Constable model in the form of surface

kinetics relations has been used for oxidation of carbon. It is proven

that the contribution of O2 to the oxidation of soot is minor. It appears,

therefore, that the oxidation of soot occurs by the OH and HO2

radicals, and NO molecules. These processes were accounted for on

the same basis, but with increased "diffusion" efficiencies. Important

steps of CO formation with the heat release effect and consumption of

oxidizing species were also accounted for.

Such a model was implemented into the KIVA-3 code and applied to

the turbocharged Volvo AH10A245 DI Diesel engine fueled by neat

DME. The concentration of soot was found to be noticeable, see

Figure 10. This is no wonder that soot is formed in the case of the

oxygenated fuel: DME represents the partially oxidized fuel rather

than the substance containing not bounded oxygen in its molecule.

Spatial destribution of soot well correlates with the fuel rich, high

temperature (T>1100 K) region.

Future research will focus on detailed comparisons with benchmark

engine data fueled with hydrocarbons.

FIGURE 10. Gas and droplet

temperatures, and concentration

of soot at different crank angles

for the Volvo AH10A245

engine. DME injection starts at

-15 ATDC with a timing of 10

CA.

I've heard that they developed a really good CFD IC engine code at UC Santa Barbara. Does anybody know anything about it? Otherwise I agree that there should be no compromise when it comes to technical issues.

I don't mind accepting the new technology. If you have more information then please let us know.

Why are you quiet now? Give us some information or you are just saying without knowing anything. We all are here to hear from you. Oh, I can't believe I am getting so famous! I am not an automotive engineer but technically (in engineering point of view) knocking out most automotive engineers. Guess how can it be possible!!

I even heard some companies are trying to use the software "WAVE" to analyze flow simulations and acoustics analysis through mufflers of an automobile. I would not be surprised if the company's sales go down due to not using state of the art softwares. After all, competition in the market is so challanging that nobody can afford to do momopoly business anymore.

In mufflers, exhaust gases pass through myriad preforated or drill small holes to attenuate sound intensity. As the exhaust gases pass through small holes, sound source is dissipated by friction or interference. Therefore, sound is reduced. Now we can see that these drill holes play vital role in reducing sound. These holes are two dimensional or three dimensional. So how can a software like "WAVE" with one dimensional capabilities handle flow simulations and pressure distributions through perforated holes!

The flow in a pipe is also 3d and we are still using the "Bernoulli Equation" for a lot of tasks. And this is a 1d description of the flow )

But if you really want to use "state of the art" physics then you will find that even the "Navier-Stokes-Equation" contains a lot of simplifications ......

And please forget all you've learned about physics at school, because the simplifications there are horrible.

For simplicity, "Bernoulli's Equation" is used for simple incompressible pipe flows, channel flows etc. But if you would like to simulate flows and pressure distributions through perforated pipe then your mentioned "Bernoulli's Equation" is not enough. You need to use 2D or 3D simulations.

I'm surprised this thread has been revived. Mr. Islam, have you seen any comparisons of the acoustic predictions of an engine simulation code versus experimental data? I have and they're quite good. The computational expense of building a model of a muffler in 3-D (there are very few mufflers which can be represented in 2-D) and running it seems excessive, like driving a Formula 1 race car down the street to pick up a gallon of milk. Sure it's state of the art, but the cost is impractical. I could build a muffler model and calculate a transmission loss in less than an afternoon using GT-Power. How long will it take to build and run a 3-D model using the software of your choice? Does it include subroutines to calculate transmission loss or do you have to write a subroutine to do that? What type of boundary condition should you use? Industry is not looking for state of the art unless it can also be productive. The additional 1-2% accuracy gained is not worth the large increase in time and expense. In fact, engine simulation codes can generate a sound file of an accelerating vehicle including the gear shifts and it sounds damn good!

Actually the only noise that can't be modeled in an engine code is flow noise generated by turbulence. But in order for a 3-D code to simulate that it would have to include all of the space between the exhaust and the microphone, which could easily turn into several million grid points.

If your specialty were acoustics I would take you more seriously, but I'd bet you have never even tried to model a muffler in either 1-D or 3-D. And just so you know, perforations are easily handled by these codes. The people who wrote them were well aware of their presence in mufflers.

"I'd bet you have never even tried to model a muffler in 1-D."

Answer

Let me explain to you more clearly why 1-D model will not give better results. In mufflers, there are perforated pipes. Let us consider that we are doing simulations in x-directions. But the exhaust gases passes through the perforated holes in another direction. So there is no way you can get better results using 1-D model. I think you have a big problem with your conception.

You are right 1-D will not give you better results than 2-D. But it will give you acceptable results at a much lower cost than a 2 or 3-D will. A right angled branch or junction with multiple pipes is a basic building block of any 1-D model. Maybe 1-D is a limiting term. A network of pipes can be assembled, so maybe "network solution" is better. No engine company would license software that can't model branching. If you don't believe that is can be done, check out SAE technical paper 910072. Don't knock it until you try it.

..."Actually the only noise that can't be modeled in an engine code is flow noise generated by turbulence. But in order for a 3-D code to simulate that it would have to"...

Can someone explain how to simulate turbulence flow noise. Is it involved turbulence energy level?