NOx Model

Downloaded from SAE International by University Of Wales Trinity Saint, Friday, January 17, 2014 09:29:36 AM

SOCIETY OF AUTOMOTIVE ENGINEERS, INC.
400 Commonwealth Drive, Warrendale, Pa. 15096

A Computer Program for
Calculating Properties of
Equilibrium Combustion
Products with Some

Applications to I.C. Engines
Cherian Olikara and Gary L. Borman

University of Wisconsin

Automotive Engineering Congress and Exposition

Detroit, Michigan

February 24-28, 1975

750468

Downloaded from SAE International by University Of Wales Trinity Saint, Friday, January 17, 2014 09:29:36 AM

Copyright © Society of Automotive Engineers, Inc. 1975

All rights reserved.

Downloaded from SAE International by University Of Wales Trinity Saint, Friday, January 17, 2014 09:29:36 AM

750468

A Computer Program for
Calculating Properties of
Equilibrium Combustion
Products with Some

Applications to I.C. Engines
Cherian Olikara and Gary L. Borman

University of Wisconsin

MANY MODELS FOR ENGINE COMBUSTION PROBLEMS use

the First Law of Thermodynamics as applied to

either the entire cylinder contents or to sub

systems. Examples include models for spark ig nition engines (1)*, diesel engines (2) and stratified charge engines (3). Typically the major species of the products of combustion may be assumed to follow a shifting equilibrium process for thermodynamic purposes. For pur

are given by References 6-9. The NASA-Lewis program (9) is very extensive and includes ther modynamic data for hundreds of species.
The purpose of the present program is very narrow by comparison to the NASA program. Its purpose is to specifically deal only with the gas phase products of combustion of hydrocarbon

fuels (containing, C, H, O, N atoms) and air.
The program is however extended to calculate the

partial derivatives of internal energy and molec ular weight which are helpful when numerically at the equilibrium concentration (4). Because solving the first law as a differential equation of this wide spread use it is important to have in time using the method of Reference 10. Most a rapid means of calculating the equilibrium in importantly, the program is to provide a rapid ternal energy, molecular weight and species mole means of calculation comparable to the use of fractions of the products of hydrocarbon-air com the regression analysis equations given in Ref poses of chemical kinetics calculations, many of the major species may also be assumed to be

bustion.

erence 11.

Calculation of the equilibrium composition and internal energy of combustion products for

ENERGY EQUATION FORMULATION

engine cycles goes back to the early work of

Tizard and Pye (5) and has been the subject of numerous computer studies, a sampling of which
*Numbers in parentheses designate References at end of paper

In order to treat the cylinder gas system

thermodynamically the system may be divided into

cells such that the composition and temperature

of each cell is uniform. This implies that the molar average temperature of the cell may be

ABSTRACT

bon fuel and air is described.

A subroutine is

A computer program which rapidly calculates the equilibrium mole fractions and the partial

also given which calculates the gas constant,

for the products of combustion of any hydrocar

pressure and equivalence ratio. Some examples of the uses of the programs are also given.

derivatives of the mole fractions with respect to temperature, pressure and equivalence ratio

enthalpy, internal energy and the partial deriv atives of these with respect to temperature,

Downloaded from SAE International by University Of Wales Trinity Saint, Friday, January 17, 2014 09:29:36 AM

2 used to obtain the internal energy of the cell.
As an example, some models used for the spark ignition engine products divide the products in

to subsystems (1) but assume the same pressure

for each subsystem while others (11) assign one temperature for the entire product mass. The first law for each system of products is

necessarily be true for the reactants (12).
The specific gas constant, where M is the mean molecular weight, is a function of p, T,
F. The derivatives dp/dt in Equation 2 may be obtained by differentiation of the ideal gas law and substitution of:

for the dR/dt term.

This gives:

The ideal gas equation cannot be explicitly solved for p in terms of density, temperature and composition. p, T, F and ? are known
If

at state "n" then the pressure at state n+1 may

be obtained by a first order approximation to
In order to solve Equation 1 it is necessary R. This gives:

to supply an equation of state and the internal

energy as a function of the composition and two state variables such as pressure and temperature.
The volume must be determined from the fact that

the sum of all system volumes is equal to the known total volume. A mass balance plus appro

priate flow equations determine the mass in the

system. The heat transfer may be modeled by the use of empirical convection and radiation equa

tions which generally will be functions of the temperature, density and composition as well as some characteristic velocity.
For a numerical solution of Equation 1 it

is convenient to introduce the internal energy as the sum of the chemical energy and sensible energy. For equilibrium the energy then can be expressed as a function of the variables p, T and F. The variable, F, is the fuel/air equiva lence ratio. For products formed directly from the reactants (fuel plus air) the value of F is the equivalence ratio of the reactants. For products of such a reaction mixed with air or additional products of the same fuel the value

Equation 4 may be solved numerically using

Equation 5 to evaluate p and appropriate differ

ential conservation equations for m and F. An example of this technique is given later, but first we shall consider the program to evaluate u and R and their derivatives.

EQUILIBRIUM THERMODYNAMICS

Let the fuel CnHmOlNk and air (13) at equiv

alence ratio F react and the products subject to

temperature T and pressure p attain equilibrium.

The numbers n and m should be nonzero; l and k

may or may not be zero.

of F may be computed for a hypothetical reactant
F which would give the same atomic properties as the resulting mixture. The term du/dt may be written The partial derivative terms can be calculated

directly from the equilibrium thermodynamics as functions of T, p and F.
Little error is made for normal engine con ditions if the products are assumed to be an ideal gas mixture although the same may not

where x1 through x12 are mole fractions of the product species. The number, x13, represents

the moles of fuel that will give one mole of

products.

Downloaded from SAE International by University Of Wales Trinity Saint, Friday, January 17, 2014 09:29:36 AM

3
The left side of the equation may be written as Atom balances for the various elements give

REDUCTION IN NUMBER OF EQUATIONS

Equation (8-1) gives
Equation (8-5) gives
These are now used to eliminate X12 and

x13 in Equations 8 and 9.

The constraint that the mole fraction of all the

products add up to unity requires that

To solve for the 13 unknowns, we need 7 more

Denote the ratios

equations which are provided by the criteria of equilibrium among the products, expressed by the following 7 non-redundant hypothetical reactions. Now Equations 11 can be used in Equations 12 to

eliminate all variables except x4, x6, x8 and
X11. The resulting four nonlinear equations with four unknowns may be expressed symbolically by Assume we know a vector

Where p is the pressure in atmospheres.
The equilibrium constants are curve fitted from

data in JANAF Thermochemical Tables, Second Edi

tion (1971). Details of fitting are given in

Appendix A.

The expressions for equilibrium

which is reasonably close to the true solution vector constants can be rearranged to express mole

fractions of all the products of combustion in terms of x4, x6, x8 and X11 the mole-fractions

of H2, CO, O2 and N2 respectively.

The functions on the left side of the equations

Downloaded from SAE International by University Of Wales Trinity Saint, Friday, January 17, 2014 09:29:36 AM

4 can then be expanded around the known vector as a Taylor 's series.

Substituting for x6, x9 and X10, rearranging terms and dividing through by x13,

Neglecting the partial derivatives of second or

der and higher we get a set of linear equations

where ?xi are the approximate corrections.

Thus

The quantity X13 can be estimated with good ac

curacy from the condition that the mole fractions should add up to unity.

where functions fj and their partial derivatives are evaluated at the known vector.

This set of linear equations can be solved for ?x4, ?x6, ?x8 and ?x11 using a Gaussian

For F = 1 a good estimate of X13 obtained be can from

elimination method. The improved values are then
For F > 1 an estimate of x13 can be obtained
The improved vector can now be used to evaluate the partial derivatives and functions.

from

Further corrections can then be calculated and

applied. The iterative procedure is to be con tinued until the relative changes in all the com ponents are less than a specified value.

Details

of the computation of the elements of Eq. 14 are given in Appendix B.

method, the other unknowns may be obtained di

rectly by substitution into Equations 16-1, 16-2,

INITIAL ESTIMATION OF MOLE FRACTIONS

The Newton Raphson iteration is not self

starting. Unless an approximate solution is known from a previous execution of the subroutine at conditions close to those presently encoun

tered, it is necessary to make an initial esti mation to get the iteration started. The follow

ing method of estimation developed through judi

cious simplifying assumptions is found to be re markably good.

Assume that the products are only H2, CO,

O2, H2O, CO2, N2 and Ar.

Substitution of the estimated value of x13

into Eq. 16-6 gives an equation in the single un known x8. After solving Eq. 16-6 by Newton 's

Equation 6 becomes

and 16-3.

These initial estimates can then be

used to start the solution of Equations 13 by

use of Equations 14 and 15.

PARTIAL DERIVATIVES OF THE MOLE FRACTIONS

Starting with Equations 13, the functions

and hence the mole fractions are dependent upon

temperature, pressure and equivalence ratio.
For example, taking the total derivatives of each equation with respect to temperature we get four linear simultaneous equations in four un knowns. From the C balance and Eq. 11-7
It is seen that the coefficient matrix is

the Jacobian that appeared in Newton Raphson

From the H balance and Eq. 11-6

iteration. Similarly, it is seen that the matrix equations for solving for the partial derivatives with respect to pressure and equivalence ratio would also be identical except that T is replaced by p or F. Equations for the computations of the elements of the coefficient matrix and the con

and from the N, Ar, and O balances

stant vector are given in Appendix C. Solution of the equations sets (Equations 18 and similar ones for p and F) can be accomplished in a straight forward manner using the Gaussian elim ination method. As shown in Appendix C the re maining partial derivatives of x1,x2, etc. are easily obtained by use of Equations 11 and 8-5 and direct substitution.

Downloaded from SAE International by University Of Wales Trinity Saint, Friday, January 17, 2014 09:29:36 AM

5
The computation of the equilibrium mole fractions and partial derivatives ofthe mole

fractions with respect to T, p, F iscarried

out by a FORTRAN coded program EQMD.A guide for users is given in Appendix D anda program listing is given in Appendix E.
ENTHALPY INTERNAL
AND
ENERGY

The absolute enthalpy used here is defined as The values of the specific heats and heats of

formation were taken from Ref. 14 and were used

with a table look up and linear interpolation program. Using the results of Program EQMD the en

thalpy of the mixture, h, is easily calculated as is the average molecular weight, M, and gas

constant, R.

we consider the expansion of a closed system of products. For the adiabatic constant pressure reaction we first calculate the reactant mixture enthalpy.
If the fuel heat of formation is not known it can

be estimated using the heating value and H/C ra

tio (15). For higher pressures the non-ideal ef fects (12) should be included. Having computed

the reactant enthalpy, hr, we hold p and F con

stant and solve for T

The partial derivatives are also easily computed. Using Newton 's method
For example,

The other expressions for the partiais of h, u

and R with respect to T, p and F are given in

Appendix C.
The computation of u, h, R and their par

where Tn is the first estimate and Tn+1 is the improved estimate. The first estimate can be any reasonable number but preferably larger than the expected solution. For our example, we take methane at 536.4°R, one atmosphere and F = 0.8,

0.9, 1.0, 1.1, 1.2, Figure 1 shows a plot of the tials is carried out by FORTRAN coded program PER. result and Table 1 shows the progressive esti
A users guide is given in Appendix D and a mates of T for the F = 1.1 case. A plot of program listing is given in Appendix E.
?(h-hr)/?F at the calculated F and T values of Figure 1 gives the dashed curve of Figure 1. The zero point determines the F at the maximum flame
EXAMPLE CALCULATIONS temperature. The following example of expansion of a closed system of products of combustion is given to show the use of the programs. First we con sider the classical problem of computing the con only to illustrate the application of the pro stant pressure adiabatic flame temperature. Second, grams to such calculations. The kinematic and

Two simple cases are given here as examples

Downloaded from SAE International by University Of Wales Trinity Saint, Friday, January 17, 2014 09:29:36 AM

6
Total mass of gas inside the system is
0.001244 lbm and there is no flow in or out of

the system. The fuel is C8H16 and equivalence

ratio F is 1.2.

Initial conditions are T =

5600°R and p = 533 psi at ? = 20°. The expan sion is assumed to follow shifting equilibrium.
Improved Euler method is used for solving the differential equations. The steps taken at each interval are listed below.

Let the temperature Tn and pressure Pn be

known at time tn where the derivative, dT/dt, is calculated using Eq. 4.

heat transfer models used below are extremely simplistic. are used as starting condi

tions for the subsequent time interval.

The results of the program for expansion from ? = 20° to 180° is plotted in Figure 2.
In addition, the integrated values of pdV and Q were also accumulated during the calculation.
The change in internal energy was calculated from initial and final conditions.

For a step

Downloaded from SAE International by University Of Wales Trinity Saint, Friday, January 17, 2014 09:29:36 AM

7 size of 1° crank angle:

"Engine Simulation Studies Conducted at the Uni versity of Wisconsin." Tech. Report No. 11398,
U.S. Army Tank Automotive Command, March 1972.
3.L.W. Evers, P.S. Myers, and O.A. Uyehara,
"A Search For a Low Nitric Oxide Engine." Cen tral States Combustion Institute Meeting,
Madison, Wisconsin, Spring 1974.

4.H.K. Newhall, "Kinetics of Engine-Generated

ACKNOWLEDGMENT

Nitric Oxides and Carbon Monoxide."

Twelfth

Symposium (International) on Combustion, 1969.
5.Tizard and Pye, "Empire Motor Fuels Commit tee Report." Proc. Inst. of Automotive Engrs.,

This report was submitted in partial ful fillment of Grant Number R-802589-01-0 by the
University of Wisconsin-Madison under the partial Vol. 18, No. 1, 1923. sponsorship of the Environmental Protection
6.H.N. Powell, A. Schaffer and S.N. Suciu,
Agency. Work was completed as of July 1, 1974.
"Thermodynamic Properties--Properties of Combus tion Gases, System: CnH2n-Air." General Elec

NOMENCLATURE

Aij afi/axj dQ/dt - rate of heat
-

fi

-

transfer nonlinear function of mole fractions

F - fuel/air equivalence ratio h - enthalpy per unit mass

tric Co., Cincinnati, Ohio, 1955.
7.H.K. Newhall and E.S. Starkman, "Thermody

namic Properties of Octane and Air for Engine
Performance Calculations." SAE Paper 633G, 1963.
8.R.J. Steffensen, "A FORTRAN IV Program of
Thermochemical Calculations Involving the Ele

hi hr -

enthalpy per mole of species i except

ments Al, B, Be, C, F, H, Li, Mg, N and 0 and
Their Compounds," PhD Thesis, Purdue University,

-

reactant enthalpy per unit mass

9.S. Gordon, "Complex Chemical Equilibrium
Calculations." NASA Sp-239, Kinetics and Thermo

m

-

Ki

where otherwise defined

-

partial pressure equilibrium constant

mass or number of H atoms in fuel molecule M

-

n

-

? -

molecular weight of mixture

number of carbon atoms in fuel mole cule pressure

r,r ',r" - proportions of N2, O2, Ar in air
R - specific gas constant of mixture
Ro - universal gas constant t - time

T - temperature

?xi/?xj
Tij - gas side wall surface temperature
Tw
-

u

-

internal energy per unit mass

V - volume

xi

? overdot -

mole fraction

-

gas constant correction factor indicates d/dt

-

REFERENCES

1966.

dynamics in High Temperature Gases, March 1970.

10.G.L. Borman, "Mathematical Simulation of
Internal Combustion Engine Processes and Perform

ance Including Comparisons with Experiments."

PhD Thesis, University of Wisconsin, Madison,

1964.

11.R.B. Krieger, and G.L. Borman, "The Compu

tation of Apparent Heat Release for Internal Com

bustion Engines." ASME Proc. Diesel Gas Power,

ASME Paper 66-WA/DGP-4, 1966.
12.J. Maniotes, "Ideal and Non-ideal Theoreti

cal Equilibrium Calculations for Mixtures of

Methane and Air." PhD Thesis, Purdue University
1962.

13."CRC Handbook of Chemistry and Physics."
71st Edition, Sea level atmospheric composition

of dry atmosphere, pp. F-147, 1970-71.

14."JANAF Thermochemical Tables." Second Edi

tion, The Dow Chemical Co., Midland, Michigan,
1971.

15.H.N. Powell, "Applications of an Enthalpy

1.P. Blumberg and J.T. Kummer, "Prediction of Fuel/Air Ratio Diagram to 'First Law ' Combustion
NO Formation in Spark-Ignited Engines--An Analy Problems." Trans. ASME, pp. 1129-1138, July
1957.
sis of Methods of Control." C.S.T., Vol. 4, pp. 73-95, 1971.
16.I. Klotz, "Introduction to Chemical Ther
2.G.L. Borman, P.S. Myers, and O.A. Uyehara, modynamics." W.A. Benjamin Co., New York, 1964.

Downloaded from SAE International by University Of Wales Trinity Saint, Friday, January 17, 2014 09:29:36 AM

8
APPENDIX A
CURVE FITTING OF EQUILIBRIUM CONSTANTS
The equilibrium constants data was taken

from JANAF Thermochemical Tables (14), where lNew Text og10Kp (formation) for all species are tabulated as functions of the absolute temperature (°K).
Equilibrium constants for the reactions consid ered in the equilibrium thermodynamics were cal culated using the relation

where T is the absolute temperature and A, B, C,
D, E are constants.

This model was used to fit the tabulated

data by means of a least squares fitting program.

A trade-off between the conflicting interests of

obtaining either high accuracy in a narrow range of temperature or lower accuracy in a wider range

led to choosing 600 to 4000°K (1080 to 7200°R) as

the range most relevant to the study of combus tion phenomena in engines.

The log Kp predicted by the equations were

compared with the original data and the devia tions were less than 0.0009. (The original data is tabulated only to the third decimal place and so there is an inherent uncertainty of 0.0005.
Hence the deviations were not considered signi

Theoretical studies (16) suggest a function relationship of the type

ficant.)
A transformed temperature TA defined as

0.005 T/9 where T is in °R was used for fitting.
The constants A, B, C, D and E are listed in
Table A-1.

Downloaded from SAE International by University Of Wales Trinity Saint, Friday, January 17, 2014 09:29:36 AM

9
APPENDIX B
MOLE FRACTION COMPUTATION EQUATIONS
In the specific problem under consideration, we define for convenience, the following partial derivatives From Eqs. 11,

we get

With the above notations, the elements of

Eq. 14 written as a matrix equation [A][?x] = [B] can be expressed as

Downloaded from SAE International by University Of Wales Trinity Saint, Friday, January 17, 2014 09:29:36 AM

10
APPENDIX C

PARTIAL DERIVATIVE COMPUTATIONEQUATIONS
Define

From Equations 11, Yi are seen to be functions of x4, X6, x8 and x11 only.

Substituting for xi from Equation Cl in
Equation 12 and differentiating with respect to

T, we derive

Define

The model used for fitting equilibrium constants

Partial derivatives with respect to p are similar

KP is

to above except that ?/?T is replaced by ?/?p.

Note that C5 and C7 are not functions of p and hence terms involving ?C5/?p and ?C7/?p drop out.
The parameters d2, d3 and d4 are functions of F, but none of the Ci are. Hence

From which we derive

CALCULATION OF PARTIAL DERIVATIVES OF REMAINING

MOLE FRACTIONS

From Equations 11

From the expression for

Ci and di

Downloaded from SAE International by University Of Wales Trinity Saint, Friday, January 17, 2014 09:29:36 AM

11
CALCULATION OF PARTIAL DERIVATIVES of R, h and u
From Equations 19 to 22

For partial derivatives with respect to p, substitute ?/?p where ?/?T appears above.
Example:

For partial derivatives with respect to F, substitute zero where ?Ci/?T appears and ?/?F elsewhere. Example:

Note that

the specific heats of the

species at constant pressure.

For partial derivatives with respect to p and F, replace ?/T with ?/?p and ?/?F respec tively. The exceptions are

Exception:

APPENDIX D
GUIPE TO USING SUBROUTINES EQMD AND PER

(where AN, AM, AL are the number of C, H and O atoms in fuel molecules) free carbon will defi
PURPOSE OF THE SUBROUTINES - Subroutine EQMD nitely be formed. For C8H16, this maximum F is can calculate the mole fractions xi of the prod 3. ucts of combustion of any hydrocarbon fuel and
The products of combustion are assumed to air, at equilibrium under specified conditions of be ideal gases. This assumption is not valid at extremely high pressures (12). temperature T, pressure p and equivalence ratio
HOW TO ACCESS THE SUBROUTINES - Transfer of
F. It can also optionally calculate the partial

derivatives of the xi with respect to T, p and F. information between the calling program and the

subroutines are through labelled COMMON areas
Subroutine PER can calculate the average and hence every program that calls the subrou molecular weight M, gas constant R, enthalphy h and internal energy u of the equilibrium products tines must contain of combustion. It can also optionally calculate the partial derivatives of R, h and u with re spect to t, p and F.

In addition to carbon and hydrogen, the fuel may or may not contain oxygen and nitrogen atoms.
The product species considered are H, O, N, H2,
OH, CO, NO, O2, H2O, CO2, N2 and Ar in gas phase.

LIMITATIONS OF THE SUBROUTINE - The equilib

rium constants used in the subroutines were

fitted as a function of temperature in the range

1080°R to 7200°R. The subroutines cannot be used outside this range.
The subroutine cannot handle the formation

of free carbon. lence ratio,

It can be shown that for equiva
The above variables are input to the subroutine

and should be defined at the time of calling.

Downloaded from SAE International by University Of Wales Trinity Saint, Friday, January 17, 2014 09:29:36 AM

12
The following are output variables.
IERR - Error code with meaning as specified

tered during calculation of

partial derivatives of mole

later.

XEQ(12) - Mole fractions of the product species.
XPA - The numbering code is given later.

A parameter used in the subroutine.
It represents the modes of fuel that

will give one mole of products.

fractions.

Immediately after every call to the subrou tine, the value of TERR must be checked. It should be zero. If it is not, the numbers

stored in the other output variable locations are meaningless and should not be used. Prefer rably print out the value of IERR and also the

condition (AN,AM,AL,AK,F,T,P,KLO)at which the failure occurred.

AVM - Average molecular weight of products of combustion

R - Gas constant BTU

H - Enthalpy BTU

°R-1

U - Internal energy BTU
The partial derivatives defined earlier

have units of the variable per °R, per psi , per unit fuel air equivalence ratio (dimensionless) respectively. For example,
DXT(1) has units of °R-1

DRP has units of BTU

DHF has units of BTU

psi-1
°R-1

REFERENCE STATES - For computation of en thalpies and internal energies, the following states were assigned zero value of enthalpies.
H2 (ideal gas) at 0°R
O2 (ideal gas) at 0°R
N2 (ideal gas) at 0°R
C (Graphite) at 0°R
Ar (ideal gas) at 0°R

HOW TO CALL SUBROUTINE PER - Both the

$INCLUDE cards and all three of the COMMON state

ments are required. The subroutine PER will be executed every time the following statement is encountered. CONTROL VARIABLE K-]OL KLO = 0
The subroutine will make a fresh

estimate before iterating
KLO = 1

Closeness criteria built into the subroutine will decide whether or not to make a fresh estimate. THIS
IS RECOMMENDED.

KLO = 2

The subroutine will use input values of H2, CO, O2 and N2 mole fractions as initial estimates

ERROR CODE I-]ERR

IERR 0
=

= 1
= 2

No error.

Specified temperature out side 1080 to 7200°R range.
Specified equivalence ratio too high as to cause forma tion of free carbon.

= 3

Attempted to calculate an estimate; but none found within

= 4

outputs of EQMD are also available after a call

to PER.

It is not necessary to call EQMD again.

However if only the mole fractions and/or their

partial derivatives are needed and not the prop

erties, some computation time could be saved by directly calling EQMD as explained below.
HOW TO CALL SUBROUTINE EQMD ALONE - In this case # INCLUDE CO*PER and COMMON/PROP/ cards are not needed. The subroutine EQMD will be executed each time the following statement is encountered.

A singular matrix was encoun

= 5

The subroutine PER internally calls EQMD

with the same value of JDR and all the relevant

mole fractions
Few mole fractions were cal

reasonable limits.

tered during calculation of

culated as negative during iteration. = 6

=7

Newton Raphson iteration did not converge in 25 attempts.

A singular matrix was encoun

SAMPLE PROGRAM - Listing of a sample program which calls subroutine PER is given on the fol lowing page.

Downloaded from SAE International by University Of Wales Trinity Saint, Friday, January 17, 2014 09:29:36 AM

13

Downloaded from SAE International by University Of Wales Trinity Saint, Friday, January 17, 2014 09:29:36 AM

14

Downloaded from SAE International by University Of Wales Trinity Saint, Friday, January 17, 2014 09:29:36 AM

15

Downloaded from SAE International by University Of Wales Trinity Saint, Friday, January 17, 2014 09:29:36 AM

16

Downloaded from SAE International by University Of Wales Trinity Saint, Friday, January 17, 2014 09:29:36 AM

17

Downloaded from SAE International by University Of Wales Trinity Saint, Friday, January 17, 2014 09:29:36 AM

18

Downloaded from SAE International by University Of Wales Trinity Saint, Friday, January 17, 2014 09:29:36 AM

19

Downloaded from SAE International by University Of Wales Trinity Saint, Friday, January 17, 2014 09:29:36 AM

20

Downloaded from SAE International by University Of Wales Trinity Saint, Friday, January 17, 2014 09:29:36 AM

21

This paper is subject to revision. Statements and opinions advanced in papers or discussion are the author 's and are

in Transactions. For permission to publish this paper in full or in part,
SAE

been edited by SAE for uniform styling and format. Dis cussion will be printed with the paper if it is published

tion through SAE should send the manuscript or a 300 word abstract of a pro posed manuscript to: Secretary, Engineering Activities Board, SAE.

his responsibility, not the Society 's; however, the paper has

contact the SAE Publications Division.

Persons wishing to submit papers to be considered for presentation or publica

page booklet.

Printed in U.S.A.

References: Downloaded from SAE International by University Of Wales Trinity Saint, Friday, January 17, 2014 09:29:36 AM 2