EPA-AA-SDSB-81-2
                           Technical  Report
           An Energy Demand Model for Light-Duty Vehicles,
            with Concepts for Estimating Fuel Consumption
                                 by


                            Terry Newell



                             April 1981
                               NOTICE

Technical Reports do  not  necessarily  represent  final EPA decisions
or positions.   They are intended to present  technical  analysis of
issues using  data which  are  currently  available.   The  purpose in
the release of  such reports  is  to facilitate  the exchange of tech-
nical  information and to inform  the  public  of  technical develop-
ments which may  form  the  basis for a final EPA decision, position
or regulatory action.
              Standards Development and Support Branch
                Emission Control Technology Division
            Office of Mobile Source Air Pollution Control
                  Office of Air,  Noise  and Radiaton
                U.S. Environmental Protection Agency

-------
                                                               EPA-AA-SDSB-81-2
           An Energy Demand Model for Light-Duty Vehicles,
            with Concepts for Estimating Fuel Consumption
                                 by


                            Terry Newell



                             April 1981
                               NOTICE

Technical Reports do not  necessarily  represent  final EPA decisions
or positions.   They  are intended to present  technical  analysis of
issues using  data which  are  currently  available.   The  purpose in
the release of  such  reports  is  to  facilitate  the exchange of tech-
nical  information and  to inform  the  public  of  technical develop-
ments which may form the  basis for a final EPA decision, position
or regulatory action.
              Standards Development and Support Branch
                Emission Control Technology Division
            Office of Mobile Source Air Pollution  Control
                  Office of Air,  Noise  and Radiaton
                U.S. Environmental Protection Agency

-------
 I.   Introduct ion

     This  report  presents  and  discusses  a  computer  program  that
 models the  energy demand at  the drive  wheels of  a light-duty
 vehicle  operated  over  a specified  driving  cycle.    The model  is
 based  on direct interpretation of the physical forces that act on a
 vehicle  in motion.   In  this  manner, the model  deterministically
 calculates the energy required to operate the vehicle.

     The energy model  is also  used  to estimate vehicle  fuel  con-
 sumption.  In  this application of the model,  the underlying  concept
 is  that  fuel  consumption  can  be  related  to energy demand  if  the
 characteristics of the drivetrain  are  known.   It  is because of  the
 importance  of this link that  this approach is  described as  a
 conceptual means of estimating fuel economy.

     There  are several  advantages  inherent  in estimating  vehicle
 fuel consumption by this type of approach.   The most significant is
 cost.   If a  direct  empirical  approach  is taken to  observe  small
 changes in parameters affecting vehicle fuel  economy,  the resulting
 effects  are  easily obscured by  other factors, and multiple  tests
 are required to isolate and quantify these  effects.   The  "noise" in
 a  computer simulation  is  limited and  predictable;  thus a  single
 inexpensive  computer  modeling  provides the desired  information.

     In  his  book  on computing  for scientific  and  engineering
 applications,  Richard Hamming  emphasizes  the philosophy  that "the
 purpose  of computing  is  insight,  not numbers."[1]   The  primary
 intent of this model  is to provide the user with added insight  into
 the effects  on vehicular energy demand that  occur, as vehicle  and
 driving cycle parameters are varied.

     The following sections of this report  describe  the development
 of the energy  demand modeling program, the use  of the program,  and
 the verification  of  using  the energy demand  concept to estimate
vehicle fuel consumption.  In addition, several applications of  the
model are presented and other potential applications are  discussed.

 II.  Development  of the Energy Demand Model

    A.   General

     The guiding  principle in the development of this  model  was  the
efficient determination of energy demand, which then can  be  used to
estimate fuel  consumption.  Energy demand  is  computed  by maintain-
 ing a  cumulative  total  of  the  vehicle power requirements  at each
 time interval.   These powers  are computed  from  knowledge  of  the
 forces acting  on  the  vehicle.    Therefore,  these  forces are  the
 fundamental  parameters  required  by  the model.   This program con-
siders the  three  primary forces that  act  on a vehicle that is  in
motion on a level  surface:

     F = F  + F  + F                                             (1)

-------
where:

     F   =  component  of total  force due  to  rolling resistance;

     F   =  component  of total  force due to aerodynamic resistance;
           and ,

     F =   component  of total  force necessary  to overcome inertia.

     The rolling resistance component of the total acting force is
primarily  a  function of  the  tires  used  on the  vehicle.   Tire
rolling resistance is nearly constant at velocities up to approxi-
mately 60 mi/hr. [21   Since  the  EPA urban (modified LA4) and Highway
Fuel Economy Test (HFET) cycles do not  specify driving in excess of
60 mi/hr  (the top speeds  are  56.7 mi/hr  and  59.8  mi/hr,  respec-
tively), modeling  rolling  resistance  as  velocity-independent  is
reasonable.

     Aerodynamic drag  increases proportionally with  the  square of
velocity, and the force  due to inertia is  the product of the mass
of the vehicle  and  its  instantaneous  rate  of  acceleration.   Thus,
the total  force  acting on  the moving  vehicle  can  be expressed[3]
by:
which  is  a reformulation  of  equation  (l).   This  is the  basic
equation of the model.   Energy  demand can be computed from equation
(2)  after  the  force coefficients  f   and f   and  the mass  of  the
vehicle, along with  a driving schedule, have been supplied.

     B.   The  force  coefficients

     The values of fn and f  for use in equation (2) are typically
not available  in this direct form.   Values  of these force coeffi-
cients  can  be  obtained  from  three other, more  readily available
forms of vehicle data.   Each of  these  is particularly well-suited
for  a  given  type  of simulation by the  model, and are briefly
outlined below.

     1.    Road/track coastdown data.   If  road  coastdown data have
been collected  and   analyzed for  the  vehicles of  interest,   the
resulting coefficients  of the  acceleration equation can be used to
obtain  values  of  f  and  f .    This  method  of determining  the
force equation coefficients  is  the best for modeling vehicle energy
demand for on-the-road operation.

     Analysis  of the  speed versus  time  data collected  during
coastdown testing gives an equation describing the  deceleration of
the vehicle: [3]

-------
     A = an + a2 V2.                                            (3)

Applying Newton's Second Law of Motion and distributing yields:

     f0 - tna0                                                   (4)

     f2 = ma2                                                   (5)

(Since  there  is  no conventional  assignment  of units  for a   and
a_,  care   in  applying  unit  conversion  factors  to the  right-hand
side of equations  (4)  and (5) is  necessary).   From these  numbers,
the energy demand of  the  vehicle  operated  on  the  road  according to
a given driving cycle can be modeled.

     2.   Dynamometer coastdown data.  If a vehicle-dynamometer
coastdown has been conducted for the vehicle in question,  the
results of that test can be used to determine values for fn and
f „.  This method of obtaining values for the force  coefficients  is
tfie best for use in modeling energy demand of the  vehicle  operated
on the dynamometer.   The information required is the 55 to  45  mi/hr
coastdown time At, and the actual (total) horsepower absorbed  by
the dynamometer, AHP.

     The total  force acting  on the vehicle  at  the  road-dynamometer
match point speed of 50 mi/hr is approximated by:

                                                                (6)
     F(50) = m(-^)
               At

     Of the  total  power  absorbed  by the  dynamometer (AHP),  the
power absorber unit  (PAU)  accounts for the greatest portion.    The
power  absorbed by  the PAU  is proportional  to  the  cube of  the
velocity.   Therefore, converting  dynamometer AHP from horsepower to
watts  and  factoring out  the velocity  yields  an  estimate of  the
component  of total force that is proportional  to the square of  the
velocity:
            745.7  - x AHP                                      ,..v
     F  = 	h£	                                   (7)
      A   sn El   1609.3 m-hr
          5  hr X  3600 mi-s

The  use  of dynamometer  AHP in  equation  (7) results  in a  slight
overestimation  of  the V   force term of equation (2),  since  it
includes  some  minor  effects (e.g.  dynamometer  bearing  friction)
that  may not  be proportional  to V .   The PAU  setting  can  be
substituted for  the AHP  in  equation ,17)  if desired; however,  this
would  tend to  underestimate  the  V   component.   The  AHP .value
appears to be more readily  available, and the error in  the  V   term
associated with its use is relatively small.

-------
     Equations  (6)  and (7)  can then be  used to obtain  values  of
f0 and f2:

     fQ = F(50) - FA                                            (8)


        _ ^A, V = 50 mi/hr = 22.35 m/s                          (9)
      2 = V2

and  vehicle  energy  demand  over  the  driving cycle,  on  the  dyna-
mometer, can then be modeled.

     3.   Vehicle design  parameters.  Modeling  energy requirements
using  force  coefficients derived  from  vehicle  design  parameters,
such as aerodynamic drag coefficient,  allows certain analyses  to  be
performed concerning design  goals  and future possibilities.   Using
this  type  of  input  to the  model, the  drag coefficient  (or  mass
reduction, etc.) required for  a targeted  fuel economy increase can
be estimated.   Conversely,  the impact  on  energy demand  and  fuel
economy of  intended  changes  in design  parameters can  be approxi-
mated.  In this case, the force coefficients are:

    fQ = (mg)(RRC)                                               (10)

where:

     m = vehicle mass (kg)
                                              2
     g = acceleration due to gravity (9.81 m/s )

   RRC = rolling resistance coefficient  of tires; and


    f  =1
     2   2 p  V

where:

     p = air  density  (kg/m )

    C  = coefficient  of aerodynamic drag

     A = vehicle frontal area.

     Equation  (2),  which forms  the  basis of   the  energy demand
calculations, can then be  reformulated as:
                               2    A                           (12)
     F = (mg)(RRC)  +  (   p  CA)V

-------
From  equations  (10)  and  (11)  it  is  apparent  that the  force  co-
efficients can be computed given the mass of  the vehicle,  the
rolling  resistance  of  the  tires,  and the  aerodynamic  parameters
of the  vehicle.  Note that the  important information for cal-
culation of  f  is the product of  the  frontal  area and drag  co-
efficient.

     Equations  (10)  and  (11)  also provide  a means of  estimating
values  of  the vehicle  design  parameters from road  or  dynamometer
coastdown data:
     RRC = fQ/mg                                               (lOa)

     CDA = 2f2/p                                               (lla)

The  rolling  resistance  coefficient  obtained  from  (lOa) will
include  wheel  bearing  and other  losses,  and  as  such  will be
slightly higher than the measured value for the tires only.   If  the
vehicle frontal area is known, then  an estimate  of C  can  also be
derived from equation (lla).

     C.   The Energy Demand

     After all  of  the  information necessary  for  solving the force
equation,
has been assembled,  the vehicle energy demand can be modeled.   This
basic equation  of  the forces acting  on  the vehicle is solved  at
each  second for  the duration  of the  specified  driving cycle.

     The  physical  interpretation of  this  solution  depends  upon
the signs of the total force at  velocity V,  F(V), and the  inertial
term of the right-hand side, m dv/dt.  There  are three combinations
of the  signs  of these terms  that  can occur,  each  representing  a
different physical  situation.

     If the vehicle  is  accelerating  or is maintaining a  constant
nonzero velocity,  then m  dv/dt _> 0 and F(V)  > 0.    In this  case,
all of  the forces acting on the vehicle result  in power being
demanded of the engine.   The  current  (instantaneous) power  demand
is obtained by multiplying total  acting force by current velocity:
P = FV.

     When  the  acceleration is  negative (i.e., the vehicle  is
decelerating),  the inertia component  of total  force  is also  nega-
tive.   The situation has  a different  physical meaning dependent  on
whether the total  acting force is positive or negative.

-------
     If  the  acceleration is negative  but the magnitude of  the
rolling resistance and aerodynamic terms is sufficient to result in
a  net  positive acting force, then  power  is  still being  demanded
from the engine.  Physically  the vehicle is operating  under powered
deceleration:   the  rate  of deceleration is slower than that of  a
"free" coastdown, and  power  from  the engine  is required  to main-
tain the prescribed  driving schedule.

     During  rapid decelerations,  especially  at   low speeds,   the
magnitude of the negative inertia  term can  equal or exceed the  sum
of  the  (positive) rolling resistance and  aerodynamic  components,
resulting in a zero  or  negative total force F. The  vehicle is
either freely coasting or braking, and no energy is being  demanded
from  the engine  to keep  the  vehicle  moving in the  prescribed
fashion.

III. Computer Operation of the Model

     This  section of  the report  describes  the  Fortran  computer
program used  to model  LDV energy  demand,  the required  input,  and
the program output.   A  listing  of  the program is provided as
Appendix A for reference.  The statements  having  line  numbers
enclosed in  boxes  are used  only to  estimate fuel  economy from
modeled energy  demand  (See Section  IV), and  are  deleted  if only
energy demand is desired  as output.

     A.    Input

     Two sets of  data  are necessary for running  the basic energy
demand modeling:   vehicle parameters and  a  driving cycle.    The
driving cycles that  have  been used  in the  development and  testing
of  the  model  are the  modified  (1371  second)  version of the LA-4
cycle,  which  forms the  basis  of  the  EPA urban driving cycle
(FTP),[4]  and the  765 second highway  fuel  economy test (HFET)
cycle.[5]  Any driving  cycle that  is  defined by a time versus
velocity history can be used.

     The driving  cycle  should  be  listed  as  a time-vs-velocity
table, with  time  intervals of  one  second  and velocity given in
miles per  hour.  Conversion of mi/hr  to  m/s is built into  the
program;  if  the  velocities  are  given in units other than mi/hr,
one line of the program must be modified  or  deleted.

     The required  vehicle information  consists of  the coefficients
of the force  equation  (2).  These  can be supplied  to  or  calculated
by the program, as described in the previous  section.  All of  the
calculations  in the  program are  performed using SI (metric) units;
the appropriate units   for each program variable  are  given as part
of the program listing  in Appendix A.

     One other  short  data file  must  be attached  to  the  program,
containing the number of cases  to  be run,  the length  of  the
driving cycle in seconds,  and  an  indication of the type  of vehicle

-------
tire-wheel-brake assembly  is  known,  an appropriate  adjustment  can
be  made  to  the  vehicle mass  and  entered as  part  of the vehicle
data.   When  this  information  is  not known,  as  is generally  the
case,  standard  (default) estimates are  used.[3]    These  estimates
are built into the program:

     M = 1.035 m      (road)                                    (16)

     M = m  + 0.018 m (dyno)                                    (17)

where:

     M =  total effective mass  of  vehicle (road simulation)  or of
          vehicle-dynamometer system (dynamometer simulation)

     m =  gravitational mass of vehicle

    m  =  equivalent mass  simulated  by the dynamometer flywheels.

The contribution of the rotating wheels to the  total effective mass
is, of course, less on the dynamometer than on  the road, since only
the two  driving  wheels are rotating during dynamometer operation.

     Since the acceleration dv/dt  is  negative  if V(i) < V(i-l),  the
inertia  component  of  total  force  is  frequently negative.   After
calculating the force  acting  on the  vehicle  by equation (15),  the
model tests the signs of the total  force  and  the inertia component.
As was discussed in  the  previous section,  there are  three possible
outcomes of this  dual check:   1) both  are  nonnegative, 2) the
inertia term is negative but  the total force remains positive,  and
3) both are negative.  These  three cases  are treated seperately in
the model, as described below.

     B.   Three Modes of Operation

     In the first case, where both the total force  and its inertia
component are positive,  all of the acting forces  result  in  power
being demanded of the engine.   This condition  is identified in  the
program as "A"-mode,  and can be physically interpreted as accelera-
tion or steady-speed cruise.

     A cumulative  total of  all  "instantaneous"  power demands
represents  the total  energy  demand.  When  all  of the  terms  of
equation (15) are  nonnegative,  the maintenance of  individual suras
representing rolling  resistance,  aerodynamic,  and   inertia demands
corresponds  to  the  actual  physical  breakdown of  energy demand.

     The second case,  where the inertia  component  is negative but
of insufficient magnitude to  provide  for  all  of  the demand of the
rolling resistance  and aerodynamic  components,  is identified in the
program   as  "B"-mode.   Vehicle  operation  in "B"-mode can  be
physically  interpreted  as  deceleration  without  braking.   Stored
kinetic energy resulting from the motion of the vehicle is used to

-------
data to be used and  the  type  of  operation  (dynamometer or road) to
be  simulated.   The  number  of cases refers to  the  number  of dis-
tinct  sets  of vehicle  data  to  be modeled over  the  same  driving
schedule, and  is limited  only  by time  and cost  to the  user.
The  length  of the  cycle in  seconds  is  used  to  control  the num-
ber  of loop iterations per  vehicle.  The EPA and SAE standard
driving  schedules  are all  under 2000 sec;  if  a cycle  of longer
duration is to be  used, the dimensioning of  the  velocity vec-
tor V must be increased.   The  third entry in this  file is "1," "2,"
or  "3,"  corresponding to  the use  of  dynamometer coastdown  data,
road coastdown  data,  or vehicle  design  parameters  respectively.
Simulation of dynamometer operation is  assumed if dynamometer
coastdown data  is input, and road operation  is  simulated  if the
other  two  types  of  input data are  used.   The distinction  between
road and  dynamometer simulation  is detailed  in  the  subsection  on
calculations, below.

     B.   Calculations

     After  f  and f   have been  read into or  calculated by the
model,  there is  a loop of  instructions that  is  executed  once for
each second of the specified driving cycle.  One pass through this
loop is described  here.

     The driving cycle has been  input to a vector V,  where V(i)  is
the velocity at  time t = i,  1 _<_  i _<_ number  of  seconds  in cycle.
The mean  velocity  for the i   second  and  the acceleration during
that second are  calculated as:
     V(i) = [V(O  + V(i-l

 dv/dt(i) = V(i) - V(i-l)
                                          (13)

                                          (14)
The total force acting on  the vehicle during the time segment t = i
is then the sum of the rolling  resistance, aerodynamic, and inertia
components:
          = fQ(i)
f, tv(i)]2
m dv/dt(i)
(15)
     Within the model,  the primary difference in the simulations of
dynamometer and road operation  is  in the handling  of  the  mass in
computing  total  acting force by  equation (2) or  (15).    If  road
operation  is to be  simulated,  then the gravitational mass  of the
vehicle is entered; the equivalent mass  simulated  by  the dynamom-
eter flywheels  is  entered  for dynamometer simulations.  There is
also a  difference  between  these simulations in the estimation of
the effective  equivalent masses of  the  rotating  tire-wheel-brake
assemblies,  which are required for calculation  of  the  inertia
component of total  acting force.

     In either  type of  simulation, if an experimentally measured or
calculated value of the effective equivalent mass  of  one rotating

-------
overcome  part  of the  acting  rolling and  aerodynamic  resistances,
but there is still a net positive force to be overcome.

     When the  vehicle  is in  the  "B"-mode of operation, the  total
net power demand, P =  FV,  is  added  to  the cumulative  total  just  as
is done  in "A"-mode.   However,  the  individual contributions  to the
total energy demand  resulting from rolling  resistance  and  aerody-
namic  losses  must be  treated differently  than  when  in "A"-mode.
The tire  and aerodynamic  energy demands cannot  be simply added  to
their respective component  sums,  since  this results  in the  sum  of
these energy demand components  exceeding  the  total vehicle  demand.
Consequently,   the  proportion  of  the  corresponding   steady-speed
energy demand  that  is  required is calculated, and this proportion
of the rolling  resistance  and aerodynamic terms  is added to  their
respective component sums.  That is, the  actual  total acting  force
is given by:

     F[V(i)]  = f0 + f2  [V(i)]2 + m dV/dt(i)                    (18)

where    /dt(i)  is negative.  During vehicle  operation  at the
corresponding  steady speed, the  total acting force would be:

     F'[V(i)]  = f0 + f2 [V(O]2                               (19)

The proportion of the  total steady-speed  force that is  required  at
V(i),  in "B"-mode,  is  therefore the  ratio of the  actual  total  force
to the corresponding steady-speed  force:

     F[V(i)]/F'[V(i)]                                          (20)

     Thus, it  is  this proportion  of   the  rolling  resistance and
aerodynamic  losses  that are added  to  the respective component  sums:

     FT'  = FTtF(V(i))/F'(V(i))]                                (21)
     FA'  = FA[F(V(i))/F'(V(i))]                                 (22)
      A     A
     In "B"-mode,  the inertia term of equation  (18)  is negative; no
energy  is being  demanded  of  the  engine to  overcome  inertia.
Therefore,  the  component  sum representing  energy  demand  due to
inertial  effects (kinetic energy)  is  not  incremented.

     In the event that  the  inertia component  of force is negative
and -F    > F   +  F   , so that F  < 0, no  power is demanded from
the engine.   This  is  indicated  in the model  as  "C"-mode,  and is
physically interpreted  as  the vehicle braking.   In "C"-mode, none
of the component  sums  representing  individual  energy demands are
incremented .

     All   of  these  calculations  in  the  model  are performed in
SI units;  therefore, integration of power demand over time to yield
energy demand reduces to simply summing the power demands for each
second (without problems  with  units  and conversions).  Similarly,
summing V(i) in m/s  yields  the  total distance traveled in meters.

-------
     It should be noted  that  it  is  the  existence  of the  condition
identified in  the  program as "B"-mode  that necessitates  the  sec-
ond-by-second calculations  of  the model.   If  either  of these
simplifying assumptions are made:   a)  all of the kinetic  energy  of
vehicle motion is returned for use in overcoming rolling resistance
and  aerodynamic  losses,  or  b)  none of this kinetic energy  is
returned  for  useful work;  then  the  equations  used to  determine
energy  demand  reduce  to  a  set of  definite integrals  that  depend
only  on the  characteristics of the driving cycle.   Since the
vehicle parameters would  have no  effect on the solution  of  these
integrals, they can  be  solved  once for  any given cycle.  The
solutions of these integrals and  the vehicle parameters could  then
be combined algebraically to yield  the tractive  energy  requirement
of the vehicle over  that  cycle.

     C.   Output

     The program output  can  be  considered  in  three parts:   the
vehicle information,  the energy demand and its  breakdown, and
auxiliary information.   These  are briefly discussed below.

     The vehicle information listed  includes  the identifying
numbers (a case number  for  that  run  and  an arbitrary vehicle
identification number).   The  input information  is  listed for
reference,  and as  a check  on input  accuracy.   If  vehicle  design
parameters were supplied, they are  listed with  the  calculated
values of the  force equation  coefficients.  If  the  force  equation
coefficients  were derived from road  or dynamometer coastdown data,
they are  listed  along  with backcalculated values  for  the vehicle
design parameters  from  equations (lOa) and (Ha).  The data in  this
block are labeled to indicate what was  supplied as  input  and  what
was calculated by the program.

     Energy demand  has already been discussed; the  total  and  each
of the  contributing terms are labeled  and  listed.   In  addition,
the algebraic  sum of all  of the  inertia  component  energy  terms  is
printed.  In  any driving cycle  with  equal  initial and  terminal
velocities  (zero,  for   the  EPA cycles),  the  true algebraic sum  of
all kinetic energy  terms is  zero.    Therefore,  the value printed
should be near zero; any  significant deviation  from zero is indica-
tive of errors in the modeling program.

     A  few other lines  of  information are also output.   The
time (sec) spent  in each  of  the  three  "modes"  of  operation  dis-
cussed earlier is listed.   The time spent in "A"-mode is a  function
of  the  acceleration characteristics  of the  driving  cycle  used,
and is  vehicle-independent.   For  the EPA urban  and HFET cycles,
"A"-mode consumes  544  sec and  319 sec, respectively.  The  division
of the remaining  time between  the  "B" and "C"-modes is dependent  on
the mass,  the  tire  rolling resistance, and the aerodynamic drag  of
the specific vehicle.

     Distance  traveled  over  the  cycle  is computed  in  meters and
converted  to  miles,  and both are  printed.   The maximum  power

-------
demanded by  the  vehicle  at  the  drive wheels  is listed,  along with
the  time  of its  occurrence  within  the cycle.   The  maximum  power
demands for  vehicles  over the LA4  cycle  occur during the 195th or
196th  seconds,  when  accelerating  at a rate  of approximately 3
mph/sec from velocities of 30-35  mi/hr.  Different vehicles experi-
ence  peak power  demand in  the  HFET cycle at  several  different
times,  depending  on  the  relative contributions of the aerodynamic
(V ) and inertial (dv/dt) terms.

     D.   Basic Error  Check
     A  check  for  gross  error  in  the computer  program  consisted
of  modeling  the energy demand of  two  base-value  vehicles and
several odd  variations of  those vehicles.   The base  vehicles
used  for  this were  a 1980 model  Volkswagen  Rabbit,  representing
a small car,  and a 1981 model  Ford  F150  4x4  pickup  truck,  repre-
senting a  relatively large LDT.   The  three  primary vehicle de-
scriptors  (m, RRC,  CDA) were  then  set  to zero, singly and  in
pairs,  and  the  corresponding  energy demands modeled.   In  these
configurations,  the energy demands of the vehicles can be  checked
algebraically  without  using the  model.

     A  vehicle having a mass of  zero,  while maintaining  its as-
sociated gravitational weight, would  require no kinetic energy over
the cycle:   the  inertia component F  of  total acting  force  would
always be zero.   (Maintaining the gravitational  weight associated
with the original mass is necessary to keep the energy required for
overcoming rolling resistance  from being affected.) Elimination  of
the kinetic energy requirement  means  that the total  energy demand
becomes the  sum  of  the  energy required  to overcome the  rolling
and aerodynamic resistances.

     If the base vehicle maintains  its  mass and  weight,  but  is
assumed to have a frontal area or drag coefficient of zero  so that
C A = 0, then no energy  is  required  to  overcome  aerodynamic  drag.
Similarly,  if the  rolling resistance  of  the  tires  and  tire-wheel-
brake  assemblies  is zero while mass,  weight,   and  aerodynamic
characteristics  are unchanged, then no energy  will  be  required for
overcoming  rolling resistance.

     Setting  two of  these three  primary  descriptors  equal  to
zero  at a  time,  three  other  "dummy" vehicles  are  generated.
Each  of these last  three vehicles  have a total energy demand
equal  to one  of  the  components:   rolling resistance,  aerodynamic
drag,  or kinetic.   Running  these  "dummy" cases  through  the  model
revealed  no  errors  in  the  handling of equation (15)  and its
terms.

     To this  point,  discussion has  focused  on  the concepts  that
form the basis for the calculation of energy demand,  and  on imple-
mentation  of the  model as a computer program.  The last  sections  of
the report present several suggested  applications  of  the program.

-------
IV.  Uses of the Model

     The  use  of vehicular  energy  demand modeling can  be  divided
into  two general  groupings: prediction of  the  relative  energy
requirements of vehicles of  the  future,  and  as  an aid in both the
selection of vehicles for  test programs and  the evaluation of
test results.

     A.   Future Vehicle Performance

     As was briefly discussed in Section II,  specification  of four
vehicle  design  parameters  (mass,  frontal  area,   coefficients  of
rolling resistance and aerodynamic  drag) allows  calculation of the
force equation coefficients,  and hence of cycle energy demand.  One
major advantage of any modeling exercise  is the ability to consider
systems that may not  be available  for  direct empirical  investiga-
tion.  One of the  first uses of this model was to predict the fuel
economy  that  may  be  possible with  current  and  future  optimized
technology vehicles.

     Assumptions.   Several  assumptions  were  made  to  simplify and
enhance prediction of  fuel  consumption  from  modeled  energy demand
of the  optimized  vehicles.  Foremost among  these was  the use
of a  continously  variable transmission (CVT).   Such transmis-
sions have  already been successfully  installed  in  some  European
vehicles,  and  are currently under  further  development  by  Borg-
Warner.[6]   Furthermore,  it was  assumed that the  CVT would be
controlled,  probably  through  the  use of  microprocessors,  to
always seek the point  of maximum engine efficiency possible under
the required  loading.   A  final  related  assumption  was that the
transmission and drivetrain  system had  a  constant  efficiency  of
ninety percent;  that  is, only one-tenth  of  the engine power avail-
able at the flywheel  is lost before reaching  the axle of the drive
wheels.

     Under these  assumptions the  model can  be  used to  directly
calculate fuel consumption,  if fuel consumption  rates are given as
a  function of  demanded power.   This  information  can be  obtained
from a simple one—dimensional engine map.  A  set  of  points repre-
senting fuel consumption as  a function  of power demand  are  taken
from this map, and linear  interpolation  is used to determine fuel
consumption at  intermediate levels of power demand.

     Optimized Vehicles.  Three general size categories of vehicles
were chosen for this  investigation.   In order of  increasing size
and weight,  these were (i)  two-passenger vehicles, (ii)  four-
passenger vehicles, and  (iii) five-or-six-passenger   vehicles  and
light-duty  trucks, combined in one group.   For each  of  these
classifications  based  on size, two  levels of  technology,  described
as current best (CB)  and advanced technology  (AT)  were considered.
The technological  level refers to values taken for  the four design
parameters required  to calculate  the  coefficients  of  the  force
equation.   The CB values have been achieved  in  vehicles  available

-------
today,  although not  necessarily all  simultaneously in  a single
model.   Advanced technology  is  better,  in the  sense  of  lowering
energy demand,  than  technology  in use  today;  however,  this can be
achieved within  the  framework of current  technology.   The values
chosen for the design parameters  of  these vehicles are presented in
Table 1.

     A  few  comments  regarding the values  assumed  for  the vehicle
parameters for  the current  best (CB) and  advanced  (AT)  levels of
technology should be  noted.   In  the  case of vehicle mass, there are
production vehicles  representing each of  the three  size classes
discussed having masses near or below the masses  listed for the CB
vehicles.  In  the  two-passenger  category,  the MY1980 Daihatsu Max
Cuore, with a mass of  545  kg, [7] is already nearer  to the AT than
to the CB mass.  The four-passenger MY1980 Audi L, with a carrying
capacity of about 1,000 lb, has only 910 kg mass.[7]   This is
almost exactly the mass assumed  for the CB four-passenger vehicle.
One version of  the Volkswagen Jetta,  the GLI, can carry nearly as
great  a  load (948 Ibs), yet its mass is less  than 800  kg  in a
2-door version.   The  mass  of the  MY1980  Volkswagen front-wheel-
drive  diesel  pickup  truck, a  vehicle  fitting the  largest  of
the three size categories discussed,  is only 928 kg.[8]

     Rolling   resistance  of  automobile  and light  truck  tires  is
frequently expressed  as  a  rolling  resistance coefficient (RRC).
The RRC  indicates  the  units  of  rolling  resistance   force  (lb,  N)
per thousand  units  of vertical  load.    The   assumed current  best
value,  0.008, is already  being attained, at Least in some  low
rolling  resistance  tires.   In an ongoing  EPA  test  program to
measure  the  rolling  resistance  of  tires,  data collected  to  date
indicate that there  are  tires widely  available in today's market-
place with flat-surface RRCs  in  the  range  of  0.0082  to 0.0090, [9]
when  tested  in  accordance   with  the EPA Recommended Practice  for
Determination of Tire  Rolling Resistance Coefficients  (80 percent
of design load,  at 35 psi).[10]   The continuing efforts by the tire
industry to  reduce   rolling  resistance  seem  likely  to result  in
tires  having RRCs at  or below the 0.007 assumed  for advanced
technology.

     Aerodynamic  drag coefficients  lower than  the  assumed current
best value of 0.4  are available in some vehicles today,  and coeffi-
cients lower  than  the assumed  advanced technology  value of 0.3 will
be available in  the relatively near future.  Recent research
conducted by Volkswagenwerk  AG  indicates  that reductions  in  drag
coefficients  of as much  as 45  percent  over  today's range (about
0.35  to  0.55)  should  be  achievable with  mass-produced  passenger
cars.[11]   Ford Motor Company is  continuing development of  a
prototype vehicle,  currently known  as  Probe, which  has a drag
coefficient  of 0.22.[12]   On the basis  of these statements,  the
assumed values of 0.4 (CB)  and 0.3  (AT)  certainly appear feasible.

     The  reductions  in  vehicle frontal area listed in Table  1  are
relatively  small.  Lower vehicle masses and  improved  aerodynamic

-------
                                        Table 1
Vehicle
Size
Class
Two-
passenger

Four-
passenger
5 and 6-
passenger
and
'personal1
light
trucks
Assumed
Level of
Technology
Current
best (CB)
advanced
(AT)
CB
AT
CB
AT
Mass*
(kg)
680.4
476.3
907.2
635.0
1134.
793.8
C A2
RRC D (m )
0.008 0.4 1.7
0.007 0.3 1.6
0.008 0.4 1.7
0.007 0.3 1.7
0.008 0.4 2.0
0.007 0.3 1.9
C°2 fO f?
(m ) (NJ (kg/m)
0.68 53.3 0.398
0.48 32.7 0.281
0.68 71.1 0.398
0.51 43.6 0.298
0.80 88.9 0.468
0.57 54.5 0.334
Engine
Max imum
Power
(hp)
25
18
31
23
38
28
Scaling
Factor
0.357
0.257
0.443
0.329
0.543
0.400
*    Values in this column are test weights.  Curb weights would be approximately 136 kg
(300 Ib) lower.

-------
design  are  likely to  result  in the  assumed  decreases  in  frontal
areas.

     Engine Sizing.  Maximum power  outputs  required  for  the  engines
of these six vehicles were calculated  to  allow adequate  performance
with  very little  excess power.  The minimum performance conditions
were  (i)  that  the vehicle be capable of a 55 mi/hr cruise on any
roadway having  no more than  a  five  percent  grade,  and (ii) that
zero  to 50 mi/hr  acceleration take no longer  than fifteen seconds.
The first of these conditions proved  to be the more  stringent:  if
the  engine  is  capable of maintaining  the  vehicle speed at  55
mi/hr on  a  five  percent  grade,  then zero to  50 mi/hr acceleration
will take approximately thirteen seconds.

     The power required to meet  the  most  severe of these conditions
was  then  increased to allow  for  the power demand  of accessories
that do not contribute to vehicle motion.  The increase for acces-
sories was two hp for the two-passenger  vehicle,  three  hp  for the
four-passenger,  and four hp for  the  5/6-passenger cars and personal
recreational light trucks.  These values  represent estimates of the
maximum accessory power demand,  not  the actual power  being demanded
for  accessories  while driving.    In  estimating energy  demand and
fuel  economy  for these vehicles,  it  was assumed that  an  average
in-use accessory power demand  of 0.5 hp was operating continuously.

     The total power requirements  were then rounded  up to the next
integer horsepower levels.  The  maximum power  output  of the  engines
derived from these requirements  ranged from 18 hp for the advanced
technology two-passenger  vehicle  to  38  hp  for  the current  best
technology 5/6-passenger vehicle.  These values are listed in Table
1.

     Fuel  Consumption.  The final step necessary was "development"
of  fuel  consumption  versus  power   output  maps for  the optimized
engines  of these vehicles.   The  logical starting point was  a
small,  fuel-efficient engine  for which  the required data are
currently  available.   One  such engine  is  the  seventy  horsepower
turbocharged prechamber  diesel  engine  manufactured  by  Volks-
wagen. [13]  However,  the rated power of this engine,  70 hp,  is well
in excess of the maximum power requirements of  all six  of the
optimized vehicles  when driven  over  the  EPA driving cycles.
Consequently,  the engine  maps for  these  vehicles were obtained  by
linear scaling of this map.   That  is, if an engine  with a maximum
power of 30 hp was required, it was assumed that this engine would
develop 30  hp with  the  same  thermal efficiency   that  the  70  hp
engine had at 70  hp.   Similarly, the scaled-down engine was  assumed
to have the  same thermal  efficiency at one-half maximum  power  as
did the 70 hp engine  at half power.

     Since the 70 hp  engine map used  in  deriving  fuel consumption
as a  function of  power demand is  a  prechamber engine, and direct-
injection (Dl)  engines  are more  fuel-efficient  than indirect-
injection  (IDI)  engines,   full vehicle/engine optimization suggests

-------
the use of DI engines.  There is evidence  that  the  fuel consumption
of  a  DI diesel engine  is  approximately  twelve percent  lower than
that of a comparable IDI diesel engine. [14]

     To  simulate  the maximum  fuel  economy  improvements  possible,
the fuel  consumption rates from the 70 hp  engine  map were multi-
plied by 0.88  (reduced by  12 percent) before being entered as data
for the model.   In other words, it is assumed that the engines of
these six optimized  vehicles will  be  DI  engines,  and as  such will
operate with the same decrease  in  fuel consumption relative to IDI
engines that is seen today.

     After  the  power demand P(i),  including  the 0.5  hp  accessory
load,  has been computed for a given second t=i  of the driving cycle
it  is  sent, along with  the table  of  fuel  consumption versus  de-
manded  power,  to  a  linear  interpolation  subroutine.   The  subrou-
tine determines the  interval of power demand within the table that
contains P(i) and  interpolates  when necessary to determine the fuel
consumption  rate.   This rate  (g/kWh)  is  multiplied by  the  power
demand (kW)  and divided by  3600 (s/h) to yield  the fuel consumption
(g) during the i   second.

     The resulting value  for fuel  consumption is  returned  to  the
main  program  and  added  to  the  cumulative .fuel  consumption.
This  completes  the  calculations  for the  i   second, and  the
programs  goes  on to  evaluation  of the (i+l)th  second.  Fuel
consumption is determined  by this method for all  of  the  time
that  the vehicle is  in  "A"-mode (accelerating  or cruising).
This  method  is also  used  to determine  fuel consumption when
the vehicle  is  in  "B"-mode (powered deceleration),  except  for  the
case described in  the following section.

     Idle Fuel Flow Rate.   There are several situations encountered
in  the  course  of  the EPA  cycles where the  method  described  above
cannot  be  used to  determine  fuel  consumption.   These  are:   (i)
idling,  when  the  vehicle  velocity is  zero, (ii)  operation  in
"C"-mode,  when  the calculated  power demand  is  negative,  and  (iii)
those  seconds of operation  in "B"-mode where the power demand is so
small  (approximately  0.6  kW or  less) that the interpolated value of
fuel consumption is less  than the idle fuel flow rate.

     In these  three  situations, an idle  fuel  flow rate of  0.12
gal/hr (0.107 g/s)  is assumed.   The  conversion of the measured 0.12
gal/hr  idle  fuel  flow rate  to 0.107 g/s is  based on density  of
7.078  Ib/gal for diesel #2  fuel.  This density is in turn based on
an assumed API gravity of 35",  corresponding to the midpoint of the
range  of API gravity  (33° to  37°) specified for diesel #2  as an EPA
test fuel.

    The 0.12  gal/hr idle   fuel  flow rate  was measured  in  tests
conducted  at  EPA  on an Integrated Research Volkswagen (IRVW)
safety vehicle.[15]   This vehicle was  equipped  with  a 70  hp turbo-
charged diesel engine,  which  is  equal  in  maximum power to  the
engine of the map  used  to derive  the table of fuel consumption as a

-------
function  of power  demand.   Since  the maximum  powers of  these
engines are equal,  the idle fuel  flow rate was adjusted  in the same
way  that  the  fuel  consumption values at given  power demands were
adjusted:   The base value  for idle fuel flow was multiplied by the
ratio  of  the maximum power  output  of the engine sized  for the
vehicle to the maximum  power  output  (70 hp)  of the engine used in
the  determination  of  fuel  consumption  as a  function of  power
demand .

     When   the  vehicle  is  operating  in "C"-mode,  the  scaled idle
fuel  consumption  is added  to the  cumulative  fuel consumption
without calling  the interpolation  subroutine.   The other two
situations are treated within the subroutine:   Before a one-second
fuel  consumption value is returned to the main  program,  it  is
checked against the scaled idle flow rate.   If the calculated fuel
consumption is less  than  that rate,  the scaled  idle flow  rate  is
returned instead.   This  ensures that  fuel consumption in any second
is never less than  the scaled  idle fuel  flow in one second.

     Testing  the Method.   The  general  method  was  tested by  using
data representing  a Volkswagen Rabbit.   Input  data  was in the form
of vehicle design parameters,  as shown below:

     mass         1077.3  kg
     RRC        0.0120
     CA        0.77    m
The  rolling resistance coefficient  was chosen  as representa-
tive of a low (but not exceptionally low) rolling resistance tire,
while  the values  of mass  and C A  actually describe a Rabbit.

     The fuel consumption versus  power  demand map was from a 50 hp
naturally aspirated Volkswagen diesel  engine. [13]   This  engine is
very near in power to the 52 hp diesel  engine that  is available in
MY1981  Rabbits  manufactured  by VWOA,[7]  and hence  the  fuel  con-
sumption figures were not  scaled.  Accessory  load (0.5 hp) and idle
fuel flow rate (0.107 g/s) were treated in the same way as for the
six optimized vehicles.

     The FE  projections  from  the  energy demand model were  53  MPG
for  the  LA4  cycle and  61  MPG for  the HFET cycle.  The  urban FE
value represents an  increase  of  approximately  25  percent  over  the
42 MPG achieved by the most fuel-efficient of the MY 1981 Rabbits.
[16]   Since fuel  economy  increases of  20  percent  are  possible
through use of CVTs,[6l  the modeling  approach does not appear to be
overly optimistic.

     Fuel economy (results).   The  fuel  economy projections for  the
six  optimized vehicles  are  shown  in Table 2,  along with  the cor-
responding energy demands.  The model  indicates  that fuel economy
in  the  range  of 90 MPG could  be  obtained by the two-passenger CB

-------
Table 2
Urban Cycle
HFET Cycle
Estimated
Vehicle
Size Class
Two-
passenger


Four-
passenger
•
5 or 6-
passenger
and 'personal'
Assumed level
of technology
current best
(CB)
advanced
(AT)
CB

AT
CB
AT
Energy
demand
(MJ)
2.81

1.93

3.43

2.38
4.21
2.88
Fuel
Economy
(MPG)
diesel gas
92

127

77

105
64
89
83

114

69

94
57
80
Energy
demand
(MJ)
4.88

3.36

5.38

3.84
6.48
4.46
Estimated
Fuel
Economy
(MPG)
. diesel gas
87

123

79

109
66
94
78

111

71

98
59
85
light trucks

-------
technology vehicle; this increases to approximately 120  MPG  for  the
advanced technology two-passenger vehicle.   The increases  in  FE  for
the AT  vehicles over the  CB  vehicles,  in the other  size classes,
were also in the 35 to 40 percent range.

     To estimate gasoline-equivalent  fuel  economy,  the greater heat
content per unit volume of diesel #2  fuel  over typical gasoline  was
taken  into  account.   The  ratio of  these  heating values (138,700
BTU/gal  for diesel  #2, and  125,000 BTU/gal for gasoline[17]),
138.7/125 =  1.11,  was multiplied by  the cycle fuel consumption to
estimate gasoline-equivalent  fuel  consumption.  The resulting
gasoline-equivalent  fuel   economy  projections are also  shown  in
Table 2.

     An aspect of  the optimization of all  four parameters (m, RRC,
C , A)  that  was not  anticipated was the  reversal of the ranking
or  urban and  highway  fuel  economy from the commonly  accepted
order.  For all of the  vehicles  in the  1981 Gas Mileage Guide[16],
highway FE is greater than urban FE.  This relation also holds  for
the two larger size classes of vehicles used  in these projections,
although the  gap between urban and highway  FE  for  these  vehicles is
relatively small.  For  the two-passenger vehicles, both CB and AT,
projected city FE exceeds  projected highway FE.        .

     Table 3  shows  the relative contributions  of  kinetic  and
aerodynamic   energy demand to  total   energy  demand for  these   six
vehicles, over both of  the EPA cycles.   The  kinetic  energy demand
dominates the urban cycle, while aerodynamic  drag  dominates on  the
highway cycle.  The fuel economy reversal of  the  two  cycles occurs
because for the chosen  optimized values of the vehicle  parameters,
the decreases in the kinetic energy  requirements  for  urban driving
are much  greater  than  the decreases in  the aerodynamic  energy
requirements  for expressway driving.

     In addition  to  the  use  of  the model to estimate potential
improvements   in  energy  demand  and  fuel economy,  there are other
uses  that  apply to future and  current  vehicles,  and to testing.
Sensitivity of  energy  demand  to changes  in   the  primary vehicle
parameters can be  investigated,  as discussed  in  the next section.
The last  section  briefly  touches  on some  of the other possible
applications  of the model.

     B.    Sensitivity  Analysis

     An energy  demand  model  can be useful   in  investigating   the
impact  on energy  demand of changes in different vehicle  para-
meters.    The  effect  of  small  variations in aerodynamic character-
istics (C A), for  example, may  be  obscured in the "noise" of data
from actual  tests.  This assumes  that two vehicles  that  are identi-
cal except for the parameter of  interest can  be found for testing;
this  is  often not the  case,  and in  some  instances (e.g.,  change
frontal area  only)  it  may  be  impossible.

-------
                    Table 3
Vehicle
Size
Class
Two-
passenger
Four-
passenger
5/6-passen-
ger & light
truck

Assumed level
of technology
Current best(CB)
Advanced (AT)
CB
AT
CB
AT

LA4
EK
E
0.508
0.519
0.555
0.559
0.565
0.578

cycle
EA
E
0.278
0.285
0.221
0.239
0.210
0.218

HFET
IK
E
0.162
0.164
0.195
0.192
0.203
0.206

cycle
!A
E
0.607
0.620
0.536
0.565
0.521
0.538

where:

     E

     EA
total energy demand over cycle (at drive wheels)
kinetic energy requirement
aerodynamic energy requirement

-------
     Table 4  presents  the  results of running an abbreviated sensi-
 tivity  analysis using  the  basic energy  demand  model.   The four-
 passenger vehicle  with "current best technology" design parameters
 from  the previous  section was  selected  as  the vehicle  for  this
 analysis.

     First,  the vehicle has  its  energy  demand modeled  using  the
 values of mass,  RRC,  and  C  A as given  in Table  1.  The  energy
 demands  over  both  the LA4 and HFET  cycles are then  used  as  base
 values,  along with  the maximum power  demand  in each  cycle.   The
 CB  four-passenger  vehicle  is then  run through the  model  several
 additional times;  in each  of these runs, one  of the three vehicle
 descriptors is  either  increased or decreased by ten  percent.   The
 results  of the  additional  modelings  are compared with the baseline
 results  to examine  the sensitivity of energy demand to the changes
 in the design parameters.

     The numerical  results of these modelings are  listed  in Table
 4,  along with  the  percent changes  in energy  demand  and  maximum
 power demand  associated with  ten  percent  changes in mass,  RRC,  and
 C A.   Over the LA4 cycle,  ten percent  changes in vehicle mass
 resulted in  approximately  seven percent  changes  in energy demand.
 Ten percent variations  in aerodynamic behavior, as characterized  by
 C A, caused  changes  in the  vicinity of  three  percent in  energy
 demand;  and  ten percent changes in rolling  resistance,  as  charac-
 terized  by  the RRC,  changed  energy  demand  by about  two  percent.

     In  the  HFET  cycle, with its much greater  average speed  and
 relatively minor  accelerations and  decelerations,  changes  in  the
 aerodynamic  parameter   C A  caused the  greatest  changes  in  total
 energy demand:   approximately six percent,   for  ten percent  varia-
 tions  in C A.  Increasing  or decreasing the  vehicle mass  by  ten
 percent changed energy  demand by  just  less  than  four percent.   The
 percentage effects  of  rolling  resistance changes   were  almost  the
 same as for the LA4 cycle,   around two percent.

     C.   Additional Applications

     Selection of Vehicles  for Testing.    Another potential  use  of
 an energy-demand model is  assistance in  selecting  candidate vehi-
 cles for some testing  programs.  Consider  a hypothetical  example:
Assume  that  a  test  program  has  been  planned  to   investigate  the
 differences between dynamometer simulation accuracy  for front—wheel
 drive (FWD) and rear-wheel drive  (RWD)  passenger cars;  and  further
 that sufficient test data exist to suggest the  FWD vehicles  experi-
 ence greater  loading  than  do RWD  vehicles  when tested on  dynamo-
meters.  The  primary  objective  of this hypothetical  test  program
 is to determine if  FWD vehicles actually  are more  severely  loaded
 on the  dynamometer relative to RWDs.

     By  entering  the available test  data in  this  model and com-
 puting the energy demands, valuable  insight could  be gained as  to
which vehicles  would  best  be chosen  for  testing.   Pairs of vehi-
cles,  one  FWD  and one RWD,  that exhibit  similar  internal and

-------
                              Table 4
Baseline vehicle parameters:
mass 907
RRC 0
C A 0

-------
external characteristics but  have modeled energy  demands that
differ  greatly would  be  good  test  candidates,  as  would  any  FWD
vehicle  whose  modeled  energy demands differed widely depending on
whether  road-derived  or dynamometer-derived data  were  input.  By
this exercise, the vehicles that are most likely  to demonstrate  the
suggested problem  will be selected  for testing.   (This  is  cited
only as  an  example.   The FWD/RWD simulation accuracy is dependent
on  the  distribution  of vehicle weight  between  the front  and rear
axles.)

     In  a similar  vein,  this model  could aid in the  discretionary
selection of  vehicles for  confirmatory  testing.   For all test
vehicles, the  required information  for  modeling  energy demand on
the  dynamometer  is  available.    Vehicles  that display  ratios  of
modeled  energy demand  to  fuel economy (MJ/MPG) that  are unusually
high or  low would  be obvious candidates for confirmatory testing.

     The question  that  can be answered  using the model  is:   Based
on  the modeling  parameters,  which  of the vehicles  of the possible
test group  are most likely  to  exhibit  the  highest  levels  of NOx
emissions and  the  lowest  fuel economy?   Consider  a pair  of  vehi-
cles sharing engine  and  transmission/drivetrain characteristics in
common, one of which will  be a discretionary choice  to be tested.
In  such  a case,  vehicle A may have a greater mass than vehicle B,
while vehicle  B  has  a higher  value for  total  dynamometer   power
absorption.   By modeling  the energy  demand of  these vehicles
and  choosing  the  vehicle  with  greater  energy demand for  confir-
matory testing, the lower of  the two in  fuel economy  (and probably
the greater  in NOx emissions) will  be tested.

     Test result  guidelines.    When  a  fuel economy-related  test
program  is  planned,  the  test vehicles  could  be  run  through the
model  before  testing  commenced.   The results  of the pre-test
modeling can serve as  guidelines  in evaluation of the  test   data,
flagging  test  results   that  deviate beyond  reasonable  allowances
from the  modeled  output.    Such  a  pre-test exercise might   allow
earlier  detection and correction  of errors  that otherwise may
have gone undetected, and been suspected only at the conclusion of
the program.   Another  benefit is that  more  test  time can  then be
concentrated in  areas,  or  on vehicles,  where apparently anomalous
results are  occurring.

     This is not a complete  list  of the potential  uses  of  a model
of  this  type.    In general,  it  is  attractive  because it  provides
insight  into many  aspects of vehicle  energy demand  at  low   cost.

-------
                            References

     1.      Hamming,  R.W.,   Numerical Methods for Scientists  and
Engineers, McGraw - Hill, 1962.

     2.    "The  Measurement  of Passenger  Car  Tire Rolling Resis-
tance," SAE Information Report J1270,  October  1979.

     3.    "EPA  Recommended  Practice  for Determination of Vehicle
Road Load," Attachment I  to  OMSAPC Advisory Circular 55B, December
6, 1978.

     4.    "Development  of  the  Federal Urban  Driving  Cycle," by
Ronald  E.  Kruse and  Thomas  A. Huls,  SAE  Paper 730553,  May 1973.

     5.    "Amendments to the Report  on Development  of  a Highway
Driving Cycle  for  Fuel Economy Measurements,"  by C.  Don Pausell,
EPA-AA-ECTD-74-2, April 1974.

     6.    "Belt  Drive CVT  for  1982  Model Year," Automotive En-
gineering, Vol. 88  No. 2,  February 1980.

     7.   Automobile  Club  of Italy, World  Cars  1980, Herald Books,
1980.

     8.   "Automotive Trendsetters - Fuel  Economy Reigns Supreme,"
Automotive Industries, Vol. 160 No.11,  November  1980.

     9.    "Rolling   Resistance Measurements  - 106  Passenger  Car
Tires," by Gayle Klemer,  EPA-AA-SDSB-81-3,  to be published.

     10.   "EPA Recommended Practice  for Determination of  Tire
Rolling Resistance  Coefficients,"  by Glenn  D. Thompson, March 1980,
amended August  1980.

     11.  "Necessity  and  Premises  for  Reducing  the Aerodynamic Drag
of Future  Passenger  Cars," by R.  Buchheim, K.-R.  Deutenbach,  and
H.-J. Luckoff,  SAE  Paper  810185, February 1981.

     12.   "The Fords in  Our Future," Automotive  industries,  Vol.
161 No. 3, March 1981.

     13.. "Data Base for Light-Weight Automotive Diesel Power
Plants," by B.  Wiedemann  and P.  Hofbauer, SAE  Paper  780634,  June
1978.

     14.   "Sofim  Small  High-Speed Diesel  Engines -  D.I.  Versus
I.D.I.," by V.G. Carstens, T.  Isik, G. Biaggini,  and  G.  Cornetti,
SAE Paper 810481, February 1981.

     15.  "Evaluation of Two Turbocharged Diesel Rabbits," by
Edward  A.  Earth and  James M. Kranig, EPA-AA-TEB-80-4, October
1979.

-------
                        References (cont'd)

     16.    "1981  Gas Mileage Guide  -  EPA Fuel Economy  Estimates,"
First Edition,  September 1980.

     17.    Oak  Ridge  National  Laboratory,  Transportation  Energy
Conservation Data Book: Edition  4, ORNL-5654  for  U.S.  Department  of
Energy, September 1980.

-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
3i
32
33
34
35
36
37
38
39
40
41
42
43
44
/iK
tD
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60"
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C















C
C
C
C
C
C



C
C
C




C
C
C
C
C
C
C

C
C
C

                     APPENDIX A

  ««»««   LDV ENERGY DEMAND MODEL   **»»»

  VARIABLE DIMENSIONS:  VMASS, IMASS(KG); CDA(M»»2>; RHO(KG/M«*3)i
    Dl(M); D2(MI); V» DVDT, VMEAN(M/SM AOtMI/HK-SH A2(HR/MI-S)I
    FO, FTOT, FCON, FVSQ» FORCE, FORCE2, TIREF, AEROFt FINER(N)I
    F2(KG/M); POWERt POrt£R2» TIREP, AEROP» PINER, AHPW(W)* TlMEt
    TMAX» A, B, C(S); KW, SKW(KW); ENERGY, TIRENG, AEROE, POSKIN,
    NETKIN(J); ACC, POWHP, PMAX, AHP(HP); FUEL, SFUEL(G)» IDLE,
    SIDLE(GXS); FC(G/KW-HRM MPG(MI/GAL).
  i/o ASSIGNMENTS:
REAL V(2000)
INTEGER A, B, C, I,
INTEGER NCAR, NSEC,
 6 = OUTPUT FILE
 5 = INPUT FILE:
 4 = INPUT FILE:
 3 = INPUT FILE:
 2 = INPUT FILE:
                                    DRIVING CYCLE
                                    VEHICLE DATA
                                    NCAR, NSEC, DTYP
                                    FC/PWR TABLE
REAL VMASS, IMASSf
REflL 01t D2, DVDT,
REAL TIREF* TIREP*
REAL AEROF* AEROP,
REAL FORCE* POWER,
REAL FORCES, POWER2,
 J, K»
 DTYPt
PMAX
VMEAN
TIRENG
AEROE
ENERGY
  POWHP
L
VEHID,
TMAX
REAL»8 FINER, PINER, POSKIN, NETKIN
REAL F2» RRC, CDA, RHO
REAL AHP, AHPW, TIME
REAL AO, A2, FCON* FVSQ, FTOT
INTEGER NMAP, FLAG
REAL KWC15), FC(15), MPG» FUEL* SFUEL, IDLE
REAL SKW(15), SCALE, SIDLE, BTU, MAXHP, MAPHP

  A,B,C USED TO DETERMINE TIME SPENT IN EACH MODE:
   A-MODE:  FORCE > 0 5, FINER > 0
   B-MODE:  FORCE > 0 & FINER < 0
   C-MODE:  FORCE < 0 & FINER < 0

A = 0
B •= 0
C •= 0

  READ IN KW/FC TABLE FOR INTERP

READ(2,100) MAPHP, NMAP
READ(2,103) (KW(I)fFC(I)t 1=1,NMAP)
IDLE = 0.107
BTU = 1.11
RHO = 1.17

  NCAR = NUMBER OF CASES IN RUN
  NSEC = LENGTH OF DRIVING CYCLE
  DTYP = 1 (DYNAMOMETER COASTDOWN DATA)9
         2 (ROAD COASTDOWN DATA), OR
         3 (VEHICLE DESIGN PARAMETERS)

READ(3,106) NCAR, NSEC, DTYP

  READ DRIVING CYCLE, CONVERT MI/HR TO M/S

READ(5,109) (V(I)» I=1»NSEC)

-------
                                  APPENDIX A (cont'd)

 61           DO 1  I =  1, NSEC
 62              V(I) =  0.447«Vd)
 63         1 CONTINUE
 64     c
 65     c       LOOP:  ONCE PER VEHICLE
 66     C
 67           DO 10  K = 1, NCAR
 68     C
 69           WRlTE(6tll2) K
 70     C
 71     C        TYPE OF INPUT DATA
 72     C
 73           IF (DTYP.EQ.3)  GO TO 3
 74           IF (DTYP.EQ.2)  60 TO 2
 75     C
 76     C        DYNAMOMETER COASTDOWN DATA
 77     C
 78           REftD<4,115) VEHID» MAXHP, VMASS. AHPt TIME
 79           FTOT = 
-------
                                APPENDIX A  (cont'd)

 T^T]     c
 122      C        SCALE  FC/KW ARRAY,  IDLE PC
 123      c
 124            SCALE  =  MAXHP/MAPHP
 125            SIDLE  =  IDLE»SCALE
 126            DO 5   L  =  1»  NMAP
 127               SKW(L)  = KW(L)«SCALE
 128         5  CONTINUE
 129|           WRITE<6,142)  SCALE
 130      C
 131      C        INITIALIZE  RUNNING  SUMSt  PMAX»  TMAX TO ZERO
 132      C
 133            Dl  = 0.
 134            AEP.OE  =  0.
 135            NETKIN = 0.
 136            POSKIN = 0.
 137            TIRENG = 0»
 138            ENERGY = 0.
 139            ENGINE = 0.
[1401           SFUEL  =  0.
 141            PMAX = 0.
 142            TMAX = 0
 143      C
 144      C        LOOP:  ONCE PER SEC OF CYCLE
 145      C
 146            00  9 1 - It NSEC
 147      C
 148     -C          CALCULATE DVDT,  VMEAN, &  DISTANCE TRAVELLED
 149      C                      ,
 150               IF  (I.EQ.l)   GO  TO 11
 151               J = 1-1
 152               DVDT  =  V(I)-V(J)
 153               VMEAN =  (V(I)*V(J))/2.
 154               GO  TO 22
 155        11     CONTINUE
 156               DVDT  =  V(I)
 157               VMEAN = V(I)
 158        22     CONTINUE
 159               01  =  Dl+VMEAN
 160      C
 161      C          CALCULATE FORCES AT ITH SECOND
 162      C
 163               TIREF = FO
 164               AEHOF = F2
-------
                                APPENDIX A (cont'd)

         C           ADO ACCESSORY LOAD TO POWER DEMAND
                    POWHP = POWHP+0.5
         C
                  IF (POWHP.GT.PMAX)   TMAX = I
                  IF (POWHP.GT.PMAX)  . PMAX = POWHP
                  POwt.R2 = PUWHP*745.7
                  CALL  INTERP(SKW»FC»POWER2»SIDLE»NMAP»FUEL»FLAG)
                  SFUEL = SFUEL+FUEL
                  IF (FLAG.EU.l)   WRITE(6,199) I
                  TIREP = TIREF»VMEAN
                  PINER = FINER^VMEAN
                  AEROP = AEROF»VMEAN
                  TIRENG = TIRENG+TIREP
                  NETKIN = NETKIN+PINER
                  POSKIN = POSKIN+PINER
                  AEROE = AEROE+AEROP
                  ENERGY = ENERGY+POWER
                  GO TO 8
         C
         C           FORCE IS POSITIVE AND FINER IS NEGATIVE:   PROPORTION
         C           ENERGY DEMAND TO  ROLLING AND AERODYNAMIC  RESISTANCES
         C
             6     CONTINUE
                  B  = 8*1
                  FORCE2 = T1REF+AEROF
                  TIREF = TlRtF»(FORCE/FORCE2)
                  AEROF = AEROF*(FORCE/FORCES)
                  TIREP = TIREF»VMEAN
                  PINER = FINER*VMEAN
                  AEROP = AEKOF»VMEAN
                  POWER = FORCE2*VMEAN
                  POWERS = POWER/0.9
                  POWHP = POWER2/745.7
21^      C
215      C           ADD ACCESSORY LOAD TO POWER DEMAND
216                 POWHP = POWHP+0.5
217      C
218               IF (POwHP.GT.PMAX)   TMAX = I
219               IF (POWHP.GT.PMAX)   PMAX = POWHP
220               POWER2 = POWHP*74.5.7
221               CALL  lNTERP(SKW,FCtPOWER2»SIDLE»NMAPfFUEL»FLAG)
222               SFUEL = SF.UEL*FUEL
223}              IF (FLAG.EQ.l)   WR'ITE (6,199) I
22<*               TIRENG = TIRENG*TIREP
225               NETKIN = NLlMN + PINER
226               AEROE = AEROE+AEROP
227               ENERGY = ENERGY+POWER
228               GO TO 8
229      C
230      C           FORCE IS NEGATIVE:  CALCULATE INERTIA TERMS AS CHECK
231      C
232          7     CONTINUE
233               C  = C+l
23^               PINER = FINEROVMEAN
235               NETKIN = NETKIN+PINER
r?3T]              SFUEL = SFUEL*SIDLE
237      C
238      C           (THREE BRANCHES REJOINED)
239      C
240          8     CONTINUE

-------
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
                      APPENDIX A (cont'd)
  9 CONTINUE

      OUTPUT:  BREAKDOWN OF  ENERGY DEMANDED

    WRITE(6t145) ENERGY,TIRENG,AEROE,NETKINtPOSKIN

      COMPENSATE FOR ESTIMATED CVT EFFICIENCY

    ENERGY = ENERGY/0.9
    WRlTE(6,l4b) ENERGY
    02 = D1/16U9.34
    WKITE(6,151) 01, 02
    WRTTE(6,154) A, B» C
    WRITE(6,157) PMAX, TMAX

      CONVERT PC TO FE

    SFUEL = SFUEL/3218.
    MPG = D2/SFUEL
    WRlTE(b,160) SFUEL, MPG
    SFUEL = SFUEL*BTU
    MPG = D2/SFUEL
    WKITE(6,163) SFUEL, MPG

      RESET "MODE" INDICES

    A = 0
    B a 0
    C = 0
 10
100
103
106
109
112
115
118
121
124
127
130
133
136
139
142
145

148
151
154
157
160
163
199
                                                                      HP'
 END OA LOOP

CONTINUE
FORMAT(F5.1,16)
FORMAT(F5.1,F6.i)
FORMAT(315)
FORMATU6F5.1)
FORMAT('CASE NUMBER',14)
FORMAT(15,F5.ltF10.ltF5.ltFl0.2)
FORMAT('VEHICLE #',I5,5X,'MASS:•,F7.1,• KG',5X,'MAX HP:',F5.1,
FORMATUX,'FO =',F7.1,' N',5X,'F2 =',F7.4,» KG/M',10X,'(CALC)')
FORMAT(3X,'AHP =',F5.1,« HP',4*,'55-45 CD TIME:»,F6.2»' S',4X,'(DAT
FORMAT(3X,'RRC =',F8.5,4X,'CDA  =',F6.3,» M»»2»»10X,'(CALO')
FORMAT(I5,F5.1,F10.1,F10.4,E12.4)
FORMAT(3X,»AO =»,F7.4,' MI/HR-S',3X,•A2 =»,E12.V,' HR/Ml-S',5X,'(DA
FORMAT(I5.F5.1,F10.1»F10.5,F10.3)
FORMAT(3Xt'RRC = • ,F8.5,4X,'CDA  =»,F6.3,» M»»2',1IX,•(DATA)«)
FORMAT(3X,«(SCALING FACTOR',F6.4,')«)
FORMAT('ENERGY DEMANDS:  TOTAL:•,E12.4,3X,'TIRE:',E12.4,3X,»AERO:',
            ,E12.4,3X,'POSK:',E12.4)
            (TOTAL:»»E12.4,3X,'AT ENGINE, W/ 90* EFF CVT)')
FORMAT(16X,
FORMATU7X,
FORMATU7X,
FORMAT(17X,
FORMAT(17X,
FORMAT(17X,
FORMAT(3X,«
RETURN
END
                  DISTANCE:»,8X,F7.1,» M»,bX,F5.2,« MI')
                  TIME IN MODE:' , OR =  IDLE FC RATE)
343     C
344           IFUEL = (IFUEL*IH)/3600.
345           IF  (IFUEL .LT. IFLOW)   IFUEL = IFLOW
346           GO  TO 204
347       203 CONTINUE
348           IFtjEL = 
-------