SDSB   79-10
                           Technical Report


                             •-February 1979
                 Computer Simulation of Tire  Slip On
                   A Clayton Twin Roll Dynamometer
                                  by

                              John Yurko
                                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
technical information and to inform the public  of technical devel-
opments 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 and Waste  Management
                U.S.  Environmental Protection Agency

-------
                                -1-

Abstract

     Due to  the  occurrence of  tire  slip on  a  Clayton  twin roll
dynamometer, there  is a  difference between the velocities  of
the  front  and rear rolls  of  the  dynamometer.  This  slip  can  be
modeled  by the method  described in the following report.   The
results of this theoretical modeling show that,  over the LA-4 and
the HWFET driving  schedules the velocity of the rear  roll (which is
currently used to  determine the vehicle  speed)  exceeds  the velocity
of the  front roll (which  determines  the power abosrbed)  by  an
average of approximately 1%.  From this  difference  in velocities of
the  two  rolls, a computation of the  total energy effect,  over
transient driving  cycles,  can be obtained.  Approximately  a 2%  to
5% increase  in total energy  dissipated  over  the city  (LA-4)  or
highway driving schedules   is predicted from the  use of  the front
roll velocity to  determine the  vehicle  speed,  as  compared  to the
currently used velocity of  the rear roll.

-------
                               -2—
I.   Introduction

     When a  vehicle  is  operated  on a  twin  roll dynamometer  the
velocities of the front and  rear rolls differ.  This  difference,  or
tire slip, is  important  because  the tractive load imposed on  the
vehicle is dependent  on the velocity and acceleration  of  the  front
roll,  while  the velocity  of  the rear  roll  is  used  to  determine
the  vehicle  speed.   This report discusses a theoretical model
of  this  tire  slip  phenomenon  and investigates its effect  on
the EPA measurements  of fuel economy and emissions.

II.  Discussion

     Historically,  the  term tire slip  has  been  used to  describe
several physical  phenomena  and hence  is  somewhat  ambiguous.
For  the purpose of  this report, which only considers  the twin
roll  dynamometer,  slip  is defined  as  the difference between
the  velocities  of  the  front  and rear  rolls of   the  dynamometer.

That is:

     Slip = V - U                                  (1)

     V    = velocity  of the  rear roll

     U    = velocity  of the  front roll

     This slip has  been observed  in  an EPA tire  test program
which  showed that at  steady-state 50  MPH, the front  roll  consis-
tently travels at a  slower speed  than the rear roll  (1).   This was
confirmed for several different tires at  five  different power
absorber settings.

     This phenomenon  is significant,  since  the tractive load
imposed on the  vehicle is dependent on the  velocity  and  acceler-
ation of the  front dynamometer roll, however, the  vehicle  is  driven
over a speed  versus time  schedule based on the velocity of the rear
roll.   Consequently,   the  difference  in  the roll velocities may
cause  the vehicle  to  be  underloaded with respect to the  loading
which would  be imposed if the  same speed schedule were followed  on
the  road.   Under steady  state conditions,  tests have shown that
this under  loading  may be  more  than 1/3 HP (1).  (FY78  program
plans  include  an  investigation into  the  relationship between
the occurence of slip  on  the  twin roll  dynamometer and the  actual
road experience of the  vehicle.)

     It is also important  to consider the effect of tire slip under
transient conditions.   Since it is difficult to monitor slip  during
a  transient  cycle, a  theoretical model was  proposed  as  the most
appropriate method of  investigation.  The subsequent  subsections  of
the discussion consider the development  of the tire slip model, a

-------
                                -3-
comparison  of  some  empirical  slip results  with  the predicted
results,  a  computation of the energy  effects,  and an  estimate  of
the effect on fuel economy and emissions testing.

     A.  Theoretical Model

     In order to theoretically model the tire slip that occurs on a
twin roll dynamometer, a model equation which couples the slip to a
dynamometer  or  cycle  parameter must  be  chosen.   Steady-state  tests
show  that the  difference  between front  and rear roll  velocities
increased as the  dynamometer  power setting increased.  Conse-
quently,  it  is  assumed that  tire slip  is  directly proportional  to
the force across the tire roll interface.

     Slip = sF                                      (2)

or

     V - U = sF                                     (3)

where

     s = coefficient of slip

     F = force at the tire/roll  interface

     The  basic  model  equation  2, assumes  that  tire/roll slip  is
caused by  a  tire deformation resulting from the  tangential  forces
that exist at the tire/front-roll interface.

     Slip, during  transient driving,  can  be  calculated from  the
model equation (3) if the force  across  the interface  is known.  The
force  at  the tire/front  roll interface  includes the  dynamometer
force and the inertial force.

     F = Dyno Force + Inertia Force                 (4)

     The  dynamometer  force  is  assumed to  be  proportional to  the
square of the vehicle velocity,  that is:
                     2
     Dyno Force = b u                           (5)

where

     b  -  proportionality  constant  dependent  on the  total  dyna-
         mometer absorbed power.

     This neglects  the  constant  characteristic of the dynamometer
bearing force,  however,  this is  not critical since  the  velocities
squared characteristics of the power absorber dominate  over most  of
the speed  regions.    Hence  minor variations  in  the  assumed  dyna-

-------
                                -4-

mometer curve shape should not significantly affect the accuracy of
the model results.

     The inertia force is the mass effect of the vehicle.

     inertia force = m a

                   = m (du/dT)                      (6)

where :

     m = dynamometer simulated inertia of the vehicle.

     a = du/dt = acceleration

     Substituting equations 5 and 6 into equation 4.

     F = b u2 + m (du/dT)                       (7)

using the model equation 3 this becomes:

     v - u = s(b u2 + m (du/dT))                (8)

     A linear approximation is used to calculate du/dt  for each  one
second interval of the given  speed versus  time  schedule.   That  is;

     du/dt = u2 - Uj                        (9)

where :

     T = 1 second.

     This acceleration  is assumed  to be  the  acceleration at  the
midpoint of  the  time interval.  Therefore  the  midpoint velocities
are used in the model equation:
     v = (v2 + Vj)/2                       (10)

    -u = (u2 + Uj)/2                       (11)

Substituting 5, 6 and 7 equation 8 becomes:
(v2 + Vl)/2 - (u2 + Uj)/2 = s(b((u2 + u^/2)   + m(u2  - Uj))     (12)

Expanding and  regrouping terms  in order  to use  the  quadratic
formula:
         2
(sb/2) u2  + (sbu.  + 2 sm + I)u2 + (-(v2 + Vj ) + (sbu./2  + 2  sm +  l)u,)    (13)

-------
                                  -5-
Using the quadratic formula to solve for  u~

     u2 = (-B + (B2 - r AC)1/2)/2A                (14)

Where:

     A = sb/2                                     (15)

     B = sbuj + 2 sm + 1                          (16)

     C = (sbUj/2 - 2 sm + l)ux - (vj +  v2)         (17)

     In order to utilize this model  for  computer  simulation u, and
vj are  initialized to zero,  since  the  cycle  begins with an  idle.
vo is  entered  as  the  next speed  point from  the given  driving
schedule (see Appendix A for computer program  "slip").  Each time a
    U2  is  calculated,  v2  and u2 are replaced by v^ and uj respec-
tively  and the next  v2 is  entered.  This  iterative process is
continued for the entire driving schedule.

B.   Theoretical vs.  Experimental Results

     The model  was  used  to  compute the  dynamometer  slip,  using
input parameters  estimated  to  reflect  the vehicle parameters of a
1978 Mercury Zephyr.   The Zephyr was chosen  since this vehicle was
used in a recent test program on a Clayton dynamometer in which the
difference between the  front and rear roll velocities was recorded
over  an LA-4 driving  cycle.    Therefore,  this  allows a  direct
comparison between the  theoretical tire slip model and experimen-
tally measured dynamometer slip.

     A  mean  coefficient  of  slip was  calculated  from the data
acquired from steady-state tire tests (1).

     s = 0.0008  sec/kg

     This  program also requires the vehicle  frontal area  as an
input parameter  to calculate the appropriate  force coefficient.  In
this case,  the  actual  dynamometer power  absorber  setting  used to
test the Mercury Zephyr, was used  to back  calculate for the  appro-
priate frontal area,  according  to the equation given in the Federal
Register for non-fastback  vehicles  (2).

     HP at 50 mph = 0.50 (Frontal Area)            (18)

Where :

     Actual Dynamometer HP at 50 mph  = 9.7  HP

therefore;

-------
                               -6-
     Frontal Area = 19.4 square feet.

     The dynamometer power at  50  mph is  then used to calculate the
proportionality constant (b) using equation 5.
                    2
     Dyno Force = bv

     Dyno Power = bv

therefore

     b = (Dyno power @ 50 mph)/(50 mph)

     50 mph = 22.35 m/s

     745.7 watts = 1 hp

     b = 0.648 kg/m

     The input  parameter  for the vehicle inertia  (m)  was  the same
as  that  used  in  the  dynamometer tests  with the Mercury  Zephyr.

     m = 1600 kg

     With these parameters  the theoretical  slip  was  then  computed
for the entire LA-4 and Highway Driving cycles.

     The results  of the  tire  slip  modeling  showed  that the rear
roll travelled, on  the average,  0.09%  faster than the  front roll
over the LA-4 driving schedule.   Over the highway cycle  the rear
roll averaged 1.4% greater speed than the front roll.

     These results  can be compared  to  the  results of  a test con-
ducted on a  Clayton twin roll dynamometer with  the Mercury Zephyr
(Figure 1 shows both the  actual  and  theoretical  slip  for the first
130 seconds of the LA-4 cycle).

     There  is a  very  close comparison  between  the  theoretically
modeled slip  and  the experimentally  measured  slip.   The regions of
positive slip  correspond  to  accelerations or steady-state  modes in
the driving schedule, while the regions of negative slip correspond
to  decelerations.   A  minor  discrepancy  in  the  magnitudes  of  the
modeled slip and the actual slip is mainly due to the use of a mean
value  for  the slip coefficient,  which  appears to be  slightly  low
for this particular  vehicle.   There are  a  few minor  discrepancies
in  peak  locations which are probably caused  by  transmission shift
points.  Also,  a  very  slight  shift  in the data  is  observed which
may be due  to a  slight  miscalibration of the  chart recorder.

     Since the purpose of this report is to estimate  the total fuel
economy and emissions  effects,  these slight  inaccuracies are not a
major  concern.    During  rapid  decelerations  an  actual  mechanical

-------
                                  -7-
slippage has  also been observed.  This phenomenon may  result  in a
significant discrepancy  when comparing actual  slip  to  theoritical
slip.   However,  since mechanical slip (actual  sliding  of  the  roll
surface  across the  tire surface) primarily  occurs  during  closed
throttle modes  it will not have  a significant effect  on  fuel
economy results.   Thus the  theoretical model  may  be  used to accur-
ately predict  the  effect  of slip on  EPA  fuel  economy and emissions
testing.
C.   Computation of Energy Effects

     In order to estimate the energy effect of the tire slip over a
transient  cycle,  a  second  computer  program  is  developed.   This
model utilizes  the  total  road load equation of the  vehicle on the
dynamometer (see appendix B for program "LA4FORCE").

            f = rr + b v2 + m(dv/dT)               (19)

Where:

           rr = rolling resistance (tire force).
            o
         b v  = dynamometer force.

     m(dv/dT) = inertial force

     The input  velocity  (v)  can be either the front  roll  velocity
or the rear roll velocity, for any given driving  cycle.

     The power dissipated (P) is then calculated

     P = F * V                                     (20)

     Finally  the power  is  summed  each  second  over the entire
driving schedule to yield the total energy dissipated.  However,  if
the power is negative the vehicle is assumed  to be braking  and  this
is  summed  as  a separate  quantity referred  to as "Brake  Energy".

     This program may  then be  used  to calculate  the  total energy
that is  assumed to  be  dissipated, over a  driving  cycle, using the
rear roll velocity.   Simultaneously,  it  may be  used  to calculate
the actual energy that  the vehicle has dissipated, by inputting the
front roll  velocity, that  has  been  calculated  by  the tire  slip
model, for the same  cycle.

     This program was  utilized  in  conjunction with   the tire  slip
program in order to  calculate  the  energy  effects  of  this tire  slip
phenomenon.    Once  again,  the  input parameters were  those of  the
Mercury Zephyr.   In addition,  a rolling  resistance  coefficient  is

-------
                                 -8-

required in order to calculate  the  tire  energy  dissipated.   A mean
rolling resistance coefficient was used,  typical of that particular
tire type and size on the Mercury Zephyr(S).

     rr     = 0.010
       coef

     The total energy dissipated over the LA-4 cycle was calculated
using the veloctiy  of  the rear roll, and  then  the computation was
repeated using the velocity of the front  roll (as computed from the
tire slip program).   The results showed that 4.3% more energy was
calculated using the velocity of  the  rear  roll  as  compared  to that
calculated from velocity of the front roll (see Table 1).  The same
procedure was  repeated  using  the  highway  driving  cycle and  the
results showed a 4.1% greater total  energy absorbed  by  the  dynamo-
meter was calculated  using the velocity  of  the rear roll  as com-
pared to using that of the front roll.

III. Conclusions

     There is a difference in the velocities of the  front and rear
rolls of  a  Clayton twin  roll  dynamometer  due to  tire  slip.   This
slip is directly proportional to  the  tangential  force at the front
roll/tire interface.   It  accounts  for  approximately a  1%  greater
velocity of rear roll than the front roll,  during transient  driving
cycles.  This difference in velocities can be modeled approximately
by  the method  as described in  this report.   A  computation  of the
total energy dissipated, by a vehicle over an LA-4 or HWFET  driving
schedule indicates that approximately a  2% to 5% increase in energy
will result  from using the front roll velocity to determine  vehicle
speed rather than the currently used rear roll velocity.

IV.  Recommendation

     It is important to determine if this occurrence if  slip on the
twin roll has  a  significant  effect  on  fuel economy or emissions
testing.   It  is most  important  to  compare  this phenomenon  to the
actual road experience  of the vehicle.   If  such  studies show that
slip is an undesirable factor during vehicle  testing, some possible
alternatives are:    to  couple the  front and rear rolls  of  the
dynamometer; use the front roll to determine  vehicle speed,  or look
to the possible use of flatbed dynamometers.

-------
                                -9-
                         References
1.   Richard Burgeson, Myriam  Torres,  "Tire Slip  on  the Clayton
     Dynamometer,"  EPA Technical  Support  Report, LDTP  78-02, March
     1978.
2.   "Control of Air  Pollution from New  Motor Vehicle Engines,"
     Federal  Register.  Vol.  42, No.  176,  September  12,  1977.


3.   Richard N.  Burgeson, "Clayton Dynamometer-to-Road Tire Rolling
     Resistance Relationship," Technical  Report,  LDTP  78-09.


4.   Glenn D. Thompson,  "Investigation of  the  Requested Alternate
     Dynamometer Power Absorption for  the Ford Mercury Marquis,"
     EPA Technical Report, LDTP 78-06.

-------
                                       Figure  1
                                                        - - - - Actual slip
                                                        	 Theoretical slip
CO*
0.0
20.0
40.0
60.0
TIME
80.0
(SEC)
100.0
120.0
140.0

-------
Table 1

LA-4 Cycle
(1) Rear Roll
(2) Front Roll
(1-2) Difference
.-2)/2 % Diff.
Highway Cycle
(1) Rear Roll
(2) Front Roll
(1-2) Difference
L-2)/2 % Diff.

Total Distance
(meters)
11987.5
11877.4
110.1
0.93%

16496.2
16254.3
242.2
1.47%

Avg. Velocity
(m/sec)
8.74
8.65
0.09
1.00%

21.5
21.2
0.3
1.42%
Total Energy
(Joules)
X 10
504.0
483.0
21.0
4.3%

807.0
775.0
32.0
4.10%
Tire Energy
(Joules)
X 10
153.0
151.0
2.0
1.30%

210.0
207.0
3.0
1.45%
Aerodynamic
Energy (Joules
X 10
170.0
163.0
7.0
4.30%

552.0
527.0
25.0
4.70%
Brake Energy
i (Joules)
X 10
-181.0
-169.0
- 12.0
7.10%

-448.0
-400.0
48.0
12.00%












-------
                                    Appendix A

                                       A-l
C
r
r
C
r
r.
C
C
C
C
c
r
C
C
C
r
r
c
C
c
10'-

600
c
r:
r-
C
c
C
C
      pPOf.RAM JO CAl'/iLATF  THF TOTAL F OPCEC> » PnwE-c , AND ENERGY  QISSIPA.TEO
      n\/FP  THF LA4 S-!-"En
      V^.  TIMF SCHEDULE

      ^'DITTEN BY JOH-.i YUPKQ

      THF  FOLLOWING •'"•', T a  MUST RF SI IP'--L UTi'i :  VFHICI..E  MASS  (KC-)«  FRONTAL
       (FT**?) , POLL I
      COEFFICIENT  (D^
FOP
                                    ftY r-ir>T  f1>>TC^ T-iE TEP'^S  IN'  THIS REPOPTt
      FAREA ....... FP?V--;TAL
      PAUCOF ...... CO^'FTCTFA'T ^F AEP-OYMAf-'IC  HRAr,
                   VE-jlCi. F  MASS (w)
           EF ...... POl.i I\'G  RrST^TAMCF COEFFICIENT
         r) I MFNS I OKJ V ( 1 0 n 0 0 ) . T T ( 1 0 n o 0 )
            H (5»100) VMJ.sS.F.A
               (?F in. I.? !0.3)
             A T ( ] X , ?F 1 0 . 1 » F 1 0 . 1 , / / )
        FRONTAi  APFA TO  A  n
"FDFRAL REG
FOR
              = n.sn*FA".- A
         '1 = 0.0
       =  PwATT"/ (??

     IN VELOClTr-,TIME.  AMD
                                               LOAD COEFFICIENT  USING THF
                                              DV/DT
  99
         •10  ?o 1 = 1.1372
         -VFAD (5»?On.FMD = c:'9Q) TT ( I ) . V( I )
  20
FORMAT
                   . 1 )
  30 ">    FORMAT ( • 1 • . T6« »T!MF. i . Tl?. • VFLO-TTY* *T22.»ACCELFR.'«T33»
      1  i TTRFFORCF1 . T44. » AFROFnRr;r i . T55 , • TMEPTIM..' «T^7.
      ?  'TrTAL'»T7°, if>0'Jc.R« . //I X »T5, i (S~C) • «T12. • (M/SFC)
      1  • (HV/DT) » .T33» ' f^EMjOHS) ' . T44. • t ME '--'TOMS) •
               .T67.tFO»CEi.T7P,ifwATTS)«.//)

-------
r.
c

C
r
r
C
r
r
c
c
r,
C
  40
  70 n
CALCULATE  OOLLIMG RESISTANCE (T*»EF> ,  AERODYNAMIC
IMEPTIAL FORCE (p
JMO TOTAL
TTRFF =
^FP.OE = 0.0
fMEPF =0.0

TJPFNG =  0.0
FMFPGY =  n.o
;.1P Q 1 = ] ,M
CONVERT VFLOflrv  FROM MP^ TO M-Tt'RS/SEC

V ( T )=.447*\/ (T)
.1=1-1
If (T.FQ. 1 )GO  TO 5
     = V ( ! )-V ( 0
   TO 6
CALCULATE POWF°  AMH FNFO-Y OIS-TPATEO

POWFP = FHPCF*-" 'EftM
JJPFP = TTPF.F*l'"EAM
T i PFNG =
        AFPOF  =  A.FPOF + .'
         ^aLC'ILATF  TMF TOTAL
                             E TP4VELt"0  RY  THE  ROLL
IF(FOPCF.LF.n.n»Go  TO 7
FMFDGY = FMFPGV+PPWFD
GO TO R

"PPAKE" REPRESENTS  \/EHjri E
r-'PlTE (6*400) I • v ( I ) .nv^T, TJOFF. •• c"POF . r'lrj^P
F OPM A T(io«.4x,Tu»Fin.?,FT0.2»FH.?.F12.?.Fll.2,Fin.2.Fl3.2)
    F (6,FOO) TJ-Jl- NG.FIMEP. AEROE.f.^AKt .E'-.'FPGY
        »0» »^X« 'TlPE  EMEPr'Y = ' . F 1 0 . 3./4X , ' TNERTI AL EMERGY =  ',
FH .3»/4X« 'AFRni-YMAMir FMFPAY - ' *El?.3«/4v. »PPAKE EMERGY =  '»
r1 ?.3*/4X, iTOTM.  FiVFPGY = ' « E 1 '-• . 3 )
-•'PITE (6.700)0
COP.WAT(FX. • 01 ST^Nrp  = '.-10.1)
STOP

-------
                                 Appendix B

                                     B-l
C       POnGBAM TO C*LC!'LATE  THE \/ELOCI T Y (II?.)  OF  THE FRONT POLL  FOR  A  TWIN
C   '    P'UL DYNAMOMETER
C       GIVEN THE ^EAP  -'vLL  VELOCITY (V2) :   THAT  IS THE VELOCITY  FROM THE
C       l_a<.  SCHEDULE
C       W'-'TTTEN PY JOHN  vURKO
C
C       THE  FOLLOWING DATA MUST RF SUPPi TED:   VEHICLE MASS (VM)  IN KG*
C       FRONTAL A^EA  (F.'iPEA)
C       IM  FT**2, POLLl"-'^ RESISTANCE COEFFICIENT  (RP). VELOCITY  (V?)  IN
C       A^-Jn  SLIP
C       COFFFICIFNT  (S)
c
r"
o
c       CALCULATE COEFFTCIFNTS e AND M -OR  ROAD  LOAO EQUATION:
C       (C"rft + R»V**?t-^J->n.//DT)
r       C«"'MSTANT "A" REPRESENTS POLLING RESISTANCE AND HOES NOT
r       C'iNTPIRUTF TO S! ?P
c       THERFORE THIS TC VM T«; IGMPRED
r
       OlfFNSlON V2(lOOnO) .IT (10000)
       RE -'\n ( 5 , 1 0 0 ) \/M , F A -' - A , pp , s
       DO ?n 1=1.132
       DEin(R,?00»EMD=9^g) IT(T) .V'(T)
  200  FO.'vAT(I4,F6. 1 )
    20  C"J!-.iTlNllE
  990  \|=T-1

  10^  FO-'^'AT(2F10.1 tFlrt.3.E14.3)
       WRITE (ft » 100) V/M»P"At-E/i ,RR.S
       vi=n.o
       u l=o.o
c
C       CALCULATE" REAR PnLL  VELOCTTY
C
       no  9  1=1, N
c
C       CONVERT V2 FROM '-••PH  TO  M
C
                        *1 .n
      <-»C= (R*S*U 1/2.0-2. 0*S of* "I .p) *U1-V ( T) -VI
      l(p= (_QP + SQRT (GR?-4.n«-OA*OC) ) / (•-•.0*0 A)
      IP" (UP. RE. 0.0)  GO TO ?
      U?=0.0
      Vl=V2(I)
r*

-------
                               e-2
c      c-n CMLATF  TOTAL  USTANCF TRAVELS ON EACH ROLL
r
      \/?U )=V2(T) /
      TSl. TPrV?(I)-U?
      AQ TO 21
      W«TTP(6,300) IT (I) ,U?,V2(I) .TSL IP
  30 r FORMAT (J4,F6.1 .Fl •••.] ,FIO. D
   ?i w»iTf:(A.,7nn) IT (H ,TSLTP
  700 FORMAT ( I5.F5. 1 )
C
C      THIS PROGPAM  OUTPUTS THE rRQNT wOLL V^LrrHY IN (>'PH)  IN /
C      FOPMflT  THAT
C      CAM PF  DIPFCTLY  r
-------