NCAT

 Report  15-­‐02  

LITERATURE  REVIEW:  THE  IMPACT  OF  

PAVEMENT  ROUGHNESS  ON  VEHICLE  
OPERATING  COSTS  
 
 
 
 
 
 

Submitted  by:  
Dr.  Mary  M.  Robbins  
Dr.  Nam  Tran,  P.E.  
 

May  2015  
 LITERATURE  REVIEW:  THE  EFFECT  OF  PAVEMENT  ROUGHNESS  ON  VEHICLE  OPERATING  COSTS  

 
    By  
 
 
Mary  M.  Robbins,  Ph.D.  
Nam  Tran,  Ph.D.  
 
 
 
National  Center  for  Asphalt  Technology  
Auburn  University,  Auburn,  Alabama  
 
 
 
Sponsored  by:  
National  Asphalt  Pavement  Association  
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
May  2015  
Robbins  and  Tran      

ACKNOWLEDGEMENTS    

The   authors   wish   to   thank   the   National   Asphalt   Pavement   Association   (NAPA)   and   the   State  
Asphalt  Pavement  Associations  for  sponsoring  this  research  as  part  of  the  Determining  Service  
Life  Based  on  Comparable  IRI  research  project.    

DISCLAIMER  

The  contents  of  this  report  reflect  the  views  of  the  authors  who  are  responsible  for  the  facts  
and  accuracy  of  the  data  presented  herein.  The  contents  do  not  necessarily  reflect  the  official  
views   or   policies   of   the   sponsored   agency,   the   National   Center   for   Asphalt   Technology,   or  
Auburn   University.   This   report   does   not   constitute   a   standard,   specification,   or   regulation.  
Comments  contained  in  this  report  related  to  specific  testing  equipment  and  materials  should  
not  be  considered  an  endorsement  of  any  commercial  product  or  service;  no  such  endorsement  
is  intended  or  implied.  

iii  
Robbins  and  Tran      

Table  of  Contents  

TABLE  OF  CONTENTS  .............................................................................................................  IV  
1.   INTRODUCTION  .................................................................................................................  1  
1.1   Background  .......................................................................................................................  1  
1.2   Pavement  Roughness  ........................................................................................................  1  
2.   EFFECT  OF  PAVEMENT  ROUGHNESS  ON  VEHICLE  OPERATING  COSTS  .................................  3  
2.1   Impact  of  Pavement  Roughness  on  Fuel  Consumption  Costs  ...........................................  4  
2.1.1   Recent  U.S.  Studies  ...................................................................................................  7  
2.2   Impact  of  Pavement  Roughness  on  Tire  Wear  Costs  ........................................................  8  
2.3   Impact  of  Pavement  Roughness  on  Maintenance  and  Repair  Costs  ...............................  11  
2.4   Impact  of  Roughness  on  Oil  Consumption  Costs  .............................................................  13  
2.5   Impact  of  Pavement  Roughness  on  Depreciation  Costs  ..................................................  14  
3.   SUMMARY  .......................................................................................................................  15  
REFERENCES  ..........................................................................................................................  16  
 

iv  
Robbins  and  Tran      

1.   INTRODUCTION  
1.1   Background  
Ride   quality   as   a   measure   of   pavement   roughness   plays   an   important   role   in   users’   level   of  
comfort   and   operating   costs.   As   Swanlund   points   out   in   Public   Roads   (1),   smooth   roads   cost  
transportation  agencies  less  over  the  life  of  the  pavement  and  result  in  decreased  highway  user  
operating  costs,  delayed  costs,  decreased  fuel  consumption  and  decreased  maintenance  costs.  
Thus,  “not  only  do  our  customers  want  smooth  roads  for  comfort,  smooth  roads  cost  less  for  
both  the  owner/agency  and  the  user”  (1).    
 
This   notion   of   increased   user   and   agency   costs   was   echoed   almost   a   decade   later   by   Biehler,  
then   president   of   the   American   Association   of   State   and   Highway   Transportation   Officials  
(AASHTO).  He  stated,  “The  American  public  pays  for  poor  road  conditions  twice—first  through  
additional   vehicle   operating   costs   and   then   in   higher   repair   and   reconstruction   costs”   (2).   He  
went   on   to   elaborate,   “Driving   on   rough   roads   accelerates   vehicle   depreciation,   reduces   fuel  
efficiency,  and  damages  tires  and  suspension”  (2).  
 
More   recently,   The   Road   Information   Program   (TRIP)   used   the   international   roughness   index  
(IRI)   data   reported   in   2011   by   the   Federal   Highway   Administration   (FHWA)   to   assess   the  
roughness   of   the   nation’s   roadways   and   its   potential   effects   on   highway   user   costs   (3).   TRIP  
concluded  that  27%  of  the  “nation’s  major  urban  roads”  (interstates,  freeways,  and  other  major  
routes  in  urban  areas)  were  in  poor  condition  with  IRI  greater  than  170  in/mile,  and  42%  were  
in   mediocre   or   fair   condition   with   IRI   between   120   and   170   in/mile   (3).   TRIP   also   found   that  
“driving   on   roads   in   need   of   repair   costs   the   average   driver   $377   annually   in   extra   vehicle  
operating   costs,”   with   additional   vehicle   operating   costs   ranging   between   $178   and   $832  
annually  for  urban  areas  with  populations  greater  than  500,000.    
 
It   is   evident   that   rough   pavements   result   in   increased   users’   costs   through   vehicle   operating  
costs.   This   report   is   prepared   based   on   a   review   of   literature   to   further   discuss   the   extent   to  
which  pavement  roughness  affects  the  various  components  of  vehicle  operating  costs.    
 
1.2   Pavement  Roughness  
According   to   the   American   Society   of   Testing   and   Materials   (ASTM   E867),   pavement   roughness  
is   defined   as   “the   deviations   of   a   pavement   surface   from   a   true   planar   surface   with  
characteristic   dimensions   that   affect   vehicle   dynamics,   ride   quality,   dynamic   loads,   and  
drainage,  for  longitudinal  profile,  transverse  profile,  and  cross  slope”  (4).  Pavement  roughness  
measurement   in   terms   of   serviceability   was   first   introduced   by   the   American   Association   of  
State   Highway   Officials   (AASHO)   at   the   completion   of   the   AASHO   Road   Test   in   the   late   1950s  
(5).   In   this   measurement,   the   serviceability   of   a   pavement   can   be   expressed   as   the   Present  
Serviceability  Rating  (PSR),  which  is  the  mean  roughness  rating  on  a  scale  from  0  to  5  given  by  a  
panel  of  passengers  in  a  vehicle.  The  relationship  between  the  panel-­‐rated  PSR  and  non-­‐panel  
pavement  performance  measurements  is  represented  by  a  mathematical  model  known  as  the  
Present  Serviceability  Index  (PSI).  After  the  AASHO  Road  Test,  several  studies  were  conducted  
to   evaluate   various   non-­‐panel   measurement   systems,   such   as   roughometers,   profilometers,  
and   ride   meters,   in   order   to   replicate   the   results.   In   the   early   1980s,   an   experiment   was  

1  
Robbins  and  Tran      

commissioned   by   the   World   Bank   in   Brazil   to   determine   how   to   best   relate   profile   data   to  
pavement  quality.  The  International  Road  Roughness  Experiment  (IRRE)  was  conducted  in  1982  
by  research  teams  from  Brazil,  England,  France,  the  United  States,  and  Belgium  (6).  Researchers  
found   that   surface   profile   was   the   best   common   ground   between   all   the   different   technologies  
studied.  As  a  result  of  this  study,  a  roughness  measurement  standard,  known  as  International  
Roughness  Index  (IRI),  was  established  (7).    
 
IRI   is   an   objective   means   to   assess   pavement   roughness   “based   on   the   response   of   a   generic  
motor  vehicle  to  the  roughness  of  the  road  surface”  (8).  This  is  done  through  the  quarter-­‐car  
model,  a  mathematical  model  that  simulates  how  a  reference  wheel  traveling  at  50  mph  would  
respond   to   the   deviations   in   the   pavement   surface   (i.e.   roughness)   along   the   length   of   the  
pavement   (8).   By   applying   the   quarter-­‐car   model   to   the   measured   profile,   IRI   is   computed   as  
the  cumulative  suspension  displacement  per  unit  of  distance  traveled  and  expressed  as  m/km  
or  in/mile  (9).  In  other  words,  IRI  is  a  specified  mathematical  transform  of  a  pavement  surface  
profile  (10).  IRI  can  be  modeled  on  pavement  surface  profiles  determined  by  inertial  profilers,  
which   measure   the   distance   between   a   reference   point   on   the   profiler   and   the   pavement  
surface   using   lasers.   The   distance   is   then   adjusted   to   account   for   vertical   movement   of   the  
vehicle  captured  by  accelerometers  to  determine  the  true  relative  profile  of  which  the  quarter  
car  model  is  then  applied  to  compute  IRI.    
 
While  other  indices  (ride  number,  profile  index,  etc.)  can  be  calculated  from  a  surface  profile,  
“IRI   summarizes   the   roughness   qualities   that   impact   vehicle   response   and   relates   most   to  
overall  vehicle  operating  cost,  overall  ride  quality,  dynamic  wheel  loads  (that  is  damage  to  the  
road  from  heavy  trucks  and  braking  and  cornering  safety  limits  available  to  passenger  cars),  and  
overall  surface  condition”  (10).  As  a  result,  IRI  ranges  differ  from  one  class  of  roadway  to  the  
next.   The   ranges   of   roughness,   expressed   as   IRI,   expected   on   various   roadways,   such   as  
airfields,   new   and   aged   paved   roadways,   maintained,   unpaved   roadways   and   damaged  
roadways  are  shown  in  Figure  1.  For  paved  roads  (regardless  of  class),  the  expected  IRI  range  is  
very  broad  with  IRI  from  less  than  100  in/mile  on  interstate  pavements  to  as  high  as  700  in/mile  
on  damaged  pavements  (11).    
 

2  
Robbins  and  Tran      

 
Figure  1  IRI  Roughness  Scale  (11),  after  (12)  
 
Although   Figure   1   displays   IRI   values   that   may   be   experienced   on   different   types   of   paved  
roadways,   the   IRI   ranges   that   U.S.   agencies   consider   for   categorizing   roadways   in   their   network  
are  much  tighter.  Table  1  below  shows  the  ranges  of  IRI  considered  by  FHWA  and  Washington  
State   Department   of   Transportation   (13).   FHWA   considers   pavements   that   have   a   roughness  
measured  by  IRI  greater  than  or  equal  to  170  in/mile  as  unacceptable,  while  Washington  State  
Department   of   Transportation   (DOT)   considers   pavements   unacceptable   when   they   have  
reached  an  IRI  of  221  in/mile  (13).  
             
Table  1  IRI  Categories  of  WSDOT  and  FHWA  (13)  
IRI  Categories  of  Roughness  (in/mile)  
  WSDOT   FHWA  
Very  Good     ≤  95   ≤  60  
Good     96  -­‐  170   61  -­‐  95  
Fair     171  -­‐  220   96  -­‐  120  
Poor     221  -­‐  320   121  -­‐  170  
Very  Poor     >  320   >  170  
 
2.   EFFECT  OF  PAVEMENT  ROUGHNESS  ON  VEHICLE  OPERATING  COSTS  
In   Volume   7   of   the   World   Bank’s   manual   for   the   Highway   Development   and   Management  
(HDM-­‐4)  model,  vehicle  operating  costs  (VOC)  is  defined  as  the  “total  cost  of  road  transport”  
and   it   takes   into   account   fuel   consumption,   tire   consumption,   oil   and   lubricant   consumption,  

3  
Robbins  and  Tran      

parts   consumption,   labor   hours,   depreciation,   interest,   and   overheads   (14).   It   has   been  
suggested  that  VOC  is  influenced  by  pavement  condition,  pavement  type,  roadway  geometry,  
and  operating  speed  in  addition  to  vehicle  type  and  vehicle  technology.    
 
A   number   of   studies   have   been   completed   on   such   topics,   including   an   investigation   dating  
back  to  1877  on  rolling  resistance  on  common  roads  (15).  Much  of  the  groundwork  for  vehicle  
operating   costs   used   today   was   completed   by   Winfrey   in   1969   (16),   from   which   a  
comprehensive   set   of   running   cost   tables   was   developed.   Shortly   thereafter,   Claffey   completed  
a  comprehensive  study  of  VOC  in  the  U.S.  in  1971  (17).  Since  then,  numerous  studies  have  been  
completed   on   the   effect   of   pavement   condition   on   VOC   by   studying   the   effect   of   pavement  
roughness   on   various   components   of   VOC,   including   fuel   consumption,   tire   wear,   repair   and  
maintenance,   and   oil   consumption   costs.   The   following   sections   discuss   key   findings   of   these  
studies.   Some   of   the   studies   had   been   conducted   before   IRI   was   developed   in   the   1980s.  
Therefore,  where  the  conversion  is  known  or  is  indicated  by  the  authors  of  the  original  work,  an  
approximation   of   the   equivalent   IRI   is   provided   as   IRI   measurements   capture   the   roughness  
qualities  most  associated  with  ride  quality  and  vehicle  operating  costs  (10).  
 
2.1   Impact  of  Pavement  Roughness  on  Fuel  Consumption  Costs  
Studies   have   shown   increased   fuel   consumption   on   rough   pavements,   resulting   in   increased  
operating   costs.   The   first   major   studies   on   the   impact   of   pavement   roughness   on   fuel  
consumption   were   conducted   by   the   World   Bank   on   unpaved,   gravel,   or   earthen   roadway  
surfaces   in   developing   countries   to   develop   and   refine   the   widely-­‐used   Highway   Design   and  
Maintenance   (HDM)   model   (14,   18,   19).   Since   the   model   was   developed   based   on   roughness  
data   that   were   not   typically   seen   in   the   U.S.,   it   was   later   calibrated   to   U.S.   roadway   conditions.  
According  to  Chatti  and  Zaabar,  IRI  values  on  U.S.  roadways  typically  range  from  1  to  5  m/km  
(63.4  t0  317  in/mile)  (20).  
 
De  Weille  conducted  one  of  the  earliest  studies  in  1966.  He  used  data  from  an  earlier  U.S.  study  
to  reveal  that  operating  costs  were  higher  for  gravel  and  earthen  roadways  than  for  smoother  
paved  roadways  (21).  Fuel  consumption  was  reported  to  be  20%  higher  on  a  gravel  road  than  a  
paved  road  and  even  higher  (40%)  on  an  earthen  road  relative  to  a  paved  road  (21).    
 
The  World  Bank  conducted  the  first  four  primary  cost  studies  between  1970  and  1982  in  Kenya,  
the   Caribbean,   India,   and   Brazil.   Table   A2.2   of   a   2001   World   Bank   report   reveals   that   the   range  
of  IRI  used  for  the  fuel  consumption  experiments  in  these  countries  ranged  from  as  low  as  2.0  
m/km  (126.8  in/mile)  in  the  Caribbean  to  as  high  as  22.1  m/km  (1401.1  in/mile)  in  Kenya  (14).  
This   range   falls   into   the   category   of   poor   on   the   low   end   and   very   poor   on   the   high   end  
according   to   the   FHWA’s   criteria   (22).   The   range   included   roughness   data   for   various   surface  
types,   including   paved,   gravel,   and   earthen   roads.   According   to   Chester   and   Harrison   (18),  
unpaved  roadways  could  have  IRI  as  low  as  1.6  m/km  (101.4  in/mile)  if  in  excellent  condition.  
The  effect  of  roughness  was  built  into  both  the  fuel  and  speed  equations  for  the  Brazilian  model  
(18).   This   model   also   showed   a   much   more   pronounced   effect   of   roughness   than   those  
developed  from  the  Indian,  Caribbean,  and  Kenyan  studies  (18).  
 

4  
Robbins  and  Tran      

The  World  Bank  developed  the  HDM  model,  which  was  later  refined  primarily  using  data  from  
the   earlier   study   in   Brazil   (23)   to   produce   the   HDM-­‐III   model   (19).   The   effect   of   roadway  
characteristics   (including   roughness)   was   incorporated   into   the   HDM-­‐III   model.   For   use   in   the  
fuel  consumption  model  of  the  HDM-­‐III  model,  data  from  the  study  in  Brazil  were  analyzed  to  
determine   the   effect   of   roughness   on   rolling   resistance   (19).   In   that   study,   rolling   resistance   for  
light   and   heavy   vehicles   was   considered   for   four   road   sections,   including   paved   and   unpaved  
roadways   with   two   levels   of   roughness:   smooth   and   rough   (19).   Average   roughness   ranged  
between   29   (2.2   m/km,   or   141   in/mile)   for   the   smooth   paved   road   and   178   (13.7   m/km   or   868  
in/mile)   for   the   unpaved,   rough   road,   expressed   as   quarter   car   index   (QI)   or   counts/km,   such  
that  the  authors  approximated  IRI  by  QI/13,  where  IRI  is  in  m/km.  Comparing  these  values  with  
current   FHWA   IRI   categories   (see   Table   1),   it   is   evident   that   the   lowest   IRI   values   for   paved  
roads   in   the   Brazilian   study   would   be   considered   poor.   However,   it   should   be   noted   that   the  
authors   identify   a   roughness   range   for   paved   roads   as   25   to   125   QI   and   50   to   250   QI   for  
unpaved  roads  to  be  used  with  their  predictive  models  if  measurements  are  not  available.  For  
paved  roads,  very  rough  corresponds  to  a  roadway  exceeding  125  QI  (roughly  equivalent  to  an  
IRI   of   9.6   m/km   or   610   in/mile)   and   smooth   corresponds   to   a   QI   of   25   (1.9   m/km   or   122  
in/mile).  For  unpaved  roads,  50  QI  (3.8  m/km  or  244  in/mile)  is  considered  smooth  and  250  QI  
(19.2  m/km  or  1219  in/mile)  is  classified  as  very  rough.  The  key  findings  of  this  research  effort  
include  (19):  
• Rolling  resistance  generally  increased  with  road  roughness.    
• Two  empirical  models  were  developed  for  the  coefficient  of  rolling  resistance  for  light  
and   heavy   vehicles   considering   a   linear   relationship   between   roughness   (QI)   and   the  
coefficient   of   rolling   resistance.   These   interactions   are   important,   as   “increasing  
roughness  (from  very  smooth  paved  to  very  rough  unpaved)  causes  the  vehicle  power  
and  fuel  consumption  to  go  up  via  an  increase  in  the  rolling  resistance  coefficient.”    
• Rolling  resistance  was  also  a  function  of  vehicle  weight,  and  due  to  this  combination,  the  
effect  of  roughness  on  fuel  consumption  was  stronger  for  loaded  trucks  than  unloaded  
ones.    
• For  paved  roads  that  were  level-­‐tangent  with  roughness  within  the  range  of  50-­‐150  QI  
(3.8  m/km  or  241  in/mile),  the  effect  of  increasing  roughness  on  rolling  resistance  was  
offset   by   the   reduction   in   air   resistance   as   a   result   of   decreased   speeds   on   rough  
surfaces.    
• For   roughness   levels   typical   of   unpaved   roads   (beyond   125   QI),   the   effect   of   rolling  
resistance  on  fuel  consumption  for  level-­‐tangent  roadways  dominated  any  reduction  in  
air  resistance  due  to  reduced  speed  on  a  rougher  surface.    
 
Results   from   a   Wisconsin   study   revealed   a   nonlinear   increase   in   fuel   consumption   with   an  
increase  in  roughness  (24).  Fuel  consumption  in  passenger  cars  was  measured  with  a  fuel  meter  
on   five   test   sites   featuring   bituminous   pavements   ranging   in   roughness   from   a   serviceability  
index   (SI)   of   0.9   to   4.4.   The   result   was   a   3%   increase   in   fuel   consumption   between   the  
smoothest  (SI  =  4.4)  and  roughest  (SI  =  0.9)  pavements  tested.  Although  this  relationship  was  
found  to  be  non-­‐linear,  for  pavement  roughness  typical  of  Wisconsin  state  trunk  highways  (SI  
from   4.5   to   1.5),   the   increase   in   fuel   consumption   could   be   estimated   with   a   linear   function,  

5  
Robbins  and  Tran      

such  that  1.5%  more  fuel  would  be  consumed  on  a  pavement  with  an  SI  of  1.5  than  one  with  an  
SI  of  4.5.  
 
A   South   African   study   found   a   strong   correlation   between   roughness   and   rolling   resistance   and  
thus,  fuel  consumption  (25).   It   was   reported   that   asphalt   and   concrete   roads   have   lower   rolling  
resistance  than  roads  with  surface  treatments,  and  for  a  truck  traveling  at  80  km/h,  18%  more  
fuel  will  be  used  on  a  gravel  road  than  on  a  paved  road  in  good  condition.    
 
Contrary  to  previous  studies,  including  a  1979  report  by  the  same  author  (26),  Zaniewski  et  al.  
concluded   in   their   1982   report   that   “roughness   does   not   have   a   measureable   effect   on   real  
world  fuel  economy”  (27).  Zaniewski  et  al.  conducted  a  comprehensive  study  to  “determine  the  
operating  costs  and  fuel  consumption  of  motor  vehicles  as  a  function  of  vehicle  and  roadway  
characteristics”   as   well   as   the   effect   of   pavement   condition   and   type   on   performance  
parameters   and   to   develop   adjustment   factors   based   on   pavement   type   and   condition   (27).  
Vehicle   operating   costs   considered   included   fuel   consumption,   oil   consumption,   tire   wear,  
maintenance  and  repair,  and  user-­‐related  depreciation.  The  goal  of  this  research  study  was  to  
determine   the   quantities   of   consumption   as   a   function   of   roadway   characteristics   (grade,  
surface   type,   and   roughness);   thereby,   the   costs   could   be   obtained   by   simply   applying   unit  
prices  to  consumption.  Fuel  consumption  was  measured  for  passenger  vehicles,  loaded  single  
unit,   and   loaded   semi   (tractor   trailer)   trucks   on   test   sections   with   a   serviceability   index   ranging  
between  1.8  and  4.5.   Fuel  consumption  was  found  to  be  slightly  higher  on  the  unpaved  section  
than   the   paved   sections.   However,   it   was   reported   that   at   the   95%   significance   level,   no  
statistically   significant   differences   in   fuel   consumption   were   found   on   the   paved   sections.  
Therefore,  it  was  concluded  that  for  the  range  of  conditions  in  the  U.S.,  the  type  or  condition  of  
paved  roads  did  not  influence  fuel  consumption.  The  authors  acknowledged  that  these  findings  
conflict   with   previous   studies   by   Claffey   (17)   and   Zaniewski   et   al.   (26),   which   reported   that  
pavement   roughness   influences   fuel   consumption   by   30%   and   10%,   respectively.   However,   it  
was   pointed   out   by   Zaniewski   et   al.   that   the   sections   used   in   those   previous   experiments   did  
not  realistically  represent  operating  conditions  in  the  U.S.  due  to  potholes,  patches,  and  badly  
broken  portions  of  the  roadway  included  in  those  studies  (27).  
 
Four  different  studies  from  Belgium,  France,  South  Africa,  and  Sweden  (28-­‐31)  on  the  topic  of  
fuel  consumption  or  rolling  resistance  and  its  relationship  to  roughness  were  presented  at  the  
International   Symposium   on   Surface   Characteristics   in   1990.   Not   all   of   these   studies   directly  
measured  fuel  consumption,  and  their  methods  of  measuring  roughness  were  not  consistent.  
However,  they  agreed  with  previous  findings  that  a  relationship  between  pavement  roughness  
and  fuel  consumption  exists.  Descornet  identified  that  the  main  influence  on  rolling  resistance  
was   surface   profile   irregularities   in   the   wavelength   range   between   macrotexture   and  
unevenness  (28).  After  converting  rolling  resistance  to  fuel  consumption,  it  was  reported  that  
9%   fuel   savings   could   be   had   on   smooth   pavements   over   rough   pavements   based   on   the   range  
of   surface   characteristics   tested.   Laganier   and   Lucas   measured   fuel   consumption   directly   and  
characterized  road  evenness  with  the  longitudinal  profile  analyzer  to  determine  the  influence  
of  pavement  evenness  and  macrotexture  on  fuel  consumption  (29).  They  concluded  that  for  the  
range   of   pavements   tested,   extra   fuel   consumption   of   0   to   0.4   liters   per   100   km   resulted   for  

6  
Robbins  and  Tran      

surfaces   with   evenness   rating   from   excellent   to   poor.   Du   Plessis   et   al.   reported   that   for   the  
range  of  road  roughness  tested  in  their  study,  an  increase  in  fuel  consumption  of  up  to  20%  can  
be   expected   for   medium   trucks   and   buses   travelling   at   80   km/h   when   driven   on   very   rough,  
unpaved  roads  compared  to  smooth,  paved  roads  (30).  Sandberg  found  that  fuel  consumption  
was   “influenced   very   much   by   road   unevenness   and   megatexture,   but   somewhat   by  
macrotexture,”  such  that  the  “fuel  consumption  varies  over  a  range  of  approximately  11%  from  
the  smoothest  to  the  roughest  road  tested  if  texture  wavelengths  in  the  range  of  0.6  to  3.5  m  
are  considered”  (31).  McLean  and  Foley  (32)  summarized  these  studies  (28-­‐31)  among  others  
(19,   24,   25,   33)   relating   roughness   to   fuel   consumption   and/or   rolling   resistance   between   1982  
and  1990.  For  those  studies  that  reported  only  a  change  in  rolling  resistance,  McLean  and  Foley  
applied   a   factor   of   five   for   cars   and   four   for   trucks   to   convert   the   results   to   an   equivalent  
change   in   fuel   consumption.   The   roughness   measurements   were   not   consistent   among   these  
studies;  therefore,  McLean  and  Foley  converted  the  roughness  measurements  to  IRI  (32).  This  
resulted   in   IRI   values   that   ranged   between   0.5   and   15   m/km,   with   the   maximum   IRI   (converted  
from  the  report  by  du  Plessis  et  al.  (30))  equivalent  to  951  in/mile,  likely  due  to  the  earthen  and  
gravel  surfaces  included  in  that  study  and  well  above  what  the  FHWA  considers  unacceptable  
(IRI   greater   than   170  in/mile)   (22).   However,   on   the   low   end,   IRI   reported   by   McLean   and   Foley  
was  on  the  order  of  32  in/mile,  falling  into  FHWA’s  category  of  very  good.  McLean  and  Foley  
reported  that  the  percent  change  in  fuel  consumption  per  unit  of  IRI  ranged  from  0.5  to  3.6%  
for  a  car  and  0.5  and  4.1%  for  a  truck  (32).  
 
2.1.1   Recent  U.S.  Studies  
Barnes   and   Langworthy   chose   not   to   include   the   effect   of   pavement   roughness   on   fuel  
consumption   in   the   development   of   vehicle   operating   costs   in   Minnesota,   basing   their   decision  
on  the  notion  that  the  results  from  previous  studies  did  not  reflect  conditions  in  the  U.S.  (34,  
35).   However,   several   studies   in   the   U.S.   have   reported   a   positive   correlation   between  
pavement   roughness   and   fuel   consumption.   Results   from   WesTrack   revealed   a   positive  
correlation   between   pavement   roughness   and   fuel   consumption   (36).   Fuel   mileage   was  
recorded   for   eight   weeks   prior   to   and   seven   weeks   after   rehabilitation   of   a   section,   with   a   10%  
reduction   in   IRI.   A   4.5%   improvement   was   reported   in   fuel   mileage   on   the   smoother,  
rehabilitated   surface   over   the   deteriorated   surface.   It   was   reported   that   prior   to   the  
rehabilitation   the   pavement   roughness   was   such   that   the   “driver   would   not   be   able   to   tolerate  
the   ride   for   more   than   a   few   hours,”   with   IRI   values   reaching   or   exceeding   150   in/mile   (Epps   et  
al.,  1999)  (37).  The  rehabilitation  reportedly  improved  the  ride  with  IRI  values  of  approximately  
75  in/mile  noted  in  the  weeks  after  rehabilitation  (37).    
 
A   2004   preliminary   report   for   a   Florida   study   also   revealed   that   a   positive   correlation   existed  
between   pavement   smoothness   and   fuel   consumption   (38).   Fuel   consumption   was   measured  
on  five  pavement  sections,  with  four  of  the  five  sections  being  asphalt  concrete  pavement  and  
having   IRI   values   between   45.7   and   54.9   in/mile   and   the   fifth   section   being   a   concrete  
pavement  averaging  an  IRI  of  148.4  in/mile.    
 
In   a   2006   study,   the   Missouri   Department   of   Transportation   (MoDOT)   reported   a   2.461%  
increase  in  fuel  efficiency  on  new,  smoother  pavements  relative  to  the  rough  pavement  prior  to  

7  
Robbins  and  Tran      

resurfacing   (39).   As   part   of   Missouri’s   Smooth   Roads   Initiative   (SRI),   a   section   of   I-­‐70   was  
resurfaced   in   the   summer   of   2006.   Pavement   smoothness   was   determined   by   measuring   the  
roughness   and   vehicle   fuel   consumption   on   an   existing   section   of   I-­‐70   prior   to   and   after  
resurfacing   using   MoDOT’s   Automated   Road   Analyzer   (ARAN).   The   average   IRI   (including  
eastbound  and  westbound  directions)  before  resurfacing  was  130.25  in/mile  and  was  reduced  
53.2%  to  an  average  IRI  of  60.99  in/mile  after-­‐resurfacing.  Prior  to  resurfacing,  the  four  dump  
trucks  averaged  5.97  miles  per  gallon,  whereas  after  resurfacing,  the  average  was  improved  to  
6.11   miles   per   gallon.   It   was   also   found   that   the   use   of   the   vehicles’   brakes   decreased   58   times  
per   night   on   average   when   driven   on   the   smoother,   resurfaced   pavement   relative   to   the  
pavement   prior   to   resurfacing.   It   has   also   been   shown   at   the   National   Center   for   Asphalt  
Technology’s  Pavement  Test  Track  that  fuel  consumption  increases  with  increased  IRI  (40).  
 
As  part  of  the  NCHRP  1-­‐45  project  (and  documented  in  the  NCHRP  Report  720),  existing  vehicle  
operating   cost   models   were   reviewed   in   order   to   select   one   for   calibration   to   U.S.   conditions  
(20).   Based   on   Chatti   and   Zaabar’s   review,   the   HDM-­‐4   model,   the   widely   adopted   model   for  
computing   total   transport   costs,   was   selected.   The   model   was   then   calibrated   to   reflect  
roughness  levels  of  U.S.  roadways  and  improvements  in  vehicle  technology.  Prior  to  calibration,  
the  HDM-­‐4  models  were  evaluated  for  U.S.  conditions  having  a  range  of  IRI  between  0.8  and  8.5  
m/km   (51   to   539   in/mile).   In   evaluating   the   effect   of   roughness   and   texture   on   fuel  
consumption,  an  analysis  of  covariance  was  conducted  to  assess  the  quality  of  fit  for  the  HDM-­‐4  
fuel  consumption  model.  It  was  found  that  roughness  was  statistically  significant.  The  HDM-­‐4  
model  was  found  to  under-­‐predict  the  effect  of  roughness  on  fuel  consumption;  however,  the  
calibrated   HDM-­‐4   model   resulted   in   improved   predictions.   A   sensitivity   analysis   was   completed  
for   the   HDM-­‐4   and   the   calibrated   HDM-­‐4   model   for   fuel   consumption   relative   to   IRI.   The  
analysis  was  completed  for  a  speed  of  55  mph  (88  km/h).  Increasing  IRI  from  1  to  3  m/km  (63.4  
to   190.2   in/mile)   at   30°C   (86°F),   while   holding   mean   profile   depth   (MPD)   at   1   mm   and   grade   at  
0%,   resulted   in   a   0.5%   and   1.8%   increase   in   fuel   consumption   for   the   original   and   calibrated  
HDM-­‐4   models,   respectively,   for   light   trucks.   Likewise,   increasing   IRI   from   1   to   3   m/km   (63.4   to  
190.2   in/mile)   under   the   same   conditions   was   found   to   result   in   a   0.9%   and   2.9%   increase   in  
fuel   consumption   based   on   the   original   and   calibrated   HDM-­‐4   models,   respectively,   for  
articulated   trucks.   Chatti   and   Zaabar   concluded   that   the   most   important   pavement   condition  
factor   relative   to   fuel   consumption   was   surface   roughness   in   terms   of   IRI.   Chatti   and   Zaabar  
reported   that   an   increase   of   1   m/km   (63.4   in/mile)   would   result   in   an   approximate   2%   increase  
in  fuel  consumption  (regardless  of  speed)  for  passenger  vehicles.  For  heavy  trucks,  it  was  found  
that  at  normal  speeds  (60  mph),  a  1%  increase  would  result  from  the  same  increase  in  IRI  and  
about   a   2%   increase   for   a   1   m/km   (63.4   in/mile)   increase   at   a   low   speed   of   35   mph.   By  
decreasing  IRI  by  1  m/km  (63.4  in/mile),  Chatti  and  Zaabar  estimated  that  as  much  as  24  billion  
dollars   could   be   saved   in   fuel   costs   per   year   in   the   U.S.   based   on   a   3%   decrease   in   fuel  
consumption  for  the  255  million  passenger  cars  in  the  U.S.  using  fuel  prices  at  the  time  of  the  
report  (2012)  (20).    
 
2.2   Impact  of  Pavement  Roughness  on  Tire  Wear  Costs  
In  the  1987  HDM  report,  it  is  suggested  that  “road  improvements  can  have  a  disproportionate  
impact  on  tire  costs  relative  to  the  other  components”  (19).  Thus,  the  influence  of  pavement  

8  
Robbins  and  Tran      

condition   on   tire   wear   cost   has   been   an   important   part   of   vehicle   operating   cost   studies.   De  
Weille  reported  that  tire  wear  on  gravel  and  earth  roadways  was  higher  than  on  paved  roads  
(level   and   tangent)   (21).   He   also   reported   that   tire   wear   is   more   significant   than   engine   oil  
consumption   when   determining   the   operating   costs   of   a   vehicle,   particularly   for   heavier  
vehicles.    
 
Zaniewski   et   al.   (27   (and   later   summarized   by   Zaniewski   and   Butler   in   1985   (41))   estimated   tire  
wear   by   updating   cost   tables   previously   developed   by   Winfrey   (16)   and   by   applying   the   slip-­‐
energy   theory   developed   by   the   Forest   Service   to   estimate   tire   wear   for   the   1982   FHWA-­‐
sponsored   study   on   vehicle   operating   costs   (27).   The   slip-­‐energy   method   was   selected   for   their  
total   VOC   tables.   However,   it   was   reported   that   there   was   a   lack   of   data   to   support   the  
selection  of  coefficients  for  use  in  this  model  to  analyze  the  effect  of  surface  type  and  condition  
on   tire   wear.   To   evaluate   roadway   effects   on   tire   costs,   Zaniewski   et   al.   instead   utilized  
relationships  developed  in  a  previous  study  in  Brazil  to  determine  cost  adjustment  factors  for  
the   “proportionate   change   in   tire   consumption   as   roadway   surfaces   vary   from   the   assumed  
baseline   condition   of   3.5   SI”   (27).   Cost   adjustment   factors   increased   with   decreasing  
serviceability   index   (SI)   for   both   passenger   vehicles   and   single   unit   trucks,   indicating   that  
rougher  pavements  result  in  a  higher  tire  expense  related  to  tire  wear  (27,  41).    
 
Watanatada  et  al.  (19)  developed  a  tire  wear  prediction  model  as  a  function  of  vehicle  and  road  
characteristics.  Data  used  to  develop  the  model  was  from  the  Brazil  study  (23),  which  included  
a  road-­‐user  cost  survey  and  was  applicable  to  buses  and  trucks  with  cross-­‐ply  tires  only.  Limited  
data  were  available  for  the  cars  and  utility  vehicles.  As  a  result,  “the  tire  wear  prediction  model  
for   cars   and   utilities   was   calibrated   as   a   simple   linear   function   of   road   roughness”   (19).  
Watanatada   et   al.   compared   results   for   tire   wear   prediction   models   developed   in   the   four  
primary   cost   studies   conducted   between   1970   and   1982   with   their   mechanistic   tire   wear  
prediction  model.  The  tire  wear  models  showed  a  general  increase  in  tire  wear  in  terms  of  the  
number   of   equivalent   new   tires   per   vehicle   for   every   1,000   km   with   road   roughness.   It   was  
reported   that   the   models   developed   from   studies   in   Kenya   and   the   Caribbean   for  
medium/heavy   trucks   and   buses   “predicted   more   than   twice   the   influence   of   roughness”   (in  
reference  to  the  rate  of  increase  or  slope  of  the  linear  relationship).  However,  these  models  did  
not  include  the  impact  of  road-­‐geometry  on  tire  wear.  The  remaining  models  (Brazil  and  India)  
for   medium/heavy   trucks   and   buses   had   similar   influence   of   pavement   roughness   on   the  
prediction   of   tire   wear   over   the   entire   range   of   roughness.   Chesher   and   Harrison   reported   that  
the   effects   of   road   roughness   were   larger   in   the   Kenyan   and   Caribbean   models   than   in   the  
Brazilian  and  Indian  models  for  truck  tire  consumption.  An  increase  in  roughness  from  5.1  to  8.5  
m/km  (323.34  to  538.9  in/mile)  would  predict  an  increase  in  tire  consumption  of  11,  10,  31  and  
27%,  respectively  for  the  Indian,  Brazilian  (medium  truck),  Caribbean,  and  Kenyan  models  (18).  
For  the  tire  wear  prediction  model  developed  in  1987,  it  was  also  reported  that  roughness  has  a  
small  effect  on  tire  wear  for  a  level  road  and  a  much  greater  effect  when  the  road  is  steep  (19).    
 
Barnes   and   Langworthy   (34)   based   their   adjustment   factors   for   vehicle   operating   costs   of  
personal   vehicles   due   to   pavement   roughness   on   the   study   conducted   by   Zaniewski   et   al.   in  
1982  (27).    Barnes  and  Langworthy  concerned  that  in  the  Zaniewski  et  al.  report,  “the  impacts  

9  
Robbins  and  Tran      

of   roughness   on   operating   costs   seem   unrealistically   large,   especially   for   smoother   pavement  
levels”  limited  the  range  of  pavement  roughness  used  to  develop  adjustment  factors  for  vehicle  
operating  costs  in  their  study  for  the  Minnesota  Department  of  Transportation  (MnDOT)  (34).  
As   a   result,   they   developed   adjustment   factors   for   passenger   vehicles   for   various   levels   of  
pavement   roughness   defined   by   PSI   and   IRI   (note,   the   conversion   between   IRI   and   PSI   is   not  
stated),   as   shown   in   Table   2.   The   authors   “assume   that   pavement   roughness   will   affect   truck  
costs  in  the  same  way  as  car  costs”  (34).  For  truck  costs,  they  used  an  increase  of  5%  for  a  PSI  of  
3.0  (IRI  of  105  in/mile),  15%  for  a  PSI  of  2.5  (IRI  of  140  in/mile),  and  25%  for  a  PSI  of  2.0  (170  
in/mile)   and   below   to   account   for   the   effect   of   pavement   roughness   on   all   vehicle   operating  
costs  with  the  exception  of  fuel  consumption  costs.    
 
Table  2  Adjustment  Factors  for  Pavement  Roughness  levels  for  Passenger  vehicles  (after  34)  
PSI   IRI   IRI  (m/km  or  mm/m)   Adjustment  multiplier  
(in/mile)  
≤  2.0   170   2.7   1.25  
2.5   140   2.2   1.15  
3.0   105   1.7   1.05  
≥  3.5   80   1.2   1.00  
 
Based   on   a   consensus   in   previous   literature   that   pavement   roughness   affected   maintenance,  
tire,  and  repair  costs,  Barnes  and  Langworthy  considered  an  effect  of  pavement  roughness  on  
all   three   cost   components,   as   well   as   depreciation   (34).   Using   a   set   of   assumptions  
(maintenance   cycles,   tire-­‐life,   fuel   costs   and   fuel   economy   corresponding   vehicle   type,   repair  
cycles   and   and   five-­‐year   repair   costs,   and   depreciation   rates),   they   estimated   baseline   per-­‐mile  
costs  (in  2003  cents)  for  a  highway  with  smooth  pavement.  They  also  calculated  the  cents  per  
mile   costs   using   the   same   assumptions   for   the   baseline   conditions   for   an   “extremely   rough”  
highway  pavement  (corresponding  to  a  PSI  of  2.0  or  170  in/mile,  which  is  consistent  with  the  
threshold  identified  by  FHWA,  in  Table  1).  Costs  associated  to  tire  wear  due  to  extremely  rough  
pavements   resulted   in   an   increase   over   the   baseline   costs   by   0.2   cents   per   mile   for  
automobiles,   0.2   cents   per   mile   for   pickup/van/SUVs,   and   0.9   cents   per   mile   for   commercial  
trucks  (34).  
 
In  applying  the  calibrated  HDM-­‐4  model  for  U.S.  conditions,  Chatti  and  Zaabar  concluded  that  
by  increasing  IRI  by  1  m/km  (63.4  in/mile),  tire  wear  for  passenger  cars  and  heavy  trucks  would  
increase   by   1%   at   88   km/h   (55   mph)   (20).   Furthermore,   it   was   found   that   based   on   the  
calibrated  VOC  models  resulting  from  the  NCHRP  1-­‐45  project,  for  the  same  IRI  value,  at  a  grade  
of  0%,  the  tire  wear  increases  with  increasing  speed  and  thus,  the  effect  of  roughness  is  greater  
at   higher   speeds.   Chatti   and   Zaabar   estimated   that   decreasing   IRI   by   1   m/km   (63.4   in/mile)  
could  result  in  savings  for  tire  wear  costs  of  340  million  dollars  per  year  based  on  255  million  
passenger  cars  with  an  average  annual  mileage  of  15,000  miles,  an  average  tire  life  of  45,000  
miles  and  an  average  cost  of  $100  per  tire.  
 
 
 

10  
Robbins  and  Tran      

2.3   Impact  of  Pavement  Roughness  on  Maintenance  and  Repair  Costs  
Based   on   vehicle   fatigue   response   testing,   pavement   roughness   influences   the   response   of  
vehicle  suspension  and  can  result  in  accelerated  fatigue  (42).  In  addition,  correlations  between  
maintenance  and  repair  costs  and  pavement  roughness  have  been  observed  in  several  studies  
in  the  U.S.  and  abroad.  However,  the  modeling  of  maintenance  parts  and  labor  costs  received  
the  least  research  attention  of  all  the  VOC  components  (14).    
 
As  was  done  for  tire  wear  expenses,  Zaniewski  et  al.  (27,  41)  updated  maintenance  and  repair  
costs  based  on  Winfrey’s  cost  tables  (16).  They  then  used  the  updated  costs  in  conjunction  with  
the   information   from   the   earlier   study   in   Brazil   to   develop   adjustment   factors   to   account   for  
changes   in   roadway   surface   conditions.   The   factors   identified   the   proportionate   changes   in  
expenses   as   the   roadway   surface   deviated   from   a   baseline   serviceability   index   of   3.5.  
Adjustment  factors  for  maintenance  and  repair  expenses  were  determined  for  passenger  cars,  
single   unit   trucks,   and   semi-­‐trucks.   As   serviceability   index   decreases,   adjustment   factors  
increase,   thus   resulting   in   additional   expenses   for   lower   serviceability   indices   (i.e.,   increased  
pavement  roughness).    
 
The   four   primary   studies   on   user   costs   indicated   an   effect   of   pavement   roughness   on  
maintenance  costs  (18).  As  a  result  of  the  Kenyan  and  Caribbean  studies,  road  roughness  was  
modeled  as  a  linear  relationship  relative  to  parts  consumption,  whereas  the  Brazilian  and  Indian  
models  revealed  an  exponential  relationship  between  parts  consumption  and  roughness.  While  
this   exponential   relationship   results   in   very   large   increases   in   parts   consumptions   for   rough  
roads,  there  are  notable  differences  between  the  studies.  The  differences  in  the  relationships  
between  parts  consumption  and  roughness  are  in  part,  attributed  to  the  surface  type  and  the  
effects   of   roughness   on   vehicle   deterioration   due   to   various   models   of   vehicles.     It   should   be  
noted  that  the  data  collected  in  the  Caribbean  and  India  were  mostly  on  paved  roads,  while  the  
data   collected   in   Kenya   were   from   paved   and   unpaved   surfaces   and   the   study   conducted   in  
Brazil  included  paved,  gravel  and  earthen  roads.    
 
According  to  Watanatada  et  al.,  an  attempt  was  made  to  develop  a  wholly  mechanistic  model  
for  the  vehicle  maintenance  parts  using  data  from  the  Brazilian  study;  however  it  could  not  be  
developed  due  in  part  to  lack  of  sufficient  data  (19).  Rather,  simpler  models  correlating  spare  
parts   and   mechanics’   labor   with   road   characteristics   were   developed.   Spare   parts   consumption  
was   found   to   be   dependent   on   road   roughness   and   vehicle   age,   and   the   effects   combined  
“multiplicatively”.   When   age   was   held   constant,   the   relationship   between   parts   consumption  
and   roughness   was   found   to   be   generally   non-­‐linear,   consistent   with   an   exponential  
relationship   reported   by   Chesher   and   Harrison   (18).   As   a   result,   the   parts   consumption   cost  
model  used  an  exponential  relationship  up  to  a  transitional  value  of  roughness  (in  QI).  Beyond  
the   transitional   value,   a   linear   relationship   was   used   to   alleviate   overprediction   of   parts  
consumption   costs   for   high   values   of   roughness.   Relative   to   the   maintenance   labor   hours  
model,  it  was  reported  that  only  buses  had  a  significant  effect  in  the  Brazilian  study.  However,  a  
coefficient  for  roughness  was  included  in  the  exponential  relationship  with  labor  hours  used  in  
the  model,  with  the  caveat  that  a  coefficient  with  value  zero  would  imply  an  insignificant  effect  
of  roughness.  Referring  to  the  maintenance  cost  component  of  the  total  vehicle  running  costs,  

11  
Robbins  and  Tran      

the  authors  note  that  it  “increases  sharply  with  increase  in  roughness.”  The  authors  observed  
that  for  a  utility  vehicle  on  a  paved  road,  the  increase  in  predicted  maintenance  costs  is  slightly  
less  than  three-­‐fold  when  increasing  the  roughness  from  QI  25  (1.92  m/km  or  121.92  in/mile,  
assuming  IRI  =  QI/13)  to  QI  125  (9.62  m/km  or  609.62  in/mile).  On  very  rough  surfaces,  it  was  
found  that  the  proportion  of  maintenance  costs  to  total  costs  increases  considerably.    
 
In   1992,   Poelman   and   Weir   published   a   report   on   vehicle   fatigue,   specifically   the   fatigue   of  
vehicle   suspension   components   due   to   pavement   roughness   (42).   In   this   study,   two   different  
instrumented   vehicles   were   driven   over   25   sites   with   PSI   ranging   from   0   to   2.5.   A   roughness  
meter   was   used   to   measure   surface   roughness   of   the   roadway,   which   resulted   in  
measurements  of  its  longitudinal  profile  in  centimeters  per  kilometer  or  inches  per  mile.  Once  
calibrated   to   the   test   location,   these   values   were   converted   to   values   of   PSI   based   on   the   state  
of   Pennsylvania’s   conversion   equations   for   asphalt   and   concrete   pavements.   The   authors  
compare  the  resulting  PSI  values  to  the  same  scale  as  PSR,  where  a  PSI  value  of  0  is  equivalent  
to   a   PSR   of   0   and   refers   to   a   pavement   that   is   impassable.   Likewise   a   PSI   value   of   2.5   is  
equivalent  to  2.5  on  the  PSR  scale  and  is  categorized  as  a  rating  of  “fair”.  The  fatigue  response  
was  measured  in  suspension  components  of  the  vehicles  as  they  were  driven  at  traffic  speeds  
ranging  between  25  and  50  miles  per  hour.  For  pavements  with  PSI  less  than  2.5  (categorized  as  
“fair”   pavements)   and   equal   to   or   greater   than   1.0,   vehicle   suspension   fatigue   was   found   to  
occur   and   greatly   accelerate   on   roadways   with   a   PSI   less   than   1.0   (categorized   as   “poor”  
pavements).  
 
Barnes  and  Langworthy  accounted  for  the  effect  of  pavement  roughness  by  applying  a  percent  
increase  in  each  vehicle  operating  cost  related  to  trucks,  including  costs  for  maintenance  and  
repair   based   on   the   level   of   PSI   (IRI)   (34).   The   adjustment   factors   developed   in   the   MnDOT  
study   imply   an   additional   maintenance   and   repair   cost   of   one   cent   per   mile   between   the  
smoothest  and  roughest  pavements  (35).  Barnes  and  Langworthy  estimated  baseline  costs  for  
each  vehicle  type  based  on  assumptions  as  described  earlier  for  tire  wear  costs.     They  went  on  
to   estimate   maintenance   and   repair   costs   in   cents   per   mile   for   an   extremely   rough   highway  
pavement  assuming  a  PSI  of  2.0  (with  IRI  equivalent  reported  as  170  in/mile)  while  holding  all  
other   assumptions   consistent   with   baseline   estimates.   In   comparing   the   maintenance   and  
repair   costs   to   the   baseline   maintenance   and   repair   costs   (smooth   highway   pavement),   an  
additional   0.8   cents   per   mile   was   reported   for   automobiles,   1.0   cents   per   mile   for  
pickup/van/SUVs,  and  1.8  cents  per  mile  was  reported  for  commercial  trucks  (34).  
 
Chatti  and  Zaabar  reported  that  for  their  calibrated  HDM-­‐4  model,  no  effect  of  roughness  was  
found  on  repair  and  maintenance  costs  up  to  an  IRI  of  3  m/km  (190.2  in/mile)  (20).  However,  
they   reported   an   increase   in   roughness   beyond   3   m/km   (190.2   in/mile)   would   result   in  
increased  repair  and  maintenance  costs  and  as  shown  in  Figure  2,  the  rate  of  increase  varies  by  
IRI  level.  For  example,  an  increase  in  IRI  from  3  to  4  m/km  (190.2  to  253.6  in/mile)  would  result  
in  a  10%  increase  in  repair  and  maintenance  costs  for  passenger  cars  and  heavy  trucks,  while  an  
increase  from  4  to  5  m/km  would  result  in  a  30  to  40%  increase  in  repair  and  maintenance  costs  
for  passenger  cars  and  heavy  trucks,  respectively.  It  was  reported  that  decreasing  IRI  by  1  m/km  
(63.4  in/mile)  could  result  in  a  savings  of  between  24.5  and  73.5  billion  dollars  per  year  in  repair  

12  
Robbins  and  Tran      

and   maintenance   costs.   These   estimates   were   made   by   applying   the   10%   and   30%   incremental  
changes  in  repair  and  maintenance  costs  for  passenger  cars  listed  above  for  IRI  greater  than  3  
m/km   (190.2   in/mile),   using   annual   repair   and   maintenance   costs   totaling   $244.8   billion   for  
passenger  cars  (20).        
 

 
Figure  2  Effect  of  Roughness  on  Repair  and  Maintenance  Costs  (20)  
 
2.4   Impact  of  Roughness  on  Oil  Consumption  Costs  
De   Weille   compared   oil   consumption   by   road   type,   which   showed   earthen   roads   resulted   in  
higher  oil  consumption  than  paved  roads  (21).  However,  De  Weille  reported  that  this  “is  by  far  
the  least  important  in  the  total  makeup  of  vehicle  operating  costs”.  Furthermore,  Chesher  and  
Harrison  stated  that  lubricant  costs  are  generally  small,  making  up  less  than  3%  of  total  VOC,  
and  they  are  difficult  to  analyze  (18).  Accordingly,  this  component  of  vehicle  operating  costs  has  
not   been   researched   extensively   and   may   explain   the   limited   research   in   the   area   of   the  
influence  of  pavement  roughness  on  oil  consumption  costs.    
 
Although  limited,  research  has  shown  that  the  influence  can  be  high  for  cars  in  India.  Chesher  
and  Harrison  summarized  the  predicted  engine  oil  consumption  based  on  the  four  primary  user  
cost   studies   (India,   Brazil,   Kenya,   and   Caribbean)   (18).   Based   on   their   Table   5.6,   it   is   deduced  
that  in  Kenya,  the  predicted  oil  consumption  for  a  truck  increased  two-­‐fold  from  an  IRI  of  2.8  to  
7.4  m/km  (177.5  to  469.2  in/mile).  The  effect  was  less  pronounced  in  Brazil—the  same  change  
in  IRI  for  a  truck  resulted  in  a  20%  increase  in  predicted  oil  consumption.  
 
Previous  truck  oil  consumption  estimates  developed  by  Winfrey  in  1969  (16)  were  updated  to  
reflect   longer   intervals   between   oil   changes   and   differences   in   oil   consumption   information  

13  
Robbins  and  Tran      

from   trucking   firms   (27,   41).   The   relationship   determined   in   the   1982   Brazilian   study   for   oil  
consumption   due   to   pavement   roughness   was   extrapolated   for   the   updated   oil   consumption  
estimates  in  the  U.S.  As  a  result,  cost  adjustment  factors  were  developed  for  the  normal  range  
of  SI  in  the  U.S.  (2-­‐4)  assuming  a  baseline  SI  of  3.5.  Decreasing  the  SI  from  4.0  to  2.0  for  trucks  
resulted  in  a  change  in  adjustment  factor  from  about  0.82  for  an  SI  of  4.0  and  about  1.1  for  an  
SI  of  2.0,  indicating  that  rougher  pavements  increase  oil  consumption  costs.  The  same  change  in  
serviceability  for  passenger  cars  resulted  in  a  much  more  severe  change  in  adjustment  factors.    
 
2.5   Impact  of  Pavement  Roughness  on  Depreciation  Costs  
Limited   research   has   been   conducted   on   the   effect   of   pavement   roughness   on   vehicle  
depreciation.   According   to   Chesher   and   Harrison,   “little   information   on   depreciation   and  
interest  costs”  was  provided  in  the  four  primary  cost  studies  (Brazil,  the  Caribbean,  India,  and  
Kenya)  (18).  The  difficulty  in  relating  roughness  to  depreciation  costs  was  explained,  “in  order  
to   estimate   the   effects   of   varying   road   characteristics   on   vehicle   depreciation   and   interest  
costs,  one  needs  to  establish  the  effects  of  road  characteristics  on  (i)  vehicle  life  and  (ii)  vehicle  
utilization.   Unfortunately,   neither   of   these   relationships   has   ever   been   properly   quantified  
empirically;   traditionally   they   have   simply   been   assumed   in   benefit-­‐cost   studies   of   road  
investments”  (19).    
 
In   the   HDM-­‐III   model,   a   model   for   vehicle   depreciation   and   interest   costs   was   developed   based  
on  the  relationships  previously  investigated  as  part  of  the  four  primary  cost  studies  (Brazil,  the  
Caribbean,  India,  and  Kenya)  (19).  While  annual  vehicle  depreciation  costs  were  quantified  with  
respect  to  the  age  of  the  vehicle  in  Kenya,  India,  and  Brazil,  the  effects  of  road  characteristics  
on   annual   vehicle   depreciation   were   not   included   (19).   However,   the   effects   of   road  
characteristics   on   depreciation   and   interest   costs   per   kilometer   were   incorporated   into   the  
Brazil  models  through  vehicle  utilization  (19).  
 
Zaniewski   et   al.   contributes   the   difficulty   in   estimating   depreciation   costs   accurately   to   the  
determination  of  “what,  if  any,  portion  of  the  expense  should  be  assigned  to  operation  on  the  
road”   (27).   In   their   1982   FHWA   study,   the   use-­‐related   expense   was   considered   by  
approximating   the   reciprocal   of   the   maximum   vehicle   life   mileage   and   data   from   the   earlier  
study  in  Brazil  to  adjust  expenses  for  pavement  conditions  through  cost  adjustment  factors  as  a  
function   of   serviceability   index   (27,   41).   To   determine   vehicle   operating   costs   in   Minnesota,  
Barnes   and   Langworthy   included   depreciation,   basing   their   decision   on   experience,   which  
suggested  “a  car  that  is  driven  almost  exclusively  on  smooth  highways  will  last  more  miles  than  
one   that   is   driven   mostly   on   rough   pavement”   (34).   They   estimated   depreciation   costs   for   each  
vehicle  type  on  an  extremely  rough  pavement  (approximately  170  in/mile)  assuming  the  same  
set  of  circumstances  as  was  assumed  for  the  baseline  estimates  (smooth  highway  pavement).  
Depreciation   costs   on   the   extremely   rough   roadway   were   approximately   25%   greater   for  
automobiles,  pickup/van/SUVs,  and  commercial  trucks  compared  to  baseline  estimates  for  each  
vehicle  type.    
 
   

14  
Robbins  and  Tran      

3.   SUMMARY  
It   has   been   shown   through   various   studies   dating   back   several   decades   that   pavement  
roughness   influences   vehicle   operating   costs.   This   report   discussed   the   effect   of   pavement  
roughness  on  the  various  components  of  vehicle  operating  costs  (fuel,  tire  wear,  maintenance  
and  repair,  oil  consumption,  and  depreciation).    
 
Extensive   work   has   been   completed   on   the   subject   of   the   effects   of   pavement   roughness   on  
fuel   consumption.   For   the   majority   of   the   literature   reviewed,   pavement   roughness   was  
reported   to   have   a   positive   relationship   with   fuel   consumption   and   its   costs,   such   that   an  
increase   in   pavement   roughness   resulted   in   increased   fuel   consumption   costs.   It   should   be  
noted,   however,   that   much   of   the   early   studies,   particularly   those   conducted   in   developing  
countries,   were   conducted   on   roadway   surfaces   with   roughness   levels   that   extend   well   beyond  
those  considered  unacceptable  in  the  U.S.  Although  a  1982  FHWA  study  reported  that  such  an  
effect   was   not   statistically   significant   (27,   41)   and   researchers   in   Minnesota   dismissed   any  
effect   of   pavement   roughness   on   fuel   consumption,   citing   that   testing   conditions   in   previous  
studies   were   inconsistent   with   U.S.   conditions   (34,   35),   there   have   been   several   studies  
conducted   in   the   U.S.   that   have   found   pavement   roughness   to   influence   fuel   consumption.  
Studies  conducted  at  WesTrack  (36),  Florida  (38),  Missouri  (39),  and  NCAT  (40)  found  that  for  
roughness  levels  seen  in  the  U.S.,  pavement  roughness  influences  fuel  consumption  and  thus,  
influences   fuel   consumption   costs.   Additionally,   the   widely   adopted   HDM-­‐4   model   for  
computing   total   transportation   costs   was   calibrated   to   U.S.   conditions,   reflecting   roughness  
levels   and   improvements   in   vehicle   technology   (20).   The   calibrated   HDM-­‐4   fuel   consumption  
model   was   used   to   determine   that   1   m/km   (63.4   in/mile)   increase   in   IRI   effects   fuel  
consumption  by  as  much  as  2%.    
 
Costs   associated   with   tire   wear,   maintenance   and   repair,   oil   consumption,   and   depreciation  
were  also  found  to  be  influenced  by  pavement  roughness.  Although  research  in  the  areas  of  oil  
consumption   and   depreciation   costs   was   limited,   early   studies   in   developing   countries   found  
that   an   increase   in   pavement   roughness   resulted   in   an   increase   in   oil   consumption   (18).  
Maintenance   and   repair   costs,   as   well   as   tire   wear   costs,   were   also   found   to   increase   with  
increases  in  roughness,  albeit  the  rate  varied  from  study  to  study.  Chatti  and  Zaabar  reported  
an   effect   of   roughness   on   repair   and   maintenance   costs   for   IRI   levels   greater   than   3   m/km  
(190.2   in/mile)   (20).   While   this   effect   varies   by   vehicle   type   and   IRI   level,   they   reported   an  
increase   in   roughness   from   3   to   4   m/km   (190.2   to   253.6   in/mile)   could   increase   repair   and  
maintenance  costs  by  10%  for  passenger  cars  and  heavy  trucks.  Chatti  and  Zaabar  also  reported  
that  a  1  m/km  (63.4  in/mile)  reduction  in  IRI  could  translate  to  a  savings  of  340  million  dollars  
per  year  in  tire  wear  costs  for  passenger  vehicles.  
 
   

15  
Robbins  and  Tran      

REFERENCES  
1. Swanlund,  M.  Enhancing  Pavement  Smoothness.  Public  Roads.  Vol.  64,  No.  2,  Federal  
Highway  Administration,  U.S.  Department  of  Transportation,  Washington,  D.C.,  2000.  
2. Biehler,  A.  Rough  Roads  Ahead:  Fix  Them  Now  or  Pay  for  Them  Later.  AASHTO,  
Washington,  D.C.,  2009.  
3. Bumpy  Roads  Ahead:  America’s  Roughest  Rides  and  Strategies  to  Make  our  Roads  
Smoother.  The  Road  Information  Program  (TRIP),  Washington,  D.C.,  2013.  
4. ASTM  E867-­‐06,  Standard  Terminology  Relating  to  Vehicle-­‐Pavement  Systems.  American  
Society  of  Testing  and  Materials  International,  West  Conshohocken,  PA,  2012.  
5. Carey,  W.  N.,  Jr.  and  P.  E.  Irick.  The  Pavement-­‐Serviceability  Concept.  Highway  Research  
Board  Bulletin  250,  National  Research  Council,  Washington,  D.C.,  1960,  pp.  40-­‐58.  
6. Sayers,  M.  W.,  T.  D.  Gillespie,  and  A.  V.  Queiroz.  The  International  Road  Roughness  
Experiment:  Establishing  Correlation  and  a  Calibration  Standard  for  Measurements.  
World  Bank  Technical  Paper  Number  45,  World  Bank,  Washington,  D.C.,  1986.  
7. Sayers,  M.  W.,  T.  D.  Gillespie,  and  C.  A.  V.  Queiroz.  The  International  Road  Roughness  
Experiment:  A  Basis  for  Establishing  a  Standard  Scale  for  Road  Roughness  
Measurements.  Transportation  Research  Record  1084,  Presented  at  65th  Annual  
Meeting  of  the  Transportation  Research  Board,  Washington,  D.C.,  1986,  pp.  76-­‐85.  
8. Gillespie,  T.  D.  Everything  You  Always  Wanted  to  Know  about  the  IRI,  But  were  Afraid  to  
Ask!  Proc.,  Road  Profile  Users  Group  Meeting,  Lincoln,  NE,  September  22-­‐24,  1992.  
9. Shahin,  M.  Y.  Pavement  Management  for  Airports,  Roads,  and  Parking  Lots,  2nd  ed.  
Springer  Science,  New  York,  NY,  2005.  
10. Sayers,  M.  W.  and  S.  M.  Karamihas.  The  Little  Book  of  Profiling:  Basic  Information  about  
Measuring  and  Interpreting  Road  Profiles.  The  Regent  of  the  University  of  Michigan,  Ann  
Arbor,  MI,  1998.  
11. Pavement  Interactive.  Pavement  Management:  Evaluation  Categories.  June  5,  2009.  
http://www.pavementinteractive.org/article/pavement-­‐evaluationevaluation-­‐
categories/.  Accessed  March  1,  2015.  
12. Sayers,  M.  W.,  T.  D.  Gillespie,  and  W.  D.  O.  Paterson.  Guidelines  for  Conducting  and  
Calibrating  Road  Roughness  Measurements.  World  Bank  Technical  Paper  Number  46,  
World  Bank,  Washington,  D.C.,  1986.  
13. WSDOT  Pavement  Roughness  (IRI)  Report:  2010.  Part  of  WSDOT  Pavement  Notebook,  
Washington  State  Department  of  Transportation,  2012.  
14. Bennett,  C.  R.  and  I.  D.  Greenwood.  Modelling  Road  User  and  Environmental  Effects  in  
HDM-­‐4.  Volume  Seven,  The  Highway  Development  and  Management  Series,  The  World  
Road  Association  (PIARC),  Paris,  France,  2001.    
15. Law,  H.  and  D.  Kinnear-­‐Clark.  The  Construction  of  Roads  and  Streets.  Crosby  Lockwood  &  
Co.,  London,  1877.  
16. Winfrey,  R.  Economic  Analysis  for  Highways.  International  Textbook  Company,  Scranton,  
PA,  1969.  
17. Claffey,  P.  NCHRP  Report  111:  Running  Costs  of  Motor  Vehicles  as  Affected  by  Road  
Design  and  Traffic.  HRB,  National  Research  Council,  Washington,  D.C.,  1971.    

16  
Robbins  and  Tran      

18. Chesher,  A.  and  R.  Harrison.  Vehicle  Operating  Costs:  Evidence  from  Developing  
Countries.  The  Highway  Design  and  Maintenance  Standards  Series,  The  Johns  Hopkins  
University  Press,  Baltimore,  MD,  1987.  
19. Watanatada,  T.,  A.  M.  Dhareshwar,  and  P.  R.  S.  Rezende-­‐Lima.  Vehicle  Speeds  and  
Operating  Costs:  Models  for  Road  Planning  and  Management.  The  Highway  Design  and  
Maintenance  Standard  Series,  The  Johns  Hopkins  University  Press,  Baltimore,  MD,  1987.  
20. Chatti,  E.  K.  and  I.  Zaabar.  NCHRP  Report  720:  Estimating  the  Effects  of  Pavement  
Condition  on  Vehicle  Operating  Costs.  Transportation  Research  Board  of  the  National  
Academies,  Washington,  D.C.,  2012.  
21. De  Weille,  J.  Quantification  of  Road  User  Savings.  World  Bank  Staff  Occasional  Papers  
Number  Two,  International  Bank  for  Reconstruction  and  Development,  The  John  
Hopkins  Press,  Baltimore,  MD,  1966.  
22. Federal  Highway  Administration.  Smoothness.  U.S.  Department  of  Transportation.  
http://www.fhwa.dot.gov/pavement/smoothness.  Accessed  April  10,  2014  
23. Research  on  the  Interrelationships  Between  Costs  of  Highway  Construction,  
Maintenance,  and  Utilization:  Final  Report  on  Brazil-­‐UNDP  Highway  Research  Project.  
Volume  5,  Final  Report,  Empresa  Brasileira  de  Planejamento  de  Transportes  (GEIPOT),  
Brasilia,  Brazil,  1982.  
24. Ross,  F.  R.  Effect  of  Pavement  Roughness  on  Vehicle  Fuel  Consumption.  In  
Transportation  Research  Record:  Journal  of  the  Transportation  Research  Board,  No.  846,  
TRB,  National  Research  Council,  Washington,  D.C.,  1982,  pp.  1-­‐6.  
25. Bester,  C.  Effect  of  Pavement  Type  and  Condition  on  the  Fuel  Consumption  of  Vehicles.  
In  Transportation  Research  Record:  Journal  of  the  Transportation  Research  Board,  No.  
1000,  TRB,  National  Research  Council,  Washington,  D.C.,  1984,  pp.  28-­‐32.  
26. Zaniewski,  J.  P.,  B.  K.  Moser,  P.  J.  de  Morasis,  and  R.  L.  Kaesehagen.  Fuel  Consumption  
Related  to  Vehicle  Type  and  Road  Conditions.  In  Transportation  Research  Record:  
Journal  of  the  Transportation  Research  Board,  No.  702,  TRB,  National  Research  Council,  
Washington,  D.C.,  1979,  pp.  328-­‐334.  
27. Zaniewski,  J.  P,  B.  C.  Butler,  G.  Cunningham,  G.  E.  Elkins,  M.  S.  Paggi  and  R.  Machemehl.  
Vehicle  Operating  Costs,  Fuel  Consumption,  and  Pavement  Type  and  Condition  Factors.  
Report  No.  FHWA/PL/82/001,  Federal  Highway  Administration,  Washington,  D.C.,  1982.  
28. Descornet,  G.  Road-­‐surface  Influence  on  Tire  Rolling  Resistance.  In  Surface  
Characteristics  of  Roadways:  International  Research  and  Technologies,  ASTM  STP  1031,  
American  Society  for  Testing  and  Materials,  Philadelphia,  PA,  1990,  pp.  401-­‐415.  
29. Laganier,  R.,  and  J.  Lucas.  The  Influence  of  Pavement  Evenness  and  Macrotexture  on  
Fuel  Consumption.  In  Surface  Characteristics  of  Roadways:  International  Research  and  
Technologies,  ASTM  STP  1031,  American  Society  for  Testing  and  Materials,  Philadelphia,  
PA,  1990,  pp.  454-­‐459.  
30. du  Plessis,  H.  W.,  A.  T.  Visser,  and  P.  C.  Curtayne.  Fuel  Consumption  of  Vehicles  as  
Affected  by  Road-­‐Surface  Characteristics.  In  Surface  Characteristics  of  Roadways:  
International  Research  and  Technologies,  ASTM  STP  1031,  American  Society  for  Testing  
and  Materials,  Philadelphia,  PA,  1990,  pp.  480-­‐496.  
31. Sandberg,  U.  S.  Road  Macro-­‐and  Megatexture  Influence  on  Fuel  Consumption.  In  
Surface  Characteristics  of  Roadways:  International  Research  and  Technologies,  ASTM  

17  
Robbins  and  Tran      

STP  1031,  American  Society  for  Testing  and  Materials,  Philadelphia,  PA,  1990,  pp.  460-­‐
479.  
32. McLean,  J.  and  G.  Foley.  Road  Surface  Characteristics  and  Condition:  Effects  on  Road  
Users.  Research  Report  ARR  314,  ARRB  Transport  Research,  1998.  
33. Young,  J.  The  Influence  of  Road  Uneveness  on  Vehicle  Fuel  Consumption.  Proceedings  of  
Seminar  E,  16th  PTRC  Summer  Annual  Meeting,  Bath,  UK,  1988,  pp.  217-­‐227.  
34. Barnes,  G.  and  P.  Langworthy.  The  Per-­‐Mile  Costs  of  Operating  Automobiles  and  Trucks.  
Final  Report  MN/RC  2003-­‐19,  Minnesota  Department  of  Transportation,  St.  Paul,  MN,  
2003.  
35. Barnes,  G.  and  P.  Langworthy.  Per  Mile  Costs  of  Operating  Automobiles  and  Trucks.  In  
Transportation  Research  Record:  Journal  of  the  Transportation  Research  Board,  No.  
1864,  TRB,  National  Research  Council,  Washington,  D.C.,  2004,  pp.  71-­‐77  
36. Epps,  J.  A.,  A.  Hand,  S.  Seeds,  T.  Schulz,  S.  Alavi,  C.  Ashmore,  J.  A.  Deacon,  J.  T.  Harvey,  
and  R.  Leahy.  NCHRP  Report  455:  Recommended  Performance-­‐Related  Specification  for  
Hot-­‐Mix  Asphalt  Construction:  Results  of  the  Westrack  Project.  Transportation  Research  
Board  of  the  National  Academies,  Washington,  D.C.,  2002.  
37. Epps,  J.  A.,  R.  B.  Leahy,  T.  Mitchell,  C.  Ashmore,  S.  Seeds,  S.  Alavi,  and  C.L.  Monismith.  
Westrack  the  Road  to  Performance-­‐Related  Specifications.  Proceedings  of  International  
Conference  on  Accelerated  Pavement  Testing,  Reno,  NV,  October  18-­‐20,  1999.    
38. Jackson.  N.  M.  An  Evaluation  of  the  Relationship  Between  Fuel  Consumption  and  
Pavement  Smoothness.  Preliminary  Report,  University  of  North  Florida,  Jacksonville,  FL,  
2004.  
39. Amos,  D.  Pavement  Smoothness  and  Fuel  Efficiency:  An  Analysis  of  the  Economic  
Dimensions  of  the  Missouri  Smooth  Road  Initiative.  Final  Report  OR07-­‐005,  Missouri  
Department  of  Transportation,  Jefferson  City,  Missouri,  2006.    
40. Heffernan,  M.  E.  Simulation,  Estimation,  and  Experimentation  of  Vehicle  Longitudinal  
Dynamics  That  Effect  Fuel  Economy.  Master’s  Thesis,  Auburn  University,  Auburn,  AL,  
2006.  
41. Zaniewski,  J.  P.  and  B.  C.  Butler.  Vehicle  Operating  Costs  Related  to  Operating  Mode,  
Road  Design,  and  Pavement  Condition.  In  Measuring  Road  Roughness  and  Its  Effects  on  
User  Cost  and  Comfort,  ASTM  STP  884,  American  Society  for  Testing  and  Materials,  
Philadelphia,  PA,  1985,  pp.  127-­‐142.  
42. Poelman,  M.  A  and  R.  P  Weir.  Vehicle  Fatigue  Induced  by  Road  Surface  Roughness.  In  
Vehicle,  Tire,  Pavement  Interface  (J.J.  Harvey  and  J.C.  Wambold,  eds.),  ASTM  STP  1164,  
American  Society  for  Testing  and  Materials,  Philadelphia,  PA,  1992,  pp.97-­‐111.  
 

18