soil paper.pdf

7/27/2019 soil paper.pdf

1/15

Proceedings of the Eastern Asia Society for Transportation Studies, Vol. 5, pp. 428 - 440,2005

PAVEMENT PERFORMANCE MODEL FORFEDERAL ROADS

Lt Col Abdul Hamid Mohd Isa, Dadang Mohamed Ma'soemResearch Associate Lecturer34, Jalan Intan Baiduri 4D Department of Civil EngineeringTaman Intan Baiduri Faculty of Engineering52100 Kuala Lumpur University Putra MalaysiaMalaysia 43400 Serdang, SelangorFax: 603-2071 3465 MalaysiaE-mail: [email protected] Fax: 603-8943 2509E-mail: [email protected] Tiek HwaLecturerDepartment of Civil EngineeringFaculty of EngineeringUniversity Putra Malaysia43400 Serdang, SelangorMalaysiaFax: 603-8943 2509E-mail: [email protected]

Abstract: Malaysia has a matured flexible pavement road network owned by thepublic sector called the Federal Road. The 14,757 .KIn road networks are providingexcellent service to road users for interstate movement. The industrialisation processof the country demands for more movements at higher load to transport products to


2/15


1. INTRODUCTIONPavement deterioration process is complex and involves not only structural fatiguebut also involves many functional distresses of pavement. It result from theinteraction between traffic, climate, material and time. Deterioration is used torepresent the change in pavement performance overtime. The ability of the road tosatisfy the demands of traffic and environment over its design life is referred to asperformance. Due to the great complexity of the road deterioration process,performance models are the best approximate predictors of expected conditions.There are many parameters that need to be acquired to successfully predict the rate ofpavement deterioration. Among others is annual average daily traffic (AADT),percent of trucks, drainage, pavement thickness, pavement strength in term ofstructural number (SN) or CBR value and mix design parameters.The objective of this study is to establish simple practicable pavement performancemodel for network level of the Malaysian Federal Road where rutting is the focus ofthe measurement. The model shall incorporate relevant variables such as pavementcondition, pavement strength, traffic loading and pavement age. Statistical analysis bymean of multiple linear regressions were conducted to test and examine the data aswell as to develop the model.1.1 Rutting and Possible Influencing Factors

Mix DesignParameters


3/15

Proceedings of the Eastern Asia Society for Transportation Studies, Vol. 5, pp. 428 - 440, 2005

1.1.1 RuttingOne of the most common distresses on flexible pavement is surface rutting. Rutting isdefined as longitudinal deformation or depression in the wheel paths, which occurafter repeated application of axle loading. It may occur in one or both wheel path of alane. It can be categorised as either traffic load associated deformation, wear relatedor the combination of both. The causes include traffic load, age of pavement anddeformation of the entire pavement structure or instability in the form of one or morepavement layers as explained by Figure 1. In engineering term the deformation occurdue to the critical strain experience by the top layer of subgrade as shown in Figure 2.Rutting make the road surface uneven, patchy and bumpy and subsequently affectsthe handling of vehicles which can lead to safety problems. The unit use to measurerut depth is millimetre (mm) .

Wheel load

Bituminous layerCompressive strain at the topof subgrade formation level,related to permanent

deformation !_f_ Granular layer_

Tensile strain at thebottom of asphalt,related to cracking

f r


4/15


1.1.3 Pavement ThicknessPavement layers and its thickness plays very important role in distributing wheel loadto underlying subgrade. More layer or thicker pavement structure would mean leastload being distributed to the subgrade and subsequently reduce vertical critical strain.1.1.4 Pavement Condition (Roughness)Present pavement condition state will influence it future performance. As pavementsurface deteriorate over time, one distress condition would add up to another distresscondition if no appropriate action taken to correct the initial condition. This wouldlead to more costly maintenance. This paper use pavement roughness as pavementcondition measure in mm/m.1.1.5Tramc loadingThe magnitude and number of wheel load passes is the main agent to deteriorate thepavement surface. Heavy and medium trucks normally fitted with large axles wouldsignificantly damage the surface as well as deform the underlying pavement layerspermanently. Normally medium truck loading is used to predict pavementdeterioration and referred as annual average daily traffic (AADT), it is measured invehicles per day (vpd).1.1.6 Pavement Strength

Another significant element that resists permanent pavement deformation is pavementstrength. The pavement can be measured in term of CBR (California Bearing Ratio)


5/15


Table 1: Samples ofData

Ser Section Name Length Rut Depth Roughness MT AADT SN(Km) (mm) (mm/m) (vpd)

1 001035000-001035999-1 1 9.20 4.37 39846 3.79 - __ - __._ - .0 _ P' __ _ .. _ _ . . . . _ .. __ __ ..2 001036000-001036999-2 1 5.13 3.02 19923 3.41_. _. -_ _ '- _ _ _ - P"p - p_ __ A .. _ __3 001036000-001036999-3 1 6.37 3.52 19923 3.41..---- -_.- --- --- - --_ .. _.". -- _ -_ ..-- ----_ -_ --_ - -- -_ --- -_ .....4.._9.Q! n Q 9 g : 9 ~ ! ~ ~ ? ~ ? ~ : ! _. _ ? _._ ~ } L ~ .. ~ .._. __ ! ? ~ ~ ~ . _ 4 : ? ~ . 5 001122000-001136999-1 15 3.09 1.98 10626 5.69-_ . -_ _ ." - p - p- - _. _ - - - - - --_ -- -_ .. -_ _ -_ -_ -.6 001137000-001138999-1 2 4.03 2.73 10626 4.16................................. _ ---. - -.- . -----.-.. ---. ----_.- ------- -.--------_ --------. - .. ----7 001155000-001156999-1 2 4.83 3.36 10626 3.98_.. _ _ -_ - _. _ _._ -- --.. -_ -- -_ --_ .. ---_.. -_. _.. - -_ .8 001157000-001160999-1 4 3.39 2.19 10317 4.97.- . . _ _ _ - _- _.- _.- --- -_.. __ .- . __ ---_ -- _A. _ _._ _. __ -._.

9 001197000-001208999-1 12 3.11 1.92 10804 5.32 '_ . _ .. ' -_ _ _. _ - - _. __ A __ __ __ - _ - _ _. __ _ _. _ 10 001270000-001270999-1 1 6.75 2.04 13480 2.95----- --_ --_.- ---" ' .. -- --_.- _ -.. ------- ._-- .-- --- --_._. ------ _ _ -----_ _-----_.. -----_..11 001318000-001319999-1 2 6.05 2.53 39718 4.20-_ .. - _ -. -- _- .-_.- _ -.-_._ _. ----_._. -_ -_.- -- . _.---_._ -. _..12 001416000-001416999-1 1 5.35 2.29 45631 4.51_ -.-.-.--_. ---.-..--------- _.. ----. --.. ---------.--.-------.. ----------- ------_ ----- .. -.13 001417000-001437999-1 21 3.17 1.84 8729 5.08.. -_.. _ _ -_ _.. _ _..'--_. _.. _. _ -_.. _.. -.. _. -_.. ---- -_._. -- .. _. _ _. -- -_. _.. _.- _..14 001438000-001441999-1 4 1.96 1.98 8729 5.06._- -- .- . __ . __ . _. --- .. -_ .. _. _._- --,--- ---_.- ._- _.. _... __ .-..-._-- -- ---------- -_.. _..._- --_.. _.. -.--- ._- ._.15 001444000-001446999-1 3 2.13 1.87 8729 5.67.-. --_.. - _.. -.- --_.- - -. -_.- --.. _ ----- ------ -_. --------_._. ----------_ .. ------ _.-_. ------- -_ _.16 001463000-001468999-1 6 3.23 1.44 9284 5.35_. __ . -_ ... -,_. _.. _.. _... _.... -_. _.'.-.. _ . ---_.- _.- -. -_ .. -_ ... __ .------- -_._ .. __ ... --- -- _.... ---- _._. -_.- ----17 001517000-001518999-3 2 3.13 2.16 5621 5.30---- ---- --- -- _...... -- -----_ .... --. ---- ._- .. _._- -----_ ... -------_._._._. -- ---_.... _.. -.- .. _.- --.- 18 001539000-001539999-2 1 3.18 2.93 6123 7.72 -.. -_ _ ._ _.- . _. - ---- --_. _.. _. --- .. -.. --- --_ .. --_._ _.. --. _.. _ 19 001539000-001539999-3 1 3.55 3.35 6123 7.72_. -_ -.._..---_ _ --. - _ _. - _. _ _. -- _. _.. _-_ --- _. --- ----. _.. -_.. ----..-_ -_..20 001580000-001581999-1 2 6.26 3.08 24135 4.87._ .... - - - ~ . _.. ..-- ...---_. -_... _.. --.. _. _... _.. _. _.. --_ ..._. --_. -- _. _.. _. --_._... _... -..._.. -_._ .....------- _...21 001610000-001611999-2 2 10.13 2.65 21618 3.51

, ---- ... -.. -. -------........ -_. --- .. _.. .. -- .. ----. -.- .-------- ...-----------.. --- . -- -... ---.----. -.. -


6/15

~ r o c e e d i n g s of the Eastern Asia Society for Transportation Studies, Vol. 5, pp. 428 - 440, 2005

3. Developing The Pavement Performance Models

3.1 Regression Analyses and Statistical TestsSeveral regression analyses were conducted on to the data set. The initial analyses aremeant to see the response of among variables and statistical tests were performed onthe developed models namely the coefficient of determinant and level of significant.The main tests conducted to validate the models are:

a. Coefficient of determinant (R2) .b. Level of significant at 0.05c. The standard error of the estimate.d. The heteroscedasticity of variance.e. Normality tests.

The coefficient of determinant (R2) and level of significant at 0.05 are the maincriteria to be achieved for the selection of good model. Coefficient of determinant(R2) describes the strength of an association between variables. An associationbetween variables means that the value of one variable can be predicted, to someextent, by the value of the other. A correlation is a special kind of association: there isa linear relation between the values ofthe variables.The standard error of the estimate is a measure of the accuracy of predictions madewith a regression line. The regression line seeks to minimise the sum of the squarederrors of prediction. A model with lowest value of standard error of estimate ispreferred.


7/15


3.2 ResultsRegression # 1 with ConstantAfter tabulating the data, first regression analysis was run using SPSS by stepwisemethod. Rut depth was assigned as dependent variables where as roughness, AADTand SN were assigned as independent variables. Significant level (a) of 0.05 was set.It is appear that the following models were suggested:

Table 2: Initial Regression Models With ConstantDependent Variable: Rut DepthModel Constant SN Roughness AADT Std

ErrI 8.620 -0.796 1.477 0.426................{Q:Q9.QL JQ.Q9.QL "'" __ .2 4.473 -0.608 1.243 1.217 0.620___ ... __ .... .(Q:Q9.Q2. .... - ( Q . Q 9 ~ 2 __ .. _... (9:9.Q9L. _. __ . .... ,' _...... __ . __ . . __ .. __ ... __ ....3 3.439 -0.473 1.046 6.373 x 10' 1.112 0.691(0.003) (0.001) (0.000) (0.005)

Value ofDurbin Watson statistic (mean constant variance) = 1.71\ Value in bracket is the significant level value (a)

The main purpose for this initial analysis is to observe the response of the variableswhile identifying the potential model. Constant was included in this model. From theanalysis Model # 3 appear to be the best among them; it can be written in the


8/15


Table 3: Second Regression Models With ConstantDependent Variable: Rut DepthModel Roughness AADT Std

Err1 1.762 1.494 0.909

(0.000).- . -- .. -- -- . -- .. -. - . . . """ .. - .. -- . "'" ' " . ' - --5 -._. . . -"""""" - ' - ' _.-.- - -..2 1.281 9.359 x 10' 1.272 0.936(0.000) (0.000) Value ofDurbin Watson statistic (mean constant variance) = ] .877 Value in bracket is the significant level value (a)

In this analysis the R2 and Durbin Watson value seem to be good; however thismodel dropped the SN variable. Since SN value is an important indicator forpavement strength, it is essential that SN value to be included in the model, howeverTable 4 suggest model # 2 as the best model and written as follows:Rut Depth = 1.281 (Roughness) +9.359 x 10'5 (AADT) (2)Regression # 3 Using Transformed Variables Without ConstantSince SN is expected to be included in the model to represent the strength of thepavement in question, transformation of variables is a possible solution. Oncetransformation of variables is done, another regression analysis was conducted atsignificant level (a) of 0.05 and without constant. The following results obtained:

Table 4: Third Regression Models ofTransformedVariables without Constant
http:///reader/full/%E5%AE%AD-.%E5%AE%95--%E5%95%AD-%E5%95%AE--%E5%95%AE.%E5%95%AD.-%E5%95%80http:///reader/full/%E5%AE%AD-.%E5%AE%95--%E5%95%AD-%E5%95%AE--%E5%95%AE.%E5%95%AD.-%E5%95%80http:///reader/full/%E5%AE%AD-.%E5%AE%95--%E5%95%AD-%E5%95%AE--%E5%95%AE.%E5%95%AD.-%E5%95%80http:///reader/full/%E5%AE%AD-.%E5%AE%95--%E5%95%AD-%E5%95%AE--%E5%95%AE.%E5%95%AD.-%E5%95%80http:///reader/full/%E5%AE%AD-.%E5%AE%95--%E5%95%AD-%E5%95%AE--%E5%95%AE.%E5%95%AD.-%E5%95%80http:///reader/full/%E5%AE%AD-.%E5%AE%95--%E5%95%AD-%E5%95%AE--%E5%95%AE.%E5%95%AD.-%E5%95%80http:///reader/full/%E5%AE%AD-.%E5%AE%95--%E5%95%AD-%E5%95%AE--%E5%95%AE.%E5%95%AD.-%E5%95%80http:///reader/full/%E5%AE%AD-.%E5%AE%95--%E5%95%AD-%E5%95%AE--%E5%95%AE.%E5%95%AD.-%E5%95%80http:///reader/full/%E5%AE%AD-.%E5%AE%95--%E5%95%AD-%E5%95%AE--%E5%95%AE.%E5%95%AD.-%E5%95%80http:///reader/full/%E5%AE%AD-.%E5%AE%95--%E5%95%AD-%E5%95%AE--%E5%95%AE.%E5%95%AD.-%E5%95%80http:///reader/full/%E5%AE%AD-.%E5%AE%95--%E5%95%AD-%E5%95%AE--%E5%95%AE.%E5%95%AD.-%E5%95%80http:///reader/full/%E5%AE%AD-.%E5%AE%95--%E5%95%AD-%E5%95%AE--%E5%95%AE.%E5%95%AD.-%E5%95%80


9/15


Rut Depth = 1.565 (Roughness) + 8.002 x 10.5 (AADT) - 0.022 (SN)2 (3)

Regression # 4 Using Transformed Variables With ConstantAnother trial was made to obtain a model with a constant in the equation. For thisanalysis the dependent variable was transformed to logarithmic. The followingmodels were suggested:

Table 5: Fourth Regression Models ofTransformedVariables with ConstantDependent Variable: Log Rut DepthModel Constant SN :z Roughness AADT StdErr

1 0.825 -0.007 0.130 0.454_. __ .. _ _. - C Q . Q 9 ~ L -CQ.Q9.QL _ _ _ _ _ _ _ _ __ . __ ..2 0.479 -0.005 0.116 0.104 0.660... --.. - -CQ:Q9.QL._. --(Q:Q9.QL (Q:Q.QQl - --.. - --T -- ' .3 0.415 -0.004 0.100 5.473 x 10" 0.095 0.723(0.000) (0.000) (0.000) (0.030)

Value ofDurbin Watson statistic (mean constant variance) = 1.470 Value in bracket is the significant level value (a)

The equation of the third model can from the above table suggest low standard errorwith reasonable value of coefficient of determinant (R2) . The equation can be writtenas:


10/15

Proceedings ofthe Eastern Asia Society for Transportation Studies, Vol. 5, pp. 428440,2005

b. Model # 2Rut Depth = 1.281 (Roughness) + 9.359 x 10-5 (AADT) (2)

R2 = 0.936Durbin Watson Value = 1.877Standard Error = 1.272c. Model # 3Rut Depth = 1.565 (Roughness) + 8.002 x 10.5 (AADT) - 0.022 (SNi (3)

R2 = 0.943Durbin Watson Value = 1.771Standard Error = 1.212d. Model # 4Log Rut Depth = 0.415 + 0.1 (Roughness) + 5.473 x 10-6 (AADT)- 0.004 (SN)2 (4)

R2 =0.723Durbin Watson Value = 1.470Standard Error = 0.954The above models have their own strength and weaknesses based on the validity inexplaining the variables and homogeneity of variance of the residual. Model # 3 hasthe highest value of R2, however the model did not represented by a constant. Model


11/15


Rutting Vs Number of ~ ~ ; - - - - ---I17.516.515.5E 14.5

.. 13.5C).5 12.5==~ 11.5

10.59.58.5

.../ '/ '/ '/ --

JI # . / "~ .// : - - - - - - - - - I - - - - - ~ - _ . - .---- - - - - - ~ - ~ - - - - - - - - - - - - - -

.... \111 2 3 4 5 6 7 8 9

Number of YearFigure 3: Rutting Prediction Graph Based on the DevelopedModel

Using this prediction, appropriate steps can be taken to schedule maintenanceactivities subsequently assist the pavement managers in budgeting and disseminating


12/15

oceedings of the Eastern Asia Society for Transportation Studies, Vol. 5, pp. 428 - 440, 2005

c. Durbin Watson Statistic CD Value). The D value for the selectedmodel is 1.470 which is close to 2. It mean the variance of the residual isalmost constant; i.e. homoscedastic.d. Normality Test. Normality test was conducted on the data set usingprobability plot and its emerged to be short tail, which means that theprobability plot have non-linear pattern. This reveals that the data set does notfit normal distribution.

3.5 Using The ModelSay, the average value of roughness and structural number (SN) for the pavementsurface in question are 4mm/m and 7 respectively, it is also forecasted for the next tenyears of the AADT for medium truck is shown in the following table. Computingvalues of rutting and structural number into the final equation (Model # 4), rut depthis predicted for the next ten years.

Table 6: Predicted Rutting using the Developed ModelYears Medium Truck AADT (vpd) Rutting (mm)

_____ ._L " ) _ ~ t 9 Q 9 __ .. ._ . ~ ~ ~ .. ._. __...... _._ _ _._ ~ . q t 9 Q 9 _ _. __ _._ _ _ ~ ~ ~ _. _ .__ _} ___ ~ ? t 9 ~ 9 """ ._ _. _ .. . ~ : ~ _ __ ..__ ...... __ _ __ _?Qt9Q9.. . . ~ : ~ _ .. _. _._____ ..? __ . _. ? _ ~ t 9 Q 9 .__ . . ._. _. ._. ~ : 9 . ._...... .? __ ."" .._. _ _ ? q t 9 ~ 9 __ __ . ! . ~ : ~ _ _........? __ _"'" _ ??t9Q9_ _._ _ _. _. __ __ . ! . ~ : ~ __ .


13/15


I Rutting Vs Number of YearI 17.516.5 + - - - - - - - - - - - - - - - - - - - ~ 15.5 +-----------

E 14.5 +-----.. 13.5en.5 12.5:::~ 11.5

.._-------_ . . . . _.._._. .--0.59.5 +--.::7""--- - - - - - 11 - - - - - - - .. --....-.---8.5 +-+---r-+-,--f-., . . . . .-I--,--" '-.--+-r-+-.--+-.., . .-+---,--t----1

1 2 3 4 5 6 7 8 9 10 INumber of Year

- - - . - 7 " " -+ - - - - - - - - - . . . . - ------ .- .-- .

Figure 3: Rutting Prediction Graph Based on the DevelopedModelUsing this prediction, appropriate steps can be taken to schedule maintenanceactivities subsequently assist the pavement managers in budgeting and disseminating


14/15


REFERENCES

a) Books

Peterson D.E. (1977) Good Road Cost Less. Utah Department ofTransport, UtahRalph Haas. (1997) Pavement Design and Management Guide. TransportationAssociation ofCanada

b) Journalpapen

Dr. Charles Nunoo. (2002) Pavement Management System, Florida InternationalUniversity, FloridaKuo et al. (2003) Development ofPavement Performance Model', TransportResearch BoardProzzi and Madanat. (2002) A Non Linear Model for Predicting PavementServiceability, Transport Research BoardSamer Madanat. (1997) Predicting Pavement Deterioration, Institute ofTransportation Studies

PAVEMENT CONDITION RATING SHEET


15/15

LOCATION DATE & DAYROAD NO TYPES OF PAVEMENTCLIMATE WIDTH

Types of Distress Rating Rating Given RemarksA. Surface Defects1. Fatty Surface 0-52. Smooth Surface 0-53. Streaking 0-54. Hungry Surface 0-5B. Cracks1. Hairline Cracks 0-52. All igator Cracks 0-53. longitudinal Cracks 0-54. Edge Cracks 0-55. Shrinkage Cracks 0-56. Reflection Cracks 0-5C. Deformation1. Slippage 0-52. Rutting 0-53. Corrugation 0-54. Shoving 0-55. Shallow Dipression 0-56. Settlement & Upheavel 0-5D. Disintegration1. Stripping 0-52. Lossof Aggragate 0-53. Ravelling 0-54. Potholes 0-55. Edge Breaking 0-5

5VERY SEVERE4

SEVERE3

2 t'-IGHT1

VERYtLlGHT0

soil paper.pdf

Documents