| Return to Main
Page Predicting Roughness Progression of Asphalt Overlays JOINT
CSHRP/ALBERTA BAYESIAN APPLICATION APPENDIX D Encoding Package Quick Memo RE: RCI Model Development using Bayesian Method Dear Experts, With the following package we are beginning to encode your expert judgement regarding the overlayed AC pavements. Your expert judgement should be encoded by filling out Figure 7 and 8 in this package. I would greatly appreciate your cooperation and help and as fast a response as possible. Due to various reasons we are a little behind the schedule. The output of this exercise should be ready for late April of this year. If you have any questions regarding the encoding procedure please contact me. Thank you very much for your help.
First Generation Bayesian Roughness Model (for pavements with granular bases in central Alberta) A model to predict the roughness of ACP
overlays on AC pavements with granular bases in central
Alberta is being developed. Judgement on the roughness
performance of overlays is being solicited from pavement
experts. This judgement will be analyzed together with
the data taken from the Department's Pavement Management
System. Bayesian statistical method will be used to
analyze the two sources of data i.e., the PMS data and
the data obtained from the Expert judgement. Your
assistance in providing your judgement is very much
appreciated. This is the first time the Bayesian method
is being used in the Department and we hope that after
this first trial it could be more often used where there
is not enough "hard" data but where the
professional expertise definitely exists. Data Base from PMS The Department's PMS datable has been
searched to develop the experimental data base for this
project. Several limitations has been established to
start with as simple model as possible and to work with a
manageable data base. The first limitation was that only
pavements with granular base would be considered. The
pavements with granular bases constitute approximately 80
percent of all pavements in the Province and so such a
limitation is justified. The second limitation was to
consider only those granular base pavements which are
located in the central part of Alberta. This way the
consideration of the environmental factor which would
definitely have influence on the roughness progression
can be avoided. Only those inventory sections which are
longer then 3 kilometres have been included. The next two
limitation was concerned with the number of times the RCI
survey was performed during the pavement life. Only those
sections which had at least five RCI measurements during
their pavement original life and those which were tested
at least three times after the overlay have been
considered. In all 85 pavement inventory sections
have been selected giving in total 310 roughness
observations. Dependant Variable Roughness Comfort Index (RCI) of the
overlaid AC pavement has been selected as the dependent
variable. Independent Variables The Bayesian roughness model will contain six independent variables:
Soil Roughness Factor In order to define the effect different soils have on the progression of pavement roughness the Soil Roughness Factor was developed. The factor quantifies the aggressiveness of different soil types with respect to roughness. The Soil Roughness Factor (SRF) was developed based on output in a questionnaire you have filled out. The SRF is rated on the scale from 0 to 5 where the "0" reflects a passive soil environment, i.e. soils having very little or no effect on the roughness progression, and a value of "5" reflects a soil type that contributes substantially to roughness progression. The Soil Roughness Factor for different soils is shown in Table 1. Table 1 - Soil Roughness Factors The distribution of the Soil Roughness Factor for all soil types encountered in our data base is shown in Figure 1. Figure 1 - Dist. of Roughness Factor I propose to encode your expert judgement on the following ranges of this variable: · Soil which does not contribute to RCI decrease SRF=1.0 · Soil which contributes greatly to RCI decrease SRF=4.0. Two encoding matrices are attached at
the end of this package. Each matrix characterizes soil
which have different influence on the overlay roughness
progression. Roughness Performance of Original Pavement It is generally accepted that pavements
which performed well with respect to roughness in the
past would so after the overlay. Good past performance
indicates that the pavement structure was designed
properly taking into account traffic, Subgrade conditions
and the climatic environment. In our study the performance of original pavement is described by the annual rate of change in RCI. This was calculated as follows: Such an approach assumes that the RCI
decreases linearly with time. This could be an
approximation as it is generally accepted that the RCI
progression follows a certain curvature. On the other
hand our data show that by assuming linearity we could be
not very far from the reality. The past performance of all inventory sections considered in the analysis is shown in Figure 2. This figure indicates that the past performance variable is normally distributed. It ranges from 0.02 to 0.28 with the average of 0.14. I propose to encode your expert judgement on two levels of this variable: · excellent past performance (annual change of RCI=0.05) · poor past performance (annual change of RCI=0.25).
Initial Roughness of Overlay The initial roughness of overlay indicates how smooth was the overlaid pavement in the year of overlay construction or one year after the construction. Our PMS based data base shows that this variable (see Figure 3.) is almost normally distributed, with the range from 6.0 to 8.3 and average RCI of 7.5. It is proposed that your expert judgement will be encoded at two levels of this variable: · very smooth initial overlay (RCI = 8.5) · rough initial overlay (RCI = 6.5). Overlay Thickness This variable will reflect the effect
of AC thickness on the development of pavement surface
distresses. The surface distress in turn will influence
the pavement roughness. The overlay thickness in our data base ranged from 50 to 130 mm. As shown in Figure 4 this variable is normally distributed with the average of 100 mm. It is proposed to encode the expert judgement on two levels of this variable: · thin overlays (50 mm) · thick overlays (120 mm) Traffic This variable will quantify the influence the traffic has on the progression of overlay roughness. Only traffic (in terms of ESAL/year) after the overlay is taken into account. Our data base shows that distribution of traffic was skewed towards the lower values indicating that majority of considered sections had low traffic volumes (Figure 5). It is intended that this variable be encoded on two levels: · low volume road (traffic = 50,000 ESAL/Year) · high volume road (traffic = 250,000
ESAL/Year). Age of Overlay Age influence the roughness progression in two ways. Firstly, the older the overlay the more chances are for more diverse environmental conditions (such as extremely cold winters or unusually hot summers) to occur. Such conditions will be responsible for development of certain surface distress. Secondly, with age traffic loads will have greater effect on the overlay. Our data base (Figure 6) indicates that our overlays range from 2 to 18 years of age, and that the age of overlay distribution is skewed towards younger pavements. It is proposed that your expert judgement be encoded on three levels: · very young overlay (age = 2 years) · average age (age = 8 years) · old overlay (age = 16 years). |
