Local calibration of flexible pavement performance models in michigan

Syed Waqar Haider, Wouter C. Brink, Neeraj Buch

    Research output: Contribution to journalArticle

    Abstract

    The performance prediction models in the Pavement-ME are nationally calibrated using in-service pavement material properties, pavement structure, climate and truck loading conditions, and performance data obtained from the Long Term Pavement Performance program. Generally, the nationally calibrated models may not be accurate if the inputs and performance data used to calibrate do not represent a state’s local conditions and practices. Therefore, each state highway agency (SHA) should evaluate the nationally calibrated performance models to determine the adequacy of predicted field performance before implementing the new M-E design procedure. If the predictions are not satisfactory, local calibration of the Pavement-ME performance models is recommended to improve the performance prediction capabilities reflecting the unique field conditions and design practices. The commonly used calibration technique such as split sampling does not necessarily provide adequate results, especially with small sample sizes. Consequently, there is a need to employ statistical methodologies that are more efficient and robust for model calibrations given the data related challenges encountered by SHAs. The bootstrap is a nonparametric and robust resampling technique for estimating standard errors and confidence intervals of a statistic. The main advantage of resampling methodologies like bootstrapping includes estimation of a parameter without making distribution assumptions. This paper presents the use of resampling techniques to locally calibrate the flexible pavement performance models and how the locally calibrated Pavement-ME models improved the performance prediction accuracy. The results of the local calibration show that the validation standard error and bias obtained from bootstrapping were much lower than other resampling techniques. In addition, the validation statistics were similar to that of the model calibration, which indicates robustness of the local model coefficients. The reliability equations after local calibration are a better representation of measured pavement performance in Michigan.

    Original languageEnglish (US)
    Pages (from-to)986-997
    Number of pages12
    JournalCanadian Journal of Civil Engineering
    Volume43
    Issue number11
    DOIs
    StatePublished - Sep 28 2016

    Profile

    pavement
    Pavements
    calibration
    Calibration
    prediction
    bootstrapping
    methodology
    Statistics
    confidence interval
    road
    sampling
    climate
    Trucks
    Materials properties
    Sampling

    Keywords

    • Bootstrapping
    • Fatigue cracking
    • IRI
    • Jackknifing
    • Local calibrations
    • Rutting

    ASJC Scopus subject areas

    • Civil and Structural Engineering
    • Environmental Science(all)

    Cite this

    Local calibration of flexible pavement performance models in michigan. / Haider, Syed Waqar; Brink, Wouter C.; Buch, Neeraj.

    In: Canadian Journal of Civil Engineering, Vol. 43, No. 11, 28.09.2016, p. 986-997.

    Research output: Contribution to journalArticle

    Haider, Syed Waqar; Brink, Wouter C.; Buch, Neeraj / Local calibration of flexible pavement performance models in michigan.

    In: Canadian Journal of Civil Engineering, Vol. 43, No. 11, 28.09.2016, p. 986-997.

    Research output: Contribution to journalArticle

    @article{cbf0d08cbf6149f8936fb4efe71a8384,
    title = "Local calibration of flexible pavement performance models in michigan",
    abstract = "The performance prediction models in the Pavement-ME are nationally calibrated using in-service pavement material properties, pavement structure, climate and truck loading conditions, and performance data obtained from the Long Term Pavement Performance program. Generally, the nationally calibrated models may not be accurate if the inputs and performance data used to calibrate do not represent a state’s local conditions and practices. Therefore, each state highway agency (SHA) should evaluate the nationally calibrated performance models to determine the adequacy of predicted field performance before implementing the new M-E design procedure. If the predictions are not satisfactory, local calibration of the Pavement-ME performance models is recommended to improve the performance prediction capabilities reflecting the unique field conditions and design practices. The commonly used calibration technique such as split sampling does not necessarily provide adequate results, especially with small sample sizes. Consequently, there is a need to employ statistical methodologies that are more efficient and robust for model calibrations given the data related challenges encountered by SHAs. The bootstrap is a nonparametric and robust resampling technique for estimating standard errors and confidence intervals of a statistic. The main advantage of resampling methodologies like bootstrapping includes estimation of a parameter without making distribution assumptions. This paper presents the use of resampling techniques to locally calibrate the flexible pavement performance models and how the locally calibrated Pavement-ME models improved the performance prediction accuracy. The results of the local calibration show that the validation standard error and bias obtained from bootstrapping were much lower than other resampling techniques. In addition, the validation statistics were similar to that of the model calibration, which indicates robustness of the local model coefficients. The reliability equations after local calibration are a better representation of measured pavement performance in Michigan.",
    keywords = "Bootstrapping, Fatigue cracking, IRI, Jackknifing, Local calibrations, Rutting",
    author = "Haider, {Syed Waqar} and Brink, {Wouter C.} and Neeraj Buch",
    year = "2016",
    month = "9",
    doi = "10.1139/cjce-2015-0556",
    volume = "43",
    pages = "986--997",
    journal = "Canadian Journal of Civil Engineering",
    issn = "0315-1468",
    publisher = "National Research Council of Canada",
    number = "11",

    }

    TY - JOUR

    T1 - Local calibration of flexible pavement performance models in michigan

    AU - Haider,Syed Waqar

    AU - Brink,Wouter C.

    AU - Buch,Neeraj

    PY - 2016/9/28

    Y1 - 2016/9/28

    N2 - The performance prediction models in the Pavement-ME are nationally calibrated using in-service pavement material properties, pavement structure, climate and truck loading conditions, and performance data obtained from the Long Term Pavement Performance program. Generally, the nationally calibrated models may not be accurate if the inputs and performance data used to calibrate do not represent a state’s local conditions and practices. Therefore, each state highway agency (SHA) should evaluate the nationally calibrated performance models to determine the adequacy of predicted field performance before implementing the new M-E design procedure. If the predictions are not satisfactory, local calibration of the Pavement-ME performance models is recommended to improve the performance prediction capabilities reflecting the unique field conditions and design practices. The commonly used calibration technique such as split sampling does not necessarily provide adequate results, especially with small sample sizes. Consequently, there is a need to employ statistical methodologies that are more efficient and robust for model calibrations given the data related challenges encountered by SHAs. The bootstrap is a nonparametric and robust resampling technique for estimating standard errors and confidence intervals of a statistic. The main advantage of resampling methodologies like bootstrapping includes estimation of a parameter without making distribution assumptions. This paper presents the use of resampling techniques to locally calibrate the flexible pavement performance models and how the locally calibrated Pavement-ME models improved the performance prediction accuracy. The results of the local calibration show that the validation standard error and bias obtained from bootstrapping were much lower than other resampling techniques. In addition, the validation statistics were similar to that of the model calibration, which indicates robustness of the local model coefficients. The reliability equations after local calibration are a better representation of measured pavement performance in Michigan.

    AB - The performance prediction models in the Pavement-ME are nationally calibrated using in-service pavement material properties, pavement structure, climate and truck loading conditions, and performance data obtained from the Long Term Pavement Performance program. Generally, the nationally calibrated models may not be accurate if the inputs and performance data used to calibrate do not represent a state’s local conditions and practices. Therefore, each state highway agency (SHA) should evaluate the nationally calibrated performance models to determine the adequacy of predicted field performance before implementing the new M-E design procedure. If the predictions are not satisfactory, local calibration of the Pavement-ME performance models is recommended to improve the performance prediction capabilities reflecting the unique field conditions and design practices. The commonly used calibration technique such as split sampling does not necessarily provide adequate results, especially with small sample sizes. Consequently, there is a need to employ statistical methodologies that are more efficient and robust for model calibrations given the data related challenges encountered by SHAs. The bootstrap is a nonparametric and robust resampling technique for estimating standard errors and confidence intervals of a statistic. The main advantage of resampling methodologies like bootstrapping includes estimation of a parameter without making distribution assumptions. This paper presents the use of resampling techniques to locally calibrate the flexible pavement performance models and how the locally calibrated Pavement-ME models improved the performance prediction accuracy. The results of the local calibration show that the validation standard error and bias obtained from bootstrapping were much lower than other resampling techniques. In addition, the validation statistics were similar to that of the model calibration, which indicates robustness of the local model coefficients. The reliability equations after local calibration are a better representation of measured pavement performance in Michigan.

    KW - Bootstrapping

    KW - Fatigue cracking

    KW - IRI

    KW - Jackknifing

    KW - Local calibrations

    KW - Rutting

    UR - http://www.scopus.com/inward/record.url?scp=84994750594&partnerID=8YFLogxK

    UR - http://www.scopus.com/inward/citedby.url?scp=84994750594&partnerID=8YFLogxK

    U2 - 10.1139/cjce-2015-0556

    DO - 10.1139/cjce-2015-0556

    M3 - Article

    VL - 43

    SP - 986

    EP - 997

    JO - Canadian Journal of Civil Engineering

    T2 - Canadian Journal of Civil Engineering

    JF - Canadian Journal of Civil Engineering

    SN - 0315-1468

    IS - 11

    ER -