|<< Previous||Contents||Next >>|
Geotechnical Aspects of Pavements Reference Manual
Appendix D: NCHRP 1-37A Design Method
The Guide for the Mechanistic-Empirical Design of New and Rehabilitated Pavement Structures developed under NCHRP Project 1-37A is the state-of-the-art procedure for the design of flexible and rigid pavement structures. The mechanistic-empirical approach at the heart of the NCHRP 1-37A methodology represents a fundamental paradigm shift for pavement design. In the mechanistic-empirical approach, the response of the pavement - defined in terms of stresses, strains, and other parameters - is analyzed using rigorous theories of mechanics. Critical response quantities - e.g., tensile strains at the bottom of an asphalt or PCC layer - are then related empirically to pavement performance - e.g., fatigue cracking.
Figure D-1 provides a flow chart for the mechanistic-empirical design approach as implemented in the NCHRP 1-37A procedures. The major steps are:
- Define the traffic, environmental, and other general design inputs for the project. In the case of rehabilitation designs, this will also include information on existing pavement conditions (e.g., distress survey, FWD testing).
- Select a trial pavement section for analysis. For rehabilitation designs, this includes identification of an appropriate rehabilitation strategy.
- Define the properties for the materials in the various pavement layers.
- Analyze the pavement response (temperature, moisture, stress, strain) due to traffic loading and environmental influences. The pavement response analysis is performed on a season-by-season basis in order to include variations in traffic loading, environmental conditions, and material behavior over time.
- Empirically relate critical pavement responses to damage and distress for the pavement distresses of interest. Damage/distress are determined on a season-by-season basis and then accumulated over the design life of the pavement.
- Adjust the predicted distresses for the specified design reliability.
- Compare the predicted distresses at the end of design life against design limits. If necessary, adjust the trial pavement section and repeat Steps 3-7 until all predicted distresses are within design limits.
The corresponding major components required to implement this mechanistic-empirical pavement design methodology are:
- Inputs-traffic, climate, materials, others.
- Pavement response models-to compute critical responses.
- Performance models or transfer functions-to predict pavement performance over the design life.
- Design reliability and variability-to add a margin of safety for the design.
- Performance criteria-to set objective distress limits against which the pavement performance will be judged.
- Software-to implement the mechanistic-empirical models and calculations in a usable form.
Figure D-1. Flow chart for mechanistic-empirical design methodology.
Each of these components will be briefly summarized in the following sections. Readers should refer to the NCHRP 1-37A final reports (NCHRP 1-37A, 2004) for more thorough coverage of each topic. In addition, Chapter 5 provides detailed information on the geotechnical inputs to the NCHRP 1-37A procedure and Chapter 6 gives several example applications.
D.2.1 Hierarchical Inputs
As described in Chapter 5, the NCHRP 1-37A design methodology incorporates a hierarchical approach for specifying all pavement design inputs. The hierarchical approach is based on the philosophy that the level of engineering effort exerted in determining design inputs should be commensurate with the relative importance, size, and cost of the design project. Three levels are provided for the design inputs in the NCHRP 1-37A procedure:
Level 1 inputs provide the highest level of accuracy and the lowest level of uncertainty. Level 1 design inputs would typically be used for heavily trafficked pavements or whenever there are serious safety or economic consequences of early failure. Level 1 material inputs require field or laboratory evaluation. Subgrade resilient modulus measured from FWD testing in the field or triaxial testing in the laboratory is one example of a Level 1 input.
Level 2 inputs provide an intermediate level of accuracy and are closest to the typical procedures used with the AASHTO Design Guides. This level could be used when resources or testing equipment are not available for Level 1 characterization. Level 2 inputs would typically be derived from a limited testing program or estimated via correlations or experience (possibly from an agency database). Subgrade resilient modulus estimated from correlations with measured CBR values is one example of a Level 2 input.
Level 3 inputs provide the lowest level of accuracy. This level might be used for designs in which there are minimal consequences of early failure (e.g., low volume roads). Level 3 material inputs typically are default values that are based on local agency experience. A default subgrade resilient modulus based on AASHTO soil class is an example of a Level 3 input.
Any given pavement design may incorporate a mix of input data of different levels. For example, measured HMA dynamic modulus values used with default resilient modulus values for the unbound materials in the pavement structure. However, the algorithms used in the design computations are identical for all input levels. In other words, the NCHRP 1-37 methodology features levels of input data but not levels of design analysis. The composite input level determines the overall accuracy and reliability of the pavement performance predictions used to judge the acceptability of a trial design.
Traffic data are key inputs for the analysis and design of pavement structures. Most existing design procedures, including all of the AASHTO Design Guides, quantify traffic in terms of equivalent single axle loads (ESALs). However, the mechanistic pavement response models in the NCHRP 1-37A methodology require the specification of the magnitudes and frequencies of the actual wheel loads that the pavement is expected to see over its design life. Consequently, traffic must be specified in terms of load spectra rather than ESALs. Load spectra are the frequency distributions of axle load magnitudes by axle configuration (single, tandem, tridem, quad) and season of year (monthly, typically).
State highway agencies typically collect two categories of traffic data. Weigh-in-motion (WIM) data provide information about the number and configuration of axles observed within a set of load groups. Automatic vehicle classification (AVC) data provide information about the number and types of vehicles that use a given roadway as counted over some period of time. Error! Reference source not found. summarizes the WIM and AVC data that are required at each of the hierarchical input levels in the NCHRP 1-37A methodology.
The traffic data required in the NCHRP 137A methodology are the same for all pavement types (flexible or rigid) and construction types (new or rehabilitated). Four categories of traffic data are required:
- Traffic volume-base year information
- Two-way annual average daily truck traffic (AADTT)
- Number of lanes in the design direction
- Percent trucks in design direction
- Percent trucks in design lane
- Vehicle (truck) operational speed
- Traffic volume adjustment factors
- Monthly adjustment
- Vehicle class distribution (see Table 6-5 for an example)
- Hourly truck distribution (see Table 6-6 for an example)
- Traffic growth factors
- Axle load distribution factors by season, vehicle class, and axle type (see Table 6-7 for an example)
- General traffic inputs
- Traffic wander data (mean wheel location and standard deviation of lateral wander; lane width)
- Number axles/trucks (see Table 6-8 for an example)
- Axle configuration (axle width and spacing; tire spacing and pressure)
- Wheelbase spacing distribution (rigid pavements only; see Table 6-11 for an example)
The NCHRP 1-37A design software takes all of these traffic inputs and computes the number of applications of each axle load magnitude by axle type (single, tandem, tridem, quad) and month. These axle load spectra are a primary input to the mechanistic pavement structural response models.
|Data Sources||Input Level|
|Traffic load / volume data||WIM data - site/segment specific||X|
|WIM data - regional default summaries||X|
|WIM data - national default summaries||X|
|AVC data - site/segment specific||X|
|AVC data - regional default summaries||X|
|AVC data - national default summaries||X|
|Vehicle counts - site/segment specific1||X||X|
|Traffic forecasting and trip generation models2||X||X||X|
- Level depends on whether regional or national default values are used for the WIM or AVC information.
- Level depends on input data and model accuracy/reliability.
Environmental conditions have a significant effect on the performance of both flexible and rigid pavements. External factors such as precipitation, temperature, freeze-thaw cycles, and depth to water table play key roles in defining the impact of environment on pavement performance. Internal factors such as the susceptibility of the pavement materials to moisture and freeze-thaw damage, drainability of the paving layers, and infiltration potential of the pavement define the extent to which the pavement will react to the external environmental conditions.
Variations in temperature and moisture profiles within the pavement structure and subgrade over the design life of a pavement are simulated in the NCHRP 1-37A design methodology via the Enhanced Integrated Climatic Model (EICM-described more fully in Section D.3.1). The EICM requires a relatively large number of input parameters. As with all other design inputs, EICM input parameters are specified using a hierarchical approach (Levels 1, 2, or 3). Since many of the EICM material property inputs are not commonly measured by most agency and geotechnical laboratories, Level 3 default values will typically be used for most designs. The inputs required by the EICM fall under the following broad categories (see Sections 5.5.2 and 5.6.2 for more detail):
- General information
- Base/subgrade construction completion date
- Pavement construction date
- Traffic opening date
- Weather-related information (Section 5.6.2)
- Hourly air temperature
- Hourly precipitation
- Hourly wind speed
- Hourly percentage sunshine (used to determine cloud cover)
- Hourly relative humidity
- Groundwater related information (Section 5.6.2)
- Groundwater table depth
- Drainage and surface properties (Section 5.5.2)
- Surface shortwave absorptivity (Section 5.6.2)
- Drainage path length
- Cross slope
- Pavement materials
- Asphalt and Portland cement concrete
- Thermal conductivity
- Heat capacity
- Unbound materials (Section 5.5.2)
- Physical properties (specific gravity, maximum dry unit weight, optimum moisture content)
- Soil water characteristic curve
- Hydraulic conductivity (permeability)
- Thermal conductivity
- Heat capacity
The weather-related information required by the EICM can be obtained from weather stations located near the project site. The software accompanying the NCHRP 1-37A Design Guide includes a database from nearly 800 weather stations throughout the United States that can be used to generate the weather-related design inputs.
D.2.4 Material Properties
The material property inputs required for the environmental effects model in the NCHRP 1-37A methodology have already been described in Section D.2.3 (and Sections 5.5.2 and 5.6.2). Additional material property inputs are required for the structural response models used to calculate the stresses and strains in the pavement. As with all other design inputs, the material property inputs can be provided at any of the hierarchical Levels (1, 2, or 3). The material property inputs are most conveniently grouped by material type:
- Asphalt concrete
- Layer thickness
- Dynamic modulus (measured value for level 1 or mixture gradation and volumetrics for Level 2 and 3 estimation)
- Asphalt binder properties (dynamic shear stiffness or viscosity for Levels 1 and 2, binder grade for Level 3)
- Mixture volumetrics (effective binder content, air voids, unit weight)
- Poisson's ratio
- Thermal cracking properties (low temperature tensile strength, creep compliance, thermal expansion coefficient)
- Portland cement concrete
- Layer thickness
- Mixture properties (cement and aggregate type, cement content, water/cement ratio, unit weight)
- Shrinkage characteristics
- Elastic modulus
- Poisson's ratio
- Compressive strength
- Modulus of rupture
- Thermal expansion coefficient
- Unbound materials (see Sections 5.3 and 5.4 for more details)
- Material type
- Layer thickness
- Unit weight
- Coefficient of lateral earth pressure
- Resilient modulus (see Section 5.4.3 for details on inputs at different hierarchical levels)
- Poisson's ratio
A variety of other input data are required for the NCHRP 1-37A methodology. Some of these inputs are dependent upon the particular pavement type (flexible vs. rigid) and construction type (new vs. rehabilitation) being considered. A brief summary of these other inputs are as follows:
- General project information
- Design life
- Latitude, longitude, and elevation (for accessing weather station database)
- Rigid pavement design features (all rigid pavement types)
- Permanent curl/warp effective temperature difference
- Base erodibility index
- JPCP design features
- Joint spacing, sealant type
- Dowel bar diameter, spacing
- Edge support (e.g., tied shoulder, widened slab)
- PCC-base interface bond condition
- CRCP design features
- Shoulder type
- Reinforcement (steel percentage, diameter, depth)
- Mean crack spacing
- Flexible pavement distress potential (new construction)
- Block cracking
- Longitudinal cracks outside wheel paths
- Pre-rehabilitation distresses (overlay over AC surface)
- Fatigue cracking within wheel path
- Longitudinal cracks outside wheel path
- Pre-rehabilitation distresses (overlay over PCC surface)
- Percent cracked slabs before, after restoration
- CRCP punchouts
- Dynamic modulus of subgrade reaction
Note that no design features are included for jointed reinforced concrete pavements (JRCP). The NCHRP 1-37A methodology does not include a design capability for this pavement type.
D.3 Pavement Response Models
There are two types of pavement response models in the NCHRP 1-37A methodology: (a) an environmental effects model for simulating the time- and depth-dependent temperature and moisture conditions in the pavement structure in response to climatic conditions; and (b) structural response models for determining the stresses and strains at critical locations in the pavement structure in response to traffic loads. The same environmental effects model is used for all pavement types. Different structural response models are employed for rigid vs. flexible pavements because of the fundamental differences in their mechanical behavior.
D.3.1 Environmental Effects
Diurnal and seasonal fluctuations in the moisture and temperature profiles in the pavement structure induced by changes in groundwater table, precipitation/infiltration, freeze-thaw cycles, and other external factors are incorporated in the NCHRP 1-37A design methodology via the Enhanced Integrated Climatic Model (EICM). The EICM is a mechanistic one-dimensional coupled heat and moisture flow analysis that simulates changes in the behavior and characteristics of pavement and subgrade materials induced by environmental factors. The EICM consists of three major components:
- The Climatic-Materials-Structural Model (CMS Model) originally developed at the University of Illinois (Dempsey et al., 1985).
- The CRREL Frost Heave and Thaw Settlement Model (CRREL Model) originally developed at the United States Army Cold Regions Research and Engineering Laboratory (CRREL) (Guymon et al., 1986).
- The Infiltration and Drainage Model (ID Model) originally developed at Texas A&M University (Lytton et al., 1990).
Each of these components has been enhanced substantially for use in the NCHRP 1-37A design methodology.
For flexible pavements, the EICM evaluates the following environmental effects:
- Seasonal changes in moisture content for all subgrade and unbound materials.
- Changes in resilient modulus, MR, of all subgrade and unbound materials caused by changes in soil moisture content.
- Changes MR due to freezing and subsequent thawing and recovery from frozen conditions.
- Temperature distributions in bound asphalt concrete layers (for determining the temperature-dependent asphalt concrete material properties).
For rigid pavements, the following additional environmental effects are simulated by the EICM:
- Temperature profiles in PCC slabs (for thermal curling prediction).
- Mean monthly relative humidity values (for estimating moisture warping PCC slabs).
One of the important outputs from the EICM for both flexible and rigid pavement design is a set of adjustment factors for unbound layer materials that account for the effects of environmental conditions such as moisture content changes, freezing, thawing, and recovery from thawing. This factor, denoted Fenv, varies with position within the pavement structure and with time throughout the analysis period. The Fenv factor modifies the resilient modulus at optimum moisture and density conditions MRopt to obtain the seasonally adjusted resilient modulus MR as a function of depth and time.
D.3.2 Structural Response
The mechanistic structural response models determine the stresses, strains, and displacements within the pavement system caused by traffic loads and as influenced by environmental conditions. Environmental influences may be direct (e.g., strains due to thermal expansion and/or contraction) or indirect (e.g., changes in material properties due to temperature and/or moisture effects).
Two flexible pavement analysis methods have been implemented in the NCHRP 1-37A computational procedures. For cases in which all materials in the pavement structure can realistically be treated as linearly elastic, multilayer elastic theory (MLET) is used to determine the pavement response. MLET provides an excellent combination of analysis capabilities, theoretical rigor, and computational speed for linear pavement analyses. In cases where the consideration of unbound material nonlinearity is desired (i.e., Level 1 resilient modulus for new construction), a nonlinear finite element (FE) methodology is employed instead for determining the pavement stresses, strains, and displacements.
A major advantage of MLET solutions is very quick computation times. Solutions for multiple wheel loads can be constructed from the fundamental axisymmetric single wheel solutions via superposition automatically by the computer program. The principal disadvantage of MLET is its restriction to linearly elastic material behavior. Real pavement materials, and unbound materials, in particular, often exhibit stress-dependent stiffness. The materials may even reach a failure condition in some locations, such as in tension at the bottom of the unbound base layer in some pavement structures. These nonlinearities vary both vertically through the thickness of the layer and horizontally within the layer. Some attempts have been made in the past to incorporate these material nonlinearity effects into MLET solutions in an approximate way, but the fundamental axisymmetric formulation of MLET makes it impossible to include the spatial variation of stiffness in a realistic manner.
Some of the limitations of MLET solutions are the strengths of FE analysis. In particular, finite element methods can simulate a wide variety of nonlinear material behavior; the underlying finite element formulation is not constrained to linear elasticity, as is the case with MLET. Stress-dependent stiffness and no-tension conditions for unbound materials can all be treated within the finite element framework. However, the FE computational times are substantially longer than for MLET analyses.
The choice of MLET vs. FE structural response model is made automatically by the NCHRP 1-37A software based on the input data from the user (i.e., whether Level 1 new construction inputs are specified for the unbound resilient modulus values). In both cases, the NCHRP 1-37A software automatically pre-processes all of the input data required for the analysis (e.g., automatically generates a finite element mesh), automatically performs the season-by-season analyses over the specified pavement design life, and automatically post-processes all of the analysis output data to compute the season-by-season values of the critical pavement responses for subsequent use in the empirical performance prediction models.
Performance prediction requires identification of the locations in the pavement structure where the critical pavement responses (stress or strain) attain their most extreme values. For multilayer flexible pavement systems, these locations can be difficult to determine. Critical responses are evaluated at several depth locations in the NCHRP 1-37A analyses, depending upon the distress type:
- Fatigue Depth Locations:
- Surface of the pavement (z = 0),
- 0.5 inches from the surface (z = 0.5),
- Bottom of each bound or stabilized layer.
- Rutting Depth Locations:
- Mid-depth of each layer/sub-layer,
- Top of the subgrade,
- Six inches below the top of the subgrade.
The horizontal locations for the extreme values of critical responses are more difficult to determine. The critical location for the simplest case of a single wheel load can usually be determined by inspection - e.g., directly beneath the center of the wheel. The critical location under multiple wheels and/or axles will be a function of the wheel load configuration and the pavement structure. Mixed traffic conditions (single plus multiple wheel/axle vehicle types) further complicate the problem, as the critical location within the pavement structure will not generally be the same for all vehicle types. The NCHRP 1-37A calculations address this problem by evaluating the pavement responses for a set of potential critical locations. Damage/distress magnitudes are calculated from the pavement responses at each location, with the final performance prediction based on the location having the maximum damage/distress at the end of the analysis period.
Finite element analysis has been proven a reliable tool for computing rigid pavement structural responses. However, the season-by-season distress/damage calculations implemented in the NCHRP 1-37A procedure requires hundreds of thousands of calculations to compute incremental damage over a design period of many years. These computations would take days to complete using existing rigid pavement finite element programs. To reduce computer time to a practical level, neural network models have been developed from a large parametric study performed using the ISLAB2000 finite element program (Khazanovich et al., 2000). The neural network models, which, in effect, are similar to regression models, make it possible to accurately compute critical stresses and deflections virtually instantaneously. This in turn makes it possible to perform detailed month-by-month incremental analysis within a practical timeframe (i.e., a few minutes). Appendix QQ in the NCHRP 1-37A final report (NCHRP, 2004) provides a detailed description of the finite element models, parametric study, and neural networks used for the structural analysis of rigid pavements.
A key feature of the rigid pavement structural response model is its treatment of the pavement foundation. The ISLAB2000 analysis program and the neural network models derived from it employ a modified version of the conventional slab-on-Winkler springs pavement structural model (also called a "dense liquid" foundation model). As shown in Figure D-2, the actual multi-layer pavement structure is replaced by an equivalent 2-layer (slab and base) pavement section resting on a Winkler spring foundation having a stiffness characterized by k, the modulus of subgrade reaction (see Section 5.4.6). The effective k value in the equivalent 2-layer pavement is determined by matching the computed surface deflections for the actual multi-layer pavement section. The surface deflection profile of the actual section is determined using MLET, modeling all layers in the structure. This computed deflection profile is then used to backcalculate the effective k value for the equivalent 2-layer section. Thus, the effective k value is an internally computed value, not a direct input to the design procedure. The exception to this is rehabilitation design, where k determined from FWD testing may be input directly.
Figure D-2. Structural model for rigid pavement structural response computations.
The effective k value used in the NCHRP 1-37A methodology is interpreted as a dynamic k value (e.g., as determined from FWD testing), which should be distinguished from the traditional static k values used in previous AASHTO design procedures.
D.4 Pavement Performance Models
Pavement performance is evaluated in terms of individual distress modes in the NCHRP 1-37A methodology. A variety of empirical distress models - also sometimes termed "transfer functions" - are incorporated in the NCHRP 1-37A methodology for the major structural distresses in flexible and rigid pavements. Empirical models are also provided for estimating smoothness as a function of the individual structural distresses and other factors.
D.4.1 Damage vs. Distress
Some distresses can be evaluated directly during the season-by-season calculations. For example, the empirical model for rutting in the asphalt layers in flexible pavements is of the form:(D.1)
|εp||= βr1 a1 Ta2βr2 Na3βr3|
|εp||=||accumulated plastic strain after N repetitions of load at the critical location|
|εr||=||resilient strain at the critical location|
|N||=||number of load repetitions|
|a1||=||regression coefficients derived from laboratory repeated load permanent deformation tests|
|βr1||=||field calibration coefficients (see Section D.4.4)|
Each asphalt layer is divided into sublayers, and Eq. (D.1) is evaluated at the midthickness of each sublayer. The contribution Δdi to total rutting Rd from sublayer i having thickness hi can then be expressed as:(D.2)
ΔRdi = εpi Δhi
The contributions of all of the sublayers l can then be summed to give the total rutting for the asphalt concrete layer:(D.3)
Other distresses cannot be evaluated directly, but must be quantified in terms of computed damage factors. For example, the empirical model for "alligator" fatigue cracking in the asphalt layers in flexible pavements is of the form:(D.4)
Nf = βf1 k1 ( εt )βf2k2 ( E )-βf3k3
|Nf||=||number of repetitions to fatigue cracking failure|
|εt||=||tensile strain at the critical location|
|E||=||asphalt concrete stiffness (at appropriate temperature)|
|k1, k2, k3||=||regression coefficients determined from laboratory fatigue tests|
|βf1, βf2, βf3||=||field calibration coefficients (see Section D.4.4)|
Computation of fatigue damage is based upon Miner's Law:(D.5)
|T||=||total number of seasonal periods|
|ni||=||actual traffic for period i|
|Nfi||=||traffic repetitions causing fatigue failure under conditions prevailing during period i|
The damage factor determined using Eq. (D.5) is then related to observed fatigue distress quantities (e.g., area of fatigue cracking within the lane) during the field calibration process (Section D.4.4).
D.4.2 Distress Models
Empirical distress prediction models are provided for the following structural distresses in the NCHRP 1-37A flexible pavement design methodology:
- Permanent deformation (rutting)
- Within asphalt concrete layers
- Within unbound base and subbase layers
- Within the subgrade
- Fatigue cracking
- Within asphalt concrete layers
- Bottom-up (classical "alligator" cracking)
- Top-down (longitudinal fatigue cracking)
- Within cement stabilized layers
- Thermal cracking
The empirical structural distress models for rigid pavements include:
- Transverse joint faulting (JPCP)
- Transverse fatigue cracking (JPCP)
- Punchouts (CRCP)
Note that reflection cracking for asphalt concrete overlays is not included in the current version of the NCRHP 1-37A methodology. At the time of the NCHRP 1-37A development, it was judged that no suitable empirical reflection cracking models yet existed. It is anticipated that a suitable model will be developed and added to the NCHRP 1-37A procedure in the future.
Pavement smoothness is often used as a composite index of pavement quality. Smoothness (or loss thereof) is influenced by nearly all of the distresses of interest in flexible and rigid pavement systems. Smoothness data is also regularly and routinely collected and stored as part of the pavement management systems at many agencies. Lastly, smoothness is directly related to overall ride quality, the factor of most importance to highway users. Because of these reasons, empirical smoothness prediction models have been incorporated in the NCHRP 1-37A design methodology.
Pavement smoothness in the NCHRP 1-37A models is characterized in terms of the International Roughness Index, or IRI. IRI is predicted as a function of the initial as-constructed IRI, the subsequent development of distresses over time, and other factors such as subgrade type and climatic conditions that may affect smoothness through mechanisms such as shrinkage or swelling of subgrade soils and frost heave. The structural distresses influencing smoothness are predicted directly by the NCHRP 1-37A mechanistic-empirical methodology. However, nonstructural distresses cannot be evaluated using mechanistic-empirical principles, so the NCHRP 1-37A procedure provides the option of specifying the overall potential for these other distresses. Smoothness loss due to soil shrinking/swelling/frost heave and other climatic factors are incorporated into the NCHRP 1-37A IRI models through the use of a "site factor."
The NCHRP 1-37A design method provides IRI prediction models as a function of pavement type (flexible vs. rigid), base type (flexible pavements), and construction type (new vs. rehabilitation). IRI models are provided for the following cases:
- AC (new construction)
- AC over granular base
- AC over asphalt-treated base
- AC over cement-stabilized material
- AC overlay (rehabilitation)
- AC over flexible pavement
- AC over rigid pavement
- JPCP (new construction)
- JPCP (rehabilitation)
- JPCP restoration
- Bonded PCC over JPCP
- Unbonded PCC over JPCP
- CRCP (new construction)
- CRCP (rehabilitation)
- CRCP restoration
- Bonded PCC over JPCP
- Unbonded PCC over CRCP (rehabilitation)
Appendix OO in the NCHRP 1-37A final documentation (NCHRP, 2004) provides a detailed description of the development of these models.
D.4.4 Field Calibration
The distress prediction models are key components of the NCHRP 1-37A mechanistic-empirical design and analysis procedure. Calibration of these models against field performance is an essential part of the model development. Calibration refers to the mathematical process by which the models are adjusted to minimize the differences between predicted and observed values of distress.
All performance models in the NCHRP 1-37A design method have been calibrated on a global level to observed field performance at a representative set of pavement test sites around North America. Test sections from the FHWA Long Term Pavement Performance (LTPP) program were used extensively in the calibration process because of the consistency of the monitored data over time and the diversity of test sections throughout North America.
However, there were some serious limitations to the NCHRP 1-37A field calibration. Many of the material property and site feature inputs required for the NCHRP 1-37A analyses were unavailable from the LTPP database. Because of the limited number of pavement test sites with complete input data, the minimal material testing available, the use of calculated properties from correlations (i.e., Level 3 inputs), and the global scope of the calibration effort, the predictions from the calibrated models still have relatively high levels of uncertainty and a limited inference space of application. The recently completed NCHRP Project 9-30 (Von Quintus et al., 2003) has formulated a plan for developing an enhanced database for future recalibration of the NCHRP 1-37A and other similar pavement models.
The NCHRP 1-37A software also includes a provision for entering local or regional field calibration factors instead of the national values derived from the LTPP database. This feature permits local agencies to adjust the mechanistic-empirical performance predictions to better reflect their local conditions.
D.5 Design Reliability
A large amount of uncertainty and variability exists in pavement design and construction, as well as in the traffic loads and climatic factors acting over the design life. In the NCHRP 1-37A mechanistic-empirical design, the key outputs of interest are the individual distress quantities. Therefore, variability of the predicted distresses is the focus of design reliability.
The incorporation of reliability in the NCHRP 1-37A procedure is similar in some respects to the way it is treated in the 1993 AASHTO Guide. In the 1993 AASHTO Guide, an overall standard deviation or "uncertainty" is specified for the design inputs (the S0 value-see Appendix C), a desired reliability level is selected based on agency policy, and the combination of the standard deviation and reliability are then used in essence to add a "margin of safety" to the design traffic W18. The NCHRP 1-37A methodology differs from the 1993 AASHTO procedure in that the standard deviations and reliability levels are set for each individual distress mode predicted in the mechanistic-empirical computations. The default value for the standard deviation of each predicted distress quantity is based on a careful analysis of the differences between the predicted versus actual distresses during the field calibration of the empirical performance models (Section D.4.4). These estimates of error represent the combined effects of input variability, variability in the construction process, and model error.
The desired level of reliability is specified along with the acceptable level of distress at the end of design life (Section D.6) to define the performance requirements for a pavement design in the NCHRP 1-37A procedure. For example, one criterion might be to limit the percent of cracked PCC slabs to 8% at a design reliability of 90%. Then, on average for 100 projects, 90 would be expected to exhibit fewer than 8% slabs cracked at the end of the design life. Different reliability levels may be specified for different distresses in the same design. For example, the designer may choose to specify 95% reliability for slab cracking, but 90% reliability for faulting and IRI. Of course, increasing design reliability will lead to more substantial pavement sections and higher initial costs. The beneficial trade-off is that future maintenance costs should be lower for the higher-reliability design.
D.6 Performance Criteria
Performance criteria are definitions of the maximum amounts of individual distress or smoothness acceptable to an agency at a given reliability level. Performance criteria are a user input in the NCHRP 1-37A methodology and depend on local design and rehabilitation policies. Default performance criteria built into the current version of the NCHRP 1-37A software are summarized in Table D-2. The designer can select all or some subset of the performance criteria to be evaluated during the design.
|Top-down (longitudinal) fatigue cracking||feet/mile||1000|
|Bottom-up (alligator) fatigue cracking||% of wheel path area||25|
|Chemically stabilized layer fatigue cracking||% of wheel path area||25|
|Total permanent deformation (rutting)||inch||0.75|
|Permanent deformation (rutting) in asphalt layer||inch||0.25|
|Transverse fatigue cracking (JPCP)||% slabs cracked||15|
|Mean joint faulting (JPCP)||inch||0.12|
|Punchouts (CRCP)||number per mile||10|
- Default value from software version 0.700 (4/7/2004).
- Default initial IRI = 63 inches/mile.
The mechanistic-empirical calculations in the NCHRP 1-37A design methodology cannot be performed by hand or simple spreadsheets. A Windows-based program has been developed to implement the NCHRP 1-37A methodology by providing: (1) an interface to input all design variables, (2) computational engines for analysis and performance prediction, and (3) results and outputs from the analyses in formats suitable for use in electronic documents or for making hard copies.
The software presents a series of information and input screens coordinated through a main program layout screen, as illustrated in Figure D-3. On this screen, all access points to the information and data input screens are color-coded to guide the designer in providing all data needed to run a design analysis. Green tags indicate screens on which the designer has already entered/reviewed data, yellow tags indicate screens containing default data that have not yet been reviewed/approved by the designer, and red tags indicate screens that have missing required data that must still be entered by the designer before the calculations can be performed. Clicking on any tag brings up the corresponding data input screen; for example, Figure D-4 shows an example data entry screen for subgrade material properties.
Figure D-3. Main input screen for NCHRP 1-37A software.
Figure D-4. Typical data entry screen for NCHRP 1-37A software.
The main program layout screen provides access to the following five groupings of information and input screens (screens are denoted by the symbol "♣", subordinate screen tabs by the symbol "♠"):
- Project Information
- ♣ General Information
- ♣ Site/Project Identification
- ♣ Analysis Parameters
- Traffic Inputs
- ♣ Traffic Volume Adjustment Factors
- ♠ Monthly Adjustment
- ♠ Vehicle Class Distribution
- ♠ Hourly Distribution
- ♠ Traffic Growth Factors
- ♣ Axle Load Distribution Factors
- ♣ General Traffic Inputs
- ♠ Number of Axles/Truck
- ♠ Axle Configuration
- ♠ Wheelbase
- Climate Inputs
- ♣ Climate
- Structure Inputs
- ♣ Structure
- ♠ Drainage and Surface Properties
- ♠ Layers
- ♣ Layer Material Properties
- ♠ Thermal Cracking
- Distress Potential
Note that the Structure Inputs listing above is for the case of a new flexible pavement design. The screens will be slightly different for other pavement and construction types, but they all conform to the general organization listed above.
Once all necessary information and input data have been entered into the program, the user clicks the Run Analysis button to carry out all the required computations. Separate areas of the main program layout screen provide (1) the status (% complete) of the analyses in progress and (2) links to summary screens for the inputs to the analyses and their results in both tabular and graphical formats. For example, the design analysis of a conventional flexible pavement design might provide output plots of HMA modulus, alligator cracking, thermal cracking, rutting, and IRI versus pavement age. Figure D-5 is an example of the type of output generated by the software. Output can be generated as either Microsoft Excel spreadsheets or as HTML documents for easy import into other engineering applications.
Figure D-5. Typical graphical output from NCHRP 1-37A software.
- Dempsey, B.J., Herlach, W.A. and Patel, A.J. (1985). The Climatic-Material-Structural Pavement Analysis Program, FHWA/RD-84/115, Vol. 3, Final Report, Federal Highway Administration, Washington D.C.
- Guymon, G.L., Berg, R.L. and Johnson, T.C. (1986). Mathematical Model of Frost Heave and Thaw Settlement in Pavement, Report: U.S. Army Cold Region Research and Engineering Laboratory, Hanover, NH.
- Khazanovich, L., Yu, H.T., Rao, S., Galasova, K., Shats, E., and Jones, R. (2000). ISLAB2000 - Finite Element Analysis Program for Rigid and Composite Pavements. User's Guide. ERES Consultants Division of Applied Research Associates, Inc., Champaign, IL.
- Lytton, R.L., Pufahl, D.E., Michalak, C.H., Liang, H.S. and Dempsey, B.J. (1990). An Integrated Model of the Climatic Effects on Pavement, Texas Transportation Institute, Texas A&M University, Report No. FHWA-RD-90-033, Federal Highway Administration, McLean, VA.
- NCHRP 1-37A (2004). Mechanistic-Empirical Design of New and Rehabilitated Pavement Structures, Draft Report, Transportation Research Board, National Research Council, Washington, D.C.
- Von Quintus, H.L., Schwartz, C.W., McCuen, R.H. and Andrei D. (2003). "Experimental Plan for Calibration and Validation of Hot Mix Asphalt Performance Models for Mix and Structural Design," Final Report, NCHRP Project 9-30, Transportation Research Board, National Research Council, Washington, D.C.
|<< Previous||Contents||Next >>|