Eyal Levenberg

Evaluating and optimizing resilience of airport pavement networks

This paper addresses the problem of assessing and maximizing the resilience of an airports runway and taxiway network under multiple potential damage-meteorological scenarios. The problem is formulated as a stochastic integer program with recourse and an exact solution methodology based on the integer L-shaped decomposition is proposed for its solution. The formulation seeks an optimal allocation of limited resources to response capabilities and preparedness actions that facilitate them. The overall aim is to quickly restore post-event takeoff and landing capacities to pre-event operational levels taking into account operational, budgetary, time, space, and physical resource limitations. Details, such as aircraft size impacts, reductions in capacity due to joint takeoff and landing maneuvers on common runways or bidirectional flows on taxiways, potential for outsourcing repair work, and multi-team response, are incorporated. The mathematical model and solution methodology are embedded within a decision support tool, the capabilities and applicability of which are demonstrated on an illustrative case study. Potential benefits to airport operators are described, including, for example: the tools utility in suggesting equipment to have at the ready, identifying the critical pavement system components, and vulnerabilities for prioritizing future facility developments.

Smoothing Asphalt Concrete Complex Modulus Test Data

A computational approach is offered for smoothing asphalt concrete complex modulus test results in a way that forces compliance with linear viscoelastic theory and thermorheological simplicity. Essentially, it is based on presupposing the shape of the relaxation spectrum with calculations carried out in the wicket domain. The suggested scheme calls for the determination of six free parameters—three of which are associated with the mechanical (viscoelastic) response, while the remaining three are related to the time-temperature shifting properties and include (as unknown) the reference temperature level. In the evaluation process, it is shown how the first three parameters can be obtained directly from the measurements and separately from the other three, while the determination of the latter set follows as a second independent step. This decoupled process simplifies the analysis and includes additional advantages. In this paper, the smoothing approach is described and discussed, and its appropriateness and relevancy for asphalt concrete materials is demonstrated using a sample data set.

Validation of NCAT Structural Test Track Experiment Using INDOT APT Facility

The National Center for Asphalt Technology (NCAT) operates a full-scale test road for studying the response and performance of asphalt pavements. During the 2003 - 2005 testing phase, NCAT instrumented eight of their test sections with stress and strain gauges. Two of the test sections were later replicated, along with embedded instrumentation, for subsequent testing in the accelerated pavement testing (APT) facility operated by the Indiana Department of Transportation. The availability of similarly constructed and instrumented pavement systems loaded in different conditions offered a unique opportunity to develop and test the forecastability of pavement models. Exploring this aspect is the topic of the present work, in which an attempt is made to use the APT experiment in conjunction with laboratory test results, and forecast resilient responses obtained at NCAT that were generated under completely different loading and environmental conditions. The modeling and analysis methodologies are outlined in detail and the calculation results are compared with NCAT measurements. Findings are discussed and recommendations for future research are given.

Quantifying the Confidence Levels of Deformation Measurements in Asphalt Concrete

An experimental approach for measuring surface strains in asphalt concrete (AC) samples is presented in this paper. It is equivalent to the local attachment of a large number of linear variable displacement transducers (LVDTs) with gage lengths ranging from 5 to 80 mm. The strain distribution properties developed under conditions of uniaxial tension are studied and analyzed for one AC mixture. It is shown that gage lengths of the order of the nominal maximum aggregate size measure a wide range of strain values: from zero strain up to 5 times the average strain. Moreover, as the gage length is increased, the scatter is reduced and the individual measurements yield values that are closer to their average. Subsequently, the confidence levels associated with using a limited number of gages with a given gage length are quantified. The results are then used to assess test protocol recommendations related to AC sample instrumentation. The size scale of the materials Representative Volume Element (RVE) is also addressed.

Dynamic backcalculation with different load-time histories

This paper focused attention to the falling weight deflectometer (FWD) load-time history. For a commonly used device, it studied the pulse generation mechanism and the influence of different load histories on backcalculation results. In this connection, a semi-analytic impact theory was first introduced for realistically simulating FWD pulse generation. Then a newly developed finite-element code was presented for FWD interpretation; the code is capable of addressing dynamics, time-dependent layer properties, and quasi-nonlinear behaviour. Both new developments were demonstrated for an experimental dataset that resulted from operating an FWD with different loading configurations. It was found that backcalculated parameters are sensitive to the FWD pulse features. Consequently, it is recommended that, whenever advanced pavement characterisation is sought, experimental attention should be placed on generating diverse FWD pulse histories. Collectively, the resulting deflection histories will contain pertinent constitutive information for supporting the calibration of more complex pavement models.

Viscoelastic genetic algorithm for inverse analysis of asphalt layer properties from falling weight deflections

The falling weight deflectometer (FWD) is a nondestructive test whose results are typically used for backcalculating in situ layer properties of pavements. Most backcalculation methods assume the pavement to be a layered elastic half-space. However, asphalt pavements behave more like multilayered viscoelastic systems, especially in response to small or short-duration load applications. Hence, although elastic analysis is computationally efficient and well accepted in the engineering community, the theory cannot produce the viscoelastic properties of the asphalt concrete (AC) layer. In this study, a new inverse analysis method is proposed to backcalculate both linear elastic and viscoelastic properties of pavement layers as well as the AC time–temperature shift factor. In this method, the FWD load–response history of a single FWD drop and variation in temperature along the depth of the AC layer during the drop are used for performing the computations. The underlying (viscoelastic) forward solver is approximate and disregards dynamic effects; these factors, in return, make it computationally efficient. A genetic algorithm–based optimization scheme is offered to search for the pavement properties. As an example, two sections from the long-term pavement performance study were selected for investigation. The back-calculation results were positive, which indicates that unless a stiff layer exists close to the surface, it should be possible to infer linear viscoelastic properties from a single FWD drop.

Inferring Pavement Properties using an Embedded Accelerometer

A high-end single-axis inertial sensor, capable of detecting minute accelerations, was embedded in an asphalt pavement system. A vehicle of known dimensions and overall weight was driven along a straight line, passing near the sensors embedment location at constant speed. Measured accelerations were analyzed as means to infer the pavement layer properties. For this purpose, instead of double integrating the accelerometer data to arrive at deflections, the raw measurements were first smoothed and then matched against computed accelerations. Vehicle speed and weight distribution between front and rear axles were also backcalculated. The paper describes the experimental details, analysis method, and results.

Strain Measurements in Asphalt Concrete Specimens towards the Development of a Fracture Model

The phenomenology of asphalt concrete (AC) fracture is studied in the laboratory under uniaxial-tension controlled-strain conditions. A specially devised method for measuring local surface strains is described and used to visualize the load-induced damage evolution. It is shown that in pre-peak stress conditions, induced damage is a result of formation, growth, and coalescence of micro-cracks, and that the process is, in essence, three dimensional and random - related to the aggregate packing within the mix. During post-peak stress conditions, it is shown that a process of damage localization occurs within the representative volume element (RVE) leading to the creation of a dominant failure surface (and eventually fracture). The cohesive crack model (CCM) is proposed as an adequate mechanistic fracture model that can account for the above described failure process. A limited testing program was conducted as an initial step towards characterizing the material according to the model. Preliminary results and findings are presented and discussed.

Viscoelastic characterisation of asphalt-aggregate mixes in diametral compression

An approach is offered for extracting linear viscoelastic (VE) properties of asphalt-aggregate mixes in diametral compression. The modelling treats the material both as VE (solid) and viscoplastic (VP), while the suggested scheme aims at inferring ‘pure’ VE properties by removing the interlaced VP effects. The characterisation calls for exposing the specimen to several load-unload-rest cycles, and then matching the measurements in the time-domain with an appropriate forward solver – controlled by a non-linear optimisation algorithm. A case study is presented to demonstrate the required testing protocol and subsequent analysis. Overall, the approach is deemed viable, capable of identifying the materials VE properties over a very broad time interval even when operating under a single temperature level.

Viscoelastic Characterization of Asphalt Concrete in Diametral Tension-Compression

AbstractThis work focuses on improving the linear viscoelastic characterization of asphalt concrete materials with a standard indirect tension setup. Three main aspects distinguish this investigation from typical efforts. First, the applied diametral force history consisted of load–unload–rest sequences; this was done to enable separation between recoverable and irrecoverable deformation components. Second, viscoelastic properties were essentially calibrated against the recoverable deformation part to guarantee agreement with the sought constitutive theory; response during rest intervals was modeled for this purpose, assuming inactivity of the irrecoverable deformation part. Third, diametral forces were alternated between tension and compression; this was done to restrain the accumulation of irrecoverable deformation and to widen the calibration domain. Detailed step-by-step guidelines are included and applied to clarify the approach. For pavement engineering purposes, the overall scheme is deemed an impr...