DARPA INITIATIVE IN PROGNOSIS

PROGNET-MD: AN INTEGRATED SYSTEM FOR ON-LINE MONITORING AND PROGNOSIS OF

TRANSPORTATION INFRASTRUCTURE FOR BATTLE MANAGEMENT AND HOMELAND

SECURITY

A joint Maryland Transportation Initiative and Protective Technology Center project proposal

EXECUTIVE SUMMARY

A multidisciplinary research team of experts drawn from several centers at the University of

Maryland, under the aegis of the Maryland Transportation Initiative (MTI), has teamed up with

the Protective Technology Center (PTC) at the Pennsylvania State University, to respond to

DARPA’s initiative in Prognosis. The team will develop and demonstrate an automated and

integrated prognosis operational decision system for on-line monitoring of transportation

infrastructure systems for use in both battle management and homeland security protection. This

effort will focus on critical components of the transportation infrastructure systems, including

bridges, tunnels, airport runways, ports/harbors, and vital highway links. While initially focused

on transportation infrastructure elements, the prognosis system can be expanded to include other

critical physical assets such as medical or supply logistics facilities, as well as weapon sites in

both enemy and home territory. The system will rely on sensor technology embedded in or

adaptively delivered onto selected components of the transportation infrastructure system, as part

of configurable sensor networks with hybrid wireless and wireline connectivity, as well as

satellite imagery and other video technology. The sensor technology can be employed to provide

instantaneous data on the state of the individual components. Satellite and other video

capabilities, on the other hand, can provide a system-wide view of the entire system, including

existing connections between components and possible location for new connections should any

component be removed from the rest of the system.

The target of the MTI-PTC development is an integrated and automated prognosis system

comprised of monitoring, forecasting, interpretation, reasoning and real-time decision-support

tools and techniques. The prognosis system will exploit multiple sources of information, with

varying degrees of reliability, accuracy, completeness and recency, in real time or near-real time,

to assess and predict structural and/or component performance, load carrying capacity and

reliability of individual key components of the transportation infrastructure, as well as to

1

evaluate the performance, reliability and other pertinent attributes of the overall transportation

system given the current and predicted states of its components.

These capabilities will (1) enable on-line bomb damage assessment (BDA) and target

vulnerability assessment (TVA), (2) provide updated information for both tactical and strategic

operational decision-making to adaptively manage and deploy available resources within a

region (e.g., in positioning troops and supplies in a battle or in developing evacuation plans for a

civilian area under attack), and (3) support immediate response to an attack to facilitate quick

recovery from the incurred damage. In addition to its on-line capabilities, off-line uses of the

system may include a complementary role to scheduled inspection and maintenance procedures,

officer training in battle management, and critical assessment of transportation infrastructure

systems for planned operations.

The focus of the MTI-PTC effort will be on the transportation infrastructure in a first phase,

with two specific target components: bridges, and airfield pavements (as well as critical highway

pavements). The objectives of the proposed effort include development of the following

methodological capabilities in support of this goal:

1. Tools and techniques for providing continuous and instantaneous information on the state of

the transportation-infrastructure system and its constituent components based on sensor data,

imaging data, data fusion algorithms and supporting numerical simulations, with data

processing in both the time and frequency domains.

2. Automated processes for recognizing symptoms of damage and diagnosing its severity;

inferring its cause(s) by mapping observed symptoms to possible failure modes; determining

current structural condition and potential unrealized damage as existing loads are

redistributed throughout the structure; and predicting future structural performance, load

carrying capacity, and reliability, considering their probabilistic evolution over time.

3. A multidimensional network-based platform and associated optimization techniques for

integrating information on the current and predicted states of critical system components (and

of the overall transportation system), to enable BDA/TVA assessments and aid in making

dynamically updated strategic and tactical operational decisions that consider precedence

relationships and the probabilistic, dynamic nature of information from a variety of sources.

4. A real-time distributed decision support system that is modular in design, defining a

sustainable prognosis system that can incorporate new models and methodologies.

2

PROGNET-MD: AN INTEGRATED SYSTEM FOR ON-LINE MONITORING AND PROGNOSIS OF

TRANSPORTATION INFRASTRUCTURE FOR BATTLE MANAGEMENT AND HOMELAND

SECURITY

A joint Maryland Transportation Initiative and Protective Technology Center project proposal

PROBLEM STATEMENT AND OVERVIEW OF APPROACH

Modern warfare and homeland security require not only current information on the state of the

various assets that affect battle conduct, but the ability to use this information to quickly

determine the capacity of those assets to perform mission-critical functions, as well as to

optimize the utilization of these and other available resources to achieve the objectives of the

mission at hand. Significant and continuing developments in (1) sensor technologies, sensor

networks, and communications, (2) methodologies for on-line decision systems and intelligent

control, alongside (3) better fundamental understanding of the mechanisms underlying the failure

behavior and properties of materials and physical structures under extreme loads (such as

explosive blasts), can now be tapped to enable breakthrough applications in battle damage

prediction and on-line tactical decision-making in battle management and homeland security.

To achieve the potential of these technologies for on-line prognosis of critical asset

capacity and reliability, knowledge from several disciplinary areas must be brought together to

create new integrative concepts and tools. A cross-disciplinary team of internationally

recognized experts from key specialty areas at the University of Maryland, under the aegis of the

Maryland Transportation Initiative (MTI), is partnered with the Protective Technology Center

(PTC) at the Pennsylvania State University to develop an automated and integrated prognosis

operational decision system for on-line monitoring of transportation infrastructure systems for

use in battle management and homeland security protection. The MTI-PTC effort will focus on

critical components of transportation infrastructure systems, including bridges, tunnels, airport

runways, ports/harbors, and vital highway links. While initially focused on transportation

infrastructure elements, the prognosis system can be expanded to include other critical physical

assets such as medical or logistical supply facilities, as well as weapon sites in both enemy and

home territory. The system will rely on sensor technology embedded in or adaptively delivered

onto selected components of the transportation infrastructure, as part of configurable sensor

networks with hybrid wireless and wireline connectivity, as well as satellite imagery and other

3

video transmission technology. The sensors will provide instantaneous data on the state of

individual components, while a more comprehensive system view could be obtained from

satellite and other video capabilities. The system will also assist military tacticians and field

operational command in deciding where, how and what type of additional sensors to deploy in

locations where additional information may be required to reduce uncertainty levels to

acceptable thresholds for the operational purpose at hand.

The target of the MTI-PTC development is an integrated and automated prognosis system

comprised of monitoring, forecasting, interpretation, reasoning and real-time decision-support

tools and techniques. The prognosis system will exploit multiple sources of information, with

varying degrees of reliability, accuracy, completeness and recency, in real time or near-real time,

to assess and predict structural and/or component performance, load carrying capacity and

reliability of individual key components of the transportation infrastructure, as well as to

evaluate the performance, reliability and other pertinent attributes of the overall transportation

system given the current and predicted states of its components. The system is envisioned to

operate in a hostile environment. Once fully operational, the prognosis system will permit

instantaneous and continuous assessment of conditions within the transportation system

infrastructure. It will provide predictions of future conditions, including the likelihood of future

system and component states and how these conditions might evolve over time. Moreover, it will

provide detection and warning of failure or predictions of near-term imminent failure of any

system component. Similarly, it is capable of detecting the lack of and possible need for

connectivity in the transportation network for any use type (e.g. troops, vehicles, tanks,…).

These capabilities will (1) enable on-line bomb damage assessment (BDA) and target

vulnerability assessment (TVA), (2) provide updated information for both tactical and strategic

operational decision-making to adaptively manage and deploy available resources within a

region (e.g., in positioning troops and supplies in a battle or in developing evacuation plans for a

civilian area under attack), and (3) support immediate response to an attack to facilitate quick

recovery from the incurred damage. In addition to its on-line capabilities, off-line uses of the

system may include a complementary role to scheduled inspection and maintenance procedures,

officer training in battle management, and critical assessment of transportation infrastructure

systems for planned operations.

4

A central feature of the MTI-PTC approach is integration through: (1) the use of a

multidimensional network representation as a common platform; and (2) natural human interface

through shared and secure web-based portal. The former provides a shared inter-operable

representation of the physical and operational infrastructure system of interest, using multiple

operational layers. This will enhance the value of the predictions produced by the prognosis

system by embedding them in a network representation which allows application of a wide-array

of well-established efficient network optimization algorithms. The second feature recognizes

recent strides in use of web portals in various service industries as well as the military forces for

delivery of logistics support services and other tactical information. Portals will provide an

integrated framework for the query and update of the network state estimates and predictions.

The design of the MTI-PTC prognosis system is predicated on the awareness that no

matter how sophisticated the sensor technologies, the information available in dynamic battle or

attack settings will often be less than perfect, and insufficient for definitive confirmation and

assessment of damage. Hence the need for (1) reasoning methodologies based on statistical and

artificial intelligence concepts to make inferences and predictions on the basis of the extracted

information; and (2) a real-time decision making framework that recognizes the varying degrees

of incompleteness and uncertainty in the available information, as well as in the predictions.

Decision analytic models can play a significant role in this context. Rather than build a system

around a single type of sensor information, the MTI-PTC approach will recognize the availability

of a wide range of possible information sources, albeit with varying degrees of completeness,

accuracy, reliability, and recency, in making predictions. Similarly, in using these predictions,

which may be generated by a range of analytical or empirical models appropriate to the type of

component, source of the damage, and specific sensor information, the MTI-PTC system will

provide an integrated platform through the multidimensional network modeling representation,

and explicitly recognize the varying quality of the information in recommending courses of

action to the commanders.

PROJECT OBJECTIVES

As noted, the overall goal of the project is to meet DARPA’s requirements for an integrated real-

time prognosis system for structural and infrastructure damage assessment in battle conditions

and homeland security protection. The focus of the MTI-PTC effort will be on the transportation

infrastructure in a first phase, with two specific target components: bridges, and airfield

5

pavements (as well as critical highway pavements). The objectives of the proposed effort include

development of the following methodological capabilities in support of this goal:

5. Tools and techniques for providing continuous and instantaneous information on the state of

the transportation-infrastructure system and its constituent components based on sensor data,

imaging data, data fusion algorithms and supporting numerical simulations, with data

processing in both the time and frequency domains.

6. Automated processes for recognizing symptoms of damage and diagnosing its severity;

inferring its cause(s) by mapping observed symptoms to possible failure modes; determining

current structural condition and potential unrealized damage as existing loads are

redistributed throughout the structure; and predicting future structural performance, load

carrying capacity, and reliability, considering their probabilistic evolution over time.

7. A multidimensional network-based platform and associated optimization techniques for

integrating information on the current and predicted states of critical system components (and

of the overall transportation system), to enable BDA/TVA assessments and aid in making

dynamically updated strategic and tactical operational decisions that consider precedence

relationships and the probabilistic, dynamic nature of information from a variety of sources.

8. A real-time distributed decision support system that is modular in design, defining a

sustainable prognosis system that can incorporate new models and methodologies.

The remainder of the proposal elaborates on selected critical aspects of the MTI-PTC team’s

approach to the problem, and presents a more detailed statement of work and milestones, and

highlights the team’s cross-disciplinary capabilities and unique qualifications for this project.

DAMAGE ACCUMULATION AND FAILURE MECHANISMS IN STRUCTURES AND PAVEMENTS

Real-time evaluation of the structural integrity of a system subjected to severe, complex loading

conditions (such as those encountered in a battlefield) requires: (a) availability of relevant field

data to assess the current damage state of a component and/or the complete structure; (b) reliable

and secure channels for the transmission of this information; (c) an understanding of the physical

mechanisms that control different types of structural failure modes; (d) assessment of the current

structural condition of the system given the available data; (e) identification of the cause of

damage; and e) reliable analytical models to predict the redistribution of loads in a structure and

future structural damage states. The integration of the aforementioned issues in a structural

performance assessment context should incorporate all relevant sources of uncertainty (e.g.,

6

uncertainties in the sensor data itself, in loading conditions, material properties, analysis models

and methods, etc.) and address the propagation of uncertainties throughout the process. This

section of the proposal focuses on issues regarding the physics of various failure modes in a

structure and the evolution of damage (e.g. damage accumulation) based on interacting failure

mechanisms and/or additional loading conditions.

Various types of structural failure modes (e.g. shear, bending, buckling failure and many

others) may lead to two distinct outcomes: loss of functionality and collapse. Loss of

functionality refers to the inability of a system to achieve its full functional capability due to

structural damage to one or several of its components. For instance, a bridge in the vicinity of an

explosion may still be usable by small vehicles but not by heavy trucks. Collapse refers to the

loss of ability of a structural system to resist gravity loads; a collapse may be “local” or “global”.

A local collapse may occur if a vertical load carrying component (e.g. columns, piers) fails by

brisance, compression, etc… Global collapse will occur if local collapses propagate (cascading

collapse) or if a structural system displaces sufficiently so that instability occurs. One of the main

challenges of this project is to predict in a probabilistic framework the redistribution of loads

after a local collapse has occurred and the potential propagation of local collapses that may lead

to the global collapse of the structure.

In practice, collapse is usually associated with an “acceptable” response quantity

(displacement, stress ratio) or the attainment of a limit value of load (or deformation) in

individual components of the structure. This approach does not permit a redistribution of

damage and does not account for the ability of the system to sustain significantly larger

deformations (e.g. due to its own weight or additional loading conditions) or changes in

structural properties (e.g. due to high temperature effects) before collapse occurs. Thus, in many

cases, there is a need to incorporate material and geometric nonlinearities in the analyses so that

damage accumulation and the evolution of the state of a structure caused by multiple failure

mechanisms can be adequately represented. For instance, Dr. Medina, a senior investigator on

the MTI-PTC team, has carried out extensive work on the probabilistic collapse assessment of

deteriorating SDOF [Ibarra et al. 2002] and MDOF systems subjected to strong ground motion

shaking. In this work, nonlinear models that incorporate history-dependent strength deterioration

and stiffness degradation are utilized in order to simulate the dynamic behavior of a structural

system under complex loading conditions [Krawinkler et al. 1999; Medina, 2003]. The models

7

developed in that research, and the insights gained, provide a starting point to incorporate some

of the critical elements needed to provide an estimate (central value and a measure of dispersion)

of the current damage state of a structural system and future failure mechanisms that may

develop under the action of varying loading conditions.

In order for the models to be adequate for a real-time prognosis system, they need to be

calibrated on experimental data. Calibration requires close and extensive collaboration between

analysts and experimentalists. One of the tasks of this project is to utilize available (in the first

18-month phase), and eventually new, experimental data to estimate the reliability of predictions

based on analytical models. The participation of the PTC in this effort ensures access to the

largest database of experimental data relevant to bomb blasts and other extreme loads.

Airfield Pavements, Materials, and Monitoring

Advances in the areas of pavement materials modeling, sensor technology, and pavement

condition monitoring, will contribute to the real-time prognosis system for assessment of airfield

pavements and critical links of the transportation infrastructure. Existing knowledge in the

physics of failure mechanisms in pavement materials under “normal” conditions will be used in

modeling material failure mechanisms under complex and extreme conditions (such as impact

from explosives and other forces), as well as damage accumulation and evolution in pavement

materials and structures. Modeling of failure mechanisms due to multiple and interacting damage

events will also be undertaken. The effort will use a probabilistic approach that brings together

the extensive knowledge and modeling from the experimental world with reliability concepts to

estimate current conditions and predict future states.

The MTI-PTC project team will consider alternative sensor network deployment and

availability scenarios for developing the prototype detection and assessment system. Some of the

principal components considered in building the prototype system are discussed next.

Pavement/Material Failure Modeling. Pavements fail under a variety of mechanisms, and in

function of the pavement types and materials used, design characteristics, loading and

environmental parameters. Pavement engineers have modeled failure mechanisms for several

years using constitutive modeling, and probabilistic modeling for pavement condition and

performance prediction. There exists a sizeable body knowledge on the physics of concrete, and

asphalt pavement failure modes, especially for cracking failure and crack propagation under

routine pavement loading. In the context of this research, the physics of failure models due to

8

extreme events (such as explosive blasting) will be addressed for the pavement/material system.

Predicting the mechanism of blast fracture of concrete has to do with stress wave theory, in

which the magnitude and shape of this rapidly moving wave depends on such factors as the type

of explosive (explosive weight and detonation velocity), blast depth, relationship of detonation

velocity to wave propagation velocity in concrete, material properties, bore hole diameter, and

distance from hole where the stress measurement is determined. Considering such factors, the

crater diameter characteristics could be estimated. The project team will consider input values

representing typical material characteristics for modeling damage mechanisms and examining

damage accumulation and evolution from multiple and interacting damage. Building on the

deterministic analysis, a stochastic process modeling framework (such as Markov Chains) will

be used in determining damage accumulation and evolution. Finally, energy analysis will be

conducted for evaluating the absorbed energy capacity of different materials used in airfield

pavements. The effects of such failure mechanisms on pavement structure will be examined

using finite element analysis and the Federal Aviation Administration LEDFAA developed for

airfield pavements. The pavement group at the University of Maryland, which is a member of the

MTI-PTC team assembled for this project, is internationally recognized for its contributions and

extensive experience in pavement analysis and design, which covers a very broad spectrum of

materials and modeling techniques.

Condition Assessment Techniques and Field Instrumentation. Detection and assessment of

damage may be achieved with remote automated visual inspection/ monitoring techniques using

portable long range cameras, other visual reconnaissance aids, and eventually satellite imaging.

Imaging techniques for evaluating pavement condition have been developed and used on a

routine basis by highway agencies. Commercially available systems include cameras and

imaging analysis software. The majority of these systems analyze video images/film, interpret

the data, and generate pavement condition indices. Adaptation and response of such visual

systems, using long range cameras combined with imaging analysis techniques will be examined

by the project team. The resolution of the images will clearly affect the accuracy of the data, and

of the resulting damage assessment. Therefore, the data and results from the visual surveys may

be used in conjunction with stochastic material damage models to provide probabilistic

assessment for use by the inference and optimization engine of the decision-support system.

Other far-field monitoring techniques are discussed in the next section.

9

When feasible, the use of field instrumentation in airfield pavements will enable

detection of incipient and advancing damage in real time. Extensive experience in monitoring

highway and airfield pavement conditions exists today with the use of LVDTs, strain gages,

thermocouples, vibrating wire gages, and other devices. Such instrumentation provides a variety

of accurate measurements for assessing pavement material condition and can provide warning

for extreme events. This instrumentation can be strategically placed in key locations of the

airfield pavement with built-in redundancy for survivability. Finally, the possibility of using

smart materials will be examined. The use of fiber optics, piezoelectric materials,

electromagnetic and acoustic sensing can be imbedded in pavement materials/structures for

monitoring the behavior and condition, detecting loud noises and failure events. The project team

will investigate the potential of smart material technology in such a system.

SENSOR AND SENSOR NETWORKS FOR INFRASTRUCTURE HEALTH MONITORING

Condition assessment of the infrastructure after blasts may require different equipment

depending on the accessibility of the structure’s location. If a structure is pre-equipped with

sensors, and data before damage exists, then a detection approach in which signals obtained

under the current condition are compared to signals from the undamaged condition can be

applied. Even for these situations, condition assessment (reliability of the current state and

remaining life of the structure) is by no means a straightforward endeavor. The major difficulty

is the low signal-to-noise ratio that typical structural damage produces, especially at the

incipience of the damage. When a structure is not pre-equipped with sensors, condition

assessment is significantly more difficult because sensors have to be placed on the structure, and

the measured signals cannot be compared to the signals before the damage occurred. Both

problems are discussed hereafter.

For convenience in this proposal, the term “near-field monitoring” refers to condition

assessment at accessible locations, and “far-field monitoring” to assessment at inaccessible

locations. Far-field monitoring methods are also applicable for structures in accessible areas.

Far-field monitoring methods include satellite imaging, aerial photogrammetry, laser Doppler

vibrometers, vibro-thermography, shearography, and holography. Current satellite images have

resolutions of up to 0.1m. This level of resolution is sufficient to determine if a bridge has

collapsed. Future improvements in resolution and fly-over frequency will improve both the

assessment accuracy and the time required to obtain the satellite image after the event. For the

10

time being, aerial photogrammetry can be used to provide better resolution, and airplanes can be

dispatched almost immediately after a major event.

Laser Doppler vibrometers can be used to measure displacement of a structure.

Displacement-based methods are typically novelty detection methods in that damage is inferred

by the change in deflection from an undamaged state. These methods are better at determining

large damage rather than incipient damages since the latter are by definition small changes in the

structure that result in negligible overall displacement changes. The need for a stable location

for the vibrometers and distortions in signals caused by the environment such as heat severely

handicap this method as a viable far-field monitoring choice.

Vibro-thermography refers to the change in temperature at damaged areas such as cracks

due to vibration that can be picked up by infrared thermography. This method has been applied

to the monitoring of machine parts with great success. In these applications, the machine part to

be tested is carefully isolated from the environment so that it reaches a steady state temperature

throughout the part. When vibrated, cracks generate sufficient heat that can be picked up by

infrared imaging. Modern equipment can pick up as little as a change in temperature of 0.01 C.

Isolation of infrastructure so it reaches a steady state is not possible. Application of this method

to structures also requires that the infrared imaging be done at a much greater distance.

Nevertheless, this approach has some merits as blasts are generally associated with large

temperature gradients that can be measured at great distances.

Digital Holography is similar to shearography. In this technique, two light sources are

used, and the interference pattern is used to determine the displacement or strain of the object.

Holographic techniques are not as stable in noisy and dynamic environments. It is more

amenable to static measurements. Shearography refers to the splitting (shearing) of an optical

image. The sheared image is then compared to the original image by using CCD cameras and

the technique of Electronic Speckle Pattern Interferometry to determine in-plane or out-of-plane

deformation and strain. Use of pulsed laser enables indeed to work in a noisy environment and

to catch transient events [Tiziani, 1996]. This technique has been applied to the damage

assessment of smaller objects such as airfoil and tires with great success. It requires the use of a

laser, mirrors and a computer. It is only suitable for short-term measurements. Therefore,

damage must occur during the test. This technique can be converted to infrastructure

measurement if a sufficiently strong laser such as the pulsed Yag laser can be deployed.

11

An alternative to the far-field monitoring approach is to use place sensors directly on the

structure. These sensors must be easy to deploy. One such concept is the use of Motes. The

Mote concept is to develop a 1 mm3 size sensor containing both the sensor, the transducer, and

the battery. Current size of the Mote is approximately 0.5 cm3. Although small and low in

power, Motes are designed to relay information from Mote to Mote. In principle, they can be

equipped with any kind of sensor and spread to the infrastructure from an airplane. A pick-off

device only needs to be placed near one of the Motes. The information from the other Motes is

transmitted through a self-forming network as long as they are within pinging range of each

other (approximately 10 m). This approach is not robust for long-term monitoring as battery

management would render the scattered sensors useless in days. When these Motes are equipped

with an accelerometer and actuator, condition assessment similar to a piezoelectric smart patch

can be produced.

For long-term infrastructure assessment, a passive and tetherless sensor costing in the

vicinity of a dollar per sensor is an excellent alternative to the far-field monitoring techniques.

These sensors combined with devices to remotely power the sensors can be used to provide

three-dimensional positions of the sensors to within a centimeter of accuracy. The measurement

of displacement is based on radio frequency. The remote powering unit works like a radio

transponder. The radio frequency is then reflected from each sensor. By measuring the time of

arrival of radio frequencies from each sensor, the position of the sensor can be defined if three or

more transponders are used. Schemes for time synchronization and filtering of reflected signals

have been elaborated successfully.

Wireless transmission of sensor information is likely to play a critical role in a battlefield

environment, especially as one seeks to perform BDA/TVA in hostile territory. Ongoing

developments in hybrid mobile networks offer a natural application in transportation

infrastructure monitoring in the context of the present prognosis system. Massively dense sensor

networks will gain increasing attention and applicability as the cost and size of devices continues

to decrease dramatically, and swarm intelligence concepts could be applied to the problem.

SENSOR DATA FUSION, INTERPRETATION AND MULTI-VALUED REASONING

Damage assessment of structures and pavements subject to explosive attack within the context of

a real-time prognosis system to support battle management faces the following four major

difficulties: (1) the environment is physically hostile and mentally confusing; (2) experts may not

12

be available at the site, yet their recommendations are needed; (3) time is limited and decisions

must be made before another attack is launched against the site, and (4) the assessment is a

difficult task which contains both ill- and well- defined parameters. Therefore the process of

drawing inferences from the data to assess damage conditions and predict future capacity calls

for logic to interpret possibly correlated or conflicting sources of information, and to engage in

the kind of reasoning process that recognizes the various sources of uncertainty and inherent

stochasticity, as well as imprecision in certain types of data. In addition to formal procedures

from Decision Analysis or statistical modeling, several mechanisms have emerged in the past 15

years that can be viewed as hybrid approaches combining ideas and methods from artificial

intelligence, statistics and data mining, as well as behavioral decision theories.

Evidential Reasoning Mechanism

The evidence-hypothesis reasoning mechanism is the task of inferring the belief in some

hypotheses by collecting relevant evidence for or against these hypotheses. The inexact

relationships among hypotheses and evidence are classified depending on the nature of evidence,

i.e., measurement or observation. Linguistic hypothesis-evidence reasoning manipulates if-then

rules to manifest the uncertainty associated with hypothesis-evidence relationships. Numerical

hypothesis-evidence reasoning deals with measurements where the inexact relationships between

evidence and hypotheses are presented by two-dimensional plots. Varying degrees of confidence

may be associated with particular evidence.

Belief Revision

Information is subject to change due to inherent uncertainty or because the environment is

volatile and dynamic. Current non-monotonic reasoning systems fall short in adequately treating

changing information, the major problem being that once a change in the knowledge base,

however minor, is performed, one must begin from scratch; this is completely inappropriate,

especially given that evidence fusion is computationally hard. Belief revision (BR) is a research

area that has developed techniques capable of dealing with changing information.

Diagnostics

Diagnostics in general refer to the task of inferring plausible explanations for a set of evidence,

or to decide which explanation accounts for given evidence, such as inferring the cause of

degradation detected in a given critical structure. A classical method of diagnostic analysis is

based on Bayesian analysis. In this case, the relations between evidence and underlying causes

13

are described by conditional probabilities, and the possible causes are assumed to be mutually

exclusive. However, the mechanical, physical, and chemical processes of degradation can act in

a synergistic manner, thereby invalidating the assumption of mutually exclusive causes. In

addition, the actual underlying causes might be something other than the defined causes that the

problem solver is aware of. Several researchers in artificial intelligence have generalized the

Bayesian theorem within the framework of evidence theory, whereby conditional probabilities

are replaced by belief functions. Another generalization is obtained by extending the

Generalized Bayesian Theorem (GBT) to handle all types of belief functions, i.e., precise,

interval, and fuzzy.

Pattern Classification

Classification can be defined, in general, as a two-fold task that consists of “learning the

invariant and common properties of a set of samples characterizing a class” (abstraction) and of

“deciding if a new sample is a possible member of the class” (generalization). Evidence theory

provides a suitable framework for combining vague and ambiguous results of several

independent classifiers, and thereby improve pattern classification accuracy. It allows to

compute a belief assignment over the set of classes, in such a way that the mass of belief

assigned to the reference set varies between 0 (perfect knowledge) and 1 (complete ignorance),

depending on the information content of the training data with respect to the class membership of

the pattern under consideration.

Closed-World Versus Open-World Assumption

The simulated performance of a system depends heavily upon the information available about

the problem under consideration. Complete information is difficult to come by, and is generally

not available even for simple applications. In general, an intelligent system must be able to make

plausible propositions that may turn out to be incorrect when more information becomes

available. The transferable belief model provides a basis for a class of methods for making such

propositions when faced with incomplete information.

The transferable belief model (TBM) is a non-probabilistic approach that derives from the

Dempster-Shafers mathematical theory of evidence. It is a means for representing quantified

degrees of belief. Degrees of belief are obtained from agents providing evidence at a given time

within a given frame of discernment. The method is capable of treating inconsistency in data by

introducing the "open-world" assumption. In TBM, a set of all propositions consists of the three

14

subsets: (1) a set of propositions known as possible (PP), (2) a set of propositions known as

impossible (IP), (3) and a set of unknown propositions (UP). The content of the subsets depends

not only on the given problem, but also on the evidence, which is available at a given time. As

evidence becomes available, propositions are redistributed between the three sets. The closed-

world assumption postulates an empty UP set. The open-world assumption admits the existence

of a non-empty UP set, and the fact that the truth might be in UP. In this assumption, unknown

referred to none of the known propositions. The degree of conflict between two or more

evidence sources implies the existence of a proposition not defined in the frame of discernment.

In the idealized closed-world assumption that amount of conflict is redistributed among the

known propositions. In the open-world assumption, the degree of conflict corresponds to the

amount of belief allocated to the proposition that none of the known propositions has the truth.

These approaches are mentioned in this proposal to illustrate the wide range of

modeling and simulation tools that the MTI-PTC team has first hand familiarity with, and the

importance of a cross-disciplinary approach to this problem.

STATEMENT OF WORK

The project objectives will be achieved through the following tasks.

1. Development of the techniques for optimally placing sensors for each class of system

components.

2. Creation of the algorithms based on concepts from system identification approaches for

synthesizing and analyzing the sensor data for each class of system components that can

be employed on-line in an automated environment to identify symptoms of damage and

to provide severity diagnosis.

3. Development of the on-line approaches based on reasoning methodologies for inferring

from the current state of a given system component and information on prior states, the

component’s probable performance, remaining structural capacity, remaining life and/or

reliability over the near future.

15

4. Development of the techniques for converting video and satellite images to employable

electronic data for use in evaluating the overall transportation-infrastructure system, in

support of other types of acquired information.

5. Creation of the methodologies based on network optimization concepts for combining the

information on current and predicted conditions from the various components of the

system with information on the overall system to provide a comprehensive depiction of

current and predicted states with associated time-varying probabilities for the entire

system.

6. Integration of the above methodologies and techniques in a single electronically

integrated and automated modular decision support system (DSS) that exploits concepts

of network optimization and systems engineering.

7. Development of a simulation test-bed for the proposed system demonstration and

evaluation.

8. Design, performance, analysis and interpretation of controlled experiments in the

simulation test-bed to assess the practical utility of the proposed monitoring and

prediction methodologies under controlled conditions, as well as the viability of the

proposed system.

9. Preparation of final report and demonstration prototype along with supporting

documentation.

PROJECT OUTCOMES AND DELIVERABLES

The outcome of these developments will be: (1) a set of tools that, for each system component,

can fuse raw data from many sensors and/or images, and supporting numerical simulations, and

can, through real-time data analysis (in both the time and frequency domains), determine the

current state of the component, including its current structural capacity and reliability; (2) a set of

methodologies for inferring the likelihood that a given system component will have a particular

16

level of reliability and structural capacity at each moment in the future over a specified time

period; (3) a set of tools that can provide a comprehensive picture of the current and predicted

states of the larger transportation-infrastructure system; and (4) a single electronically integrated

and automated DSS with modular design that will enable real-time strategic and tactical

decision-making in a hostile environment.

PROJECT PHASING

Specific targets are given specifically for the first 18 months, or Phase I, of this study.

Subsequent effort is pending additional approval of the prototype:

(1) Development of the key methodologies described above, including system integration via

network optimization-based techniques.

(2) Creation of a working model of the DSS with the embedded methodologies.

(3) Testing of the DSS to illustrate the system’s capabilities and to evaluate the feasibility of the

proposed prognosis system.

These three goals together comprise a proof of concept.

The Phase I tasks are intended for completion at the end of the 18-month period, which is

when all deliverables for the first phase will be due.

Upon successful completion of the proof of concept, a fully operational version of the

system would be developed, tested and demonstrated in a second phase, pending DARPA

approval and further input.

REFERENCES

Ibarra, L., Medina, R., Krawinkler, H. (2002). “Collapse Assessment of Deteriorating SDOF Systems”, Proceedings of the 12th European Conference on Earthquake Engineering, London, UK, Paper Ref. 665, Elsevier Science Ltd., September 9-13, 2002.

Krawinkler, H., Medina, R., Miranda, M., Ayoub, A. (1999). “Seismic Demands for Performance-Based Design”. Proceedings of the U.S. - Japan Workshop on Performance-Based Earthquake Engineering Methodology for Reinforced Concrete Building Structures, Maui, Hawaii, PEER Report 1999/10, 1999.

Medina, R. (2003). “Seismic Demands for Nondeteriorating Frame Structures and their Dependence on Ground Motions”, Ph.D. Dissertation, Department of Civil and Environmental Engineering, Stanford University, January 2003.

Tiziani, HJ., G. Pedrini, (1996). Digital holographic interferometry for shape and vibration analysis.Proceedings of the SPIE ‘96 conference, The International Society for Optical Engineers.

17

A MULTI-DISCIPLINARY RESEARCH TEAM

An impressive team that integrates all applicable disciplines and technical and methodological

capabilities required for the conception, development and implementation of a real-time

infrastructure damage assessment and prognosis system has been assembled for this project.

The team includes experts from two main entities: the Maryland Transportation Inititative (and

associated Centers) at the University of Maryland, and the Protective Technology Center

currently located at the Pennsylvania State University. The Director of the MTI, Dr. Hani S.

Mahmassani will serve as the overall Principal Investigator for the project, in collaboration with

the Director of the PTC, Dr. Theodore Krauthammer, who will lead the PTC team (as

subcontractor to the University of Maryland). Dr. Elise Miller-Hooks at the University of

Maryland, will be a co-PI on the project. In addition, 10 faculty and senior researchers at the

University of Maryland will serve as senior investigators on the project; similarly between 5 and

8 additional senior investigators will participate from the PTC. Individually and collectively, the

team brings an outstanding array of skills, expertise and relevant experience for this project that

is difficult to match anywhere. Dr. Mahmassani and Dr. Miller-Hooks of the MTI will have

primary management responsibility for the project. Figure 1 summarizes the management chart

of the project team, by primary area of expertise and institution.

Dr. Hani S. Mahmassani is the first holder of the Charles Irish Sr. Chair in Civil and

Environmental Engineering at the University of Maryland, where he is also the founding

Director of the Maryland Transportation Initiative, a new cross-disciplinary institute for

transportation systems research and education. He joined the University of Maryland in August

2002, after 20 years on the faculty at the University of Texas at Austin, where he was most

recently the A. Abou-Ayyash Centennial Professor in Transportation Engineering, Professor of

Management Science and Information Systems and Director of the Advanced Institute for

Transportation Infrastructure Engineering and Management. He specializes in the real-time

estimation and prediction of network traffic flows for intelligent transportation system operation,

dynamic network modeling and optimization, dynamics of user behavior and telematics,

telecommunication-transportation interactions, large-scale human infrastructure systems, and

real-time operation of logistics and distribution systems.

Dr. Mahmassani received his PhD in Transportation Systems from the Massachusetts Institute of

Technology in 1982 and MS in Transportation Engineering from Purdue in 1978. He chairs

18

several technical and professional committees in the US and internationally, and is the incoming

Editor-in-Chief of Transportation Science, and Associate Editor of Transportation Research C

(Emerging Technologies), and the IEEE Transactions on Intelligent Transportation Systems. He

serves on the editorial boards of the major transportation journals. He has served as the Paper

Review Chair of the Traffic Flow Theory and Characteristics of the Transportation Research

Board since 1989, and is the immediate past president of that committee. He is a past president

of the Transportation Science Section of the Institute for Operations Research and the

Management Sciences, and the immediate past President of the International Association for

Travel Behavior Research. He has published over 200 refereed articles and conference

proceedings, and over 100 technical reports. He has served as Principal investigator on over 100

funded research projects sponsored by the National Science Foundation, The Advanced

Technology Program, General Motors Research Laboratories, Texas Department of

Transportation, Texas Governor's Office, Capital Metropolitan Transportation Authority, U.S.

Department of Transportation, Federal Highway Administration, and served as consultant to

several companies and government agencies in the areas of transportation modeling, operations,

logistics and intelligent transportation systems. He has served in an advisory capacity to various

institutes and programs, and has performed several program assessments of leading international

research institutes and corporate R&D departments.

Dr. Theodor Krauthammer is Professor of Civil Engineering at the Pennsylvania State

University and Director of the Protective Technology Center. He obtained his Ph.D. in Civil

Engineering from the University of Illinois at Urbana-Champaign. His main research and

technical activities are directed at structural behavior under severe dynamic loads, including

considerations of both survivability and fragility aspects of facilities subjected to blast, shock

impact, and vibrations. He has specialized in the nonlinear behavior of structures (including

medium-structure interaction) under seismic or impulsive loads, and the development of

numerical simulations and testing techniques for structural assessment. His work has included

the development of design recommendations for enhancing structural performance, physical

security and safety of buildings, facilities and systems. Additionally, he has been involved in the

response, safety, and physical security assessment of buildings and facilities before and after

abnormal loading events. He conducted both numerical simulation, and experimental studies of

structures subjected to a broad range of loading conditions. His research has been supported by

19

the Department of Defense Explosive Safety Board, Defense Threat Reduction Agency, Federal

Highway Administration, Minnesota Department of Transportation, National Science

Foundation, NATO, Norwegian Defence Construction Service, US Air Force, US Army, US

Marine Corps, US Navy, and the Pennsylvania Department of Transportation.

Dr. Krauthammer is a Fellow of the American Concrete Institute (ACI), a member of the

American Society of Civil Engineers (ASCE) and is involved with other national and

international professional organizations. He serves on nine technical committees of ASCE, ACI,

and the American Institute of Steel Construction (AISC). Dr. Krauthammer is chair of the ASCE

Task Committee on Structural Design for Physical Security, and Chair of the Joint ACI-ASCE

Committee 421 on Design of Reinforced Concrete Slabs. He was the founding Chair of ACI

Committee 370 on Short Duration Dynamics and Vibratory Load Effects. Also, he was Chair of

the ASCE Engineering Mechanics Committee on Experimental Analysis and Instrumentation

and of the ACI Committee 444 Experimental Analysis of Concrete Structures. He was a member

of the National Research Council Defensive Architecture Committee. Dr. Krauthammer has

written more than 290 research publications, and has been invited to lecture in the USA and

abroad. He has been a consultant to industry and governments in the USA and abroad. He is

very active in organizing cooperative international scientific, technical and engineering

education activities.

Dr. Elise Miller-Hooks is currently an assistant professor of Civil and Environmental

Engineering and a member of the Operations Research Faculty at The Pennsylvania State

University. She will be joining the faculty in Civil and Environmental Engineering at the

University of Maryland starting in Summer 2003. She received her M.S. in Engineering and

Ph.D. in Civil Engineering from the University of Texas at Austin in 1994 and 1997,

respectively, and her B.S. in Civil Engineering from Lafayette College. Professor Miller-Hooks

has been principal investigator on projects funded by the National Science Foundation, United

Technologies Corporation, the Pennsylvania Department of Transportation, the National Institute

of Statistical Sciences, the Protective Technology Center; and the Sloan Foundation for research

in, among other areas: robust on-line location and routing for urban service systems; routing in

intermodal transit-based systems; routing and scheduling service technicians; development of an

intelligent evacuation, rescue and recovery system prototype for large buildings that depends on

information from building-wide sensors and real-time damage assessment capabilities; and

20

algorithms for emergency preparedness planning and on-line evacuation of large intelligent

buildings. Her research interests include operations research methods and applications in

transportation; network optimization and algorithms; stochastic and dynamic network

algorithms; reoptimization algorithms to exploit real-time information; logistics; and multi-

objective decision-making. Professor Miller-Hooks has given over 40 technical presentations at

international conferences and invited lectures, and publishes regularly in refereed journals in

Transportation and Operations Research. She has also recently published in the Proceedings of

the 30th United States Department of Defense Explosives Safety Seminar, Proceedings of the

11th International Symposium on Interaction of the Effects of Munitions with Structures, and

Proceedings of IEEE Systems, Man and Cybernetics. Professor Miller-Hooks teaches graduate

and undergraduate courses on transportation network algorithms, stochastic models in

transportation, transportation systems engineering, and optimization in civil engineering systems.

She has received several awards including her recent selection as the first Outstanding Young

Graduate of the Civil Engineering Department at the University of Texas at Austin

Dr. Bilal M. Ayyub is a Professor of Civil and Environmental Engineering and the Director of

the Center for Technology and Systems Management at the University of Maryland (College

Park). He is a researcher and consultant in the areas of the physics of failure models, damage

accumulation and evolution in materials and structures, stochastic and probabilistic effects in

materials, structural engineering, inspection methods and practices, logic and control, reliability

and risk analysis, uncertainty modeling and analysis, mathematical modeling using the theories

of probability, statistics, and fuzzy sets. He completed his B.S. degree in civil engineering in

1980, and completed both the M.S. (1981) and Ph.D. (1983) in civil engineering at the Georgia

Institute of Technology. He has completed several research projects that were funded by the

U.S. National Science Foundation, the U.S. Coast Guard, the U.S. Navy, the U.S. Department of

Defense, the U.S. Army Corps of Engineers, the Maryland State Highway Administration, the

American Society of Mechanical Engineers, and several engineering companies. Dr. Ayyub has

served the engineering community in various capacities through societies that include ASNE,

ASCE (Fellow), ASME (Fellow), SNAME (Fellow), IEEE (Senior Member), NAFIPS, Risk

Society, World Future Society, among others. He is a member of the ASME Research

Committee of Risk Technology, and the ASME Committee on Human Factors. Recently, he

chaired the ASCE Committee on the Reliability of Offshore Structures. He also was the General

21

Chairman of the first, second and third International Symposia on Uncertainty Modeling and

Analysis that were held in 1990, 1993, and 1995. Currently, he is the chairman of the design

philosophy panel of the SNAME Ship Structures Committee. He is the author and co-author of

about 350 publications in journals and conference proceedings, and reports. His publications

include five textbooks, several edited books, and many book chapters. Dr. Ayyub is the only

multiple recipient of the ASNE "Jimmie" Hamilton Award for the best papers in the Naval

Engineers Journal in 1985, 1992, 2000 and 2002. Also, he received the ASCE "Outstanding

Research Oriented Paper" in the Journal of Water Resources Planning and Management for

1987, the NAFIPS K.S. Fu Award for Professional Service in 1995, the ASCE Edmund

Friedman Award in 1989, and the ASCE Walter L. Huber Civil Engineering Research Prize in

1997. The U.S. Army Corps of Engineers recognized him for his contributions to the success of

the 1995 Corps of Engineers Structural Engineering Conference, Fort Worth District, San

Antonio, Texas, 1995. He recently received a Certificate of Appreciation, Research Task Force

on Risk-Based Inspection Guidelines, the American Society of Mechanical Engineers, 1998, and

a Certificate of Appreciation from the Risk-based Technology Research Committee of the

American Society of Mechanical Engineers (ASME) for contribution to several ASME research

projects on risk-based inspection of industrial facilities, risk-informed design of mechanical

systems, and applications of risk-based technology to U. S. Coast Guard marine systems, 1998.

The following URL addresses provide additional information of previous and ongoing activities

of Dr. Ayyub: http://ctsm.umd.edu or http://ctsm.umd.edu/ayyub

Dr. Peter Chang is on the faculty of the Civil and Environmental Engineering Department at the

University of Maryland. He is also a member of the Small Smart Structure Research Center. He

served as a program director at the National Science Foundation from 2001 to 2002 in the

Division of Civil and Mechanical Systems. He was invited to be the keynote speaker at the First

European Workshop on Structural Health Monitoring in France and the Advanced Building

Technology Conference in Hong Kong. Dr. Chang have given research seminars at the Smart

Structures Research Center of Japan, University of California, Virginia Polytechnic and many

other major universities in the US. His research interests include sensors technology and

structural health monitoring. He has done research in the area of wireless and passive sensors for

health monitoring purposes.

22

Dr. Chung C. Fu, P.E. is currently Director of the Bridge/Building Engineering Software &

Technology (BEST) Center, and Affiliate Associate Professor in the Department of Civil and

Environmental Engineering at the University of Maryland. Before he joined the University in

1987, he was an Engineering Supervisor of the Bechtel Company in Gaithersburg, MD. For the

past fifteen years, he has been a Principal Investigator or Co-Principal Investigators for over a

hundred projects with a value of seven million dollars. His areas of expertise include Structural

Engineering; Bridge Engineering; Earthquake Engineering; Computer Application in Structures;

GIS and database application; Finite Element Analysis; Concrete; Steel & Composite

Application; Foundation Engineering. Other interests include Material Engineering;

Bridge/Tunnel Construction; Software Development; Bridge Management; Bridge/Tunnel

Inspection. Dr. Fu has served as principal investigator for the National Transportation Safety

Board (NTSB) in (1) the Collapse Study of Great Miami River Temporary Bridge, Miamitown,

Ohio in 1990, (2) the Collapse Study of New Orleans Judge Seeberg Bridge under barge impact,

New Orleans, LA, in 1993, (3) the Collapse Study of Mobile AMTRAK Bridge under barge

impact, Mobile, Alabama, in 1994 and (4) the Crash Study of the N.Y Route 58 over West

Oswegatchie River due to truck impact on truss bridge elements in 2000 as well as other expert

witness in bridge failure investigations. Dr. Fu has used the reliability approach to study the

redundancy of a truss bridge system (N.Y route 58, over west Oswegatchie river) in the state of

New York. For the past decades, the BEST Center also has conducted many bridge tests using

several sensor systems and applying finite element method to simulate the behavior of the bridge

under various loading conditions. Dr. Fu and his associates are currently investigating new

generation sensors in the bridge area to develop reasoning methodologies for future capability

prediction. Also, in the 90s, he has worked with Argonne National Lab and the Army Corps of

Engineering, using artificial intelligence techniques and uncertainty modeling to detect the

bridge capacity in the European and Korean battle field theaters.

Dr. Dimitrios Goulias is an Associate Professor in the Civil Engineering Department of the

University of Maryland. The expertise and interests of Dr. Goulias are in the area of pavement

engineering and he has lead research in excess of $1,8 million (as PI and/or Co-PI) addressing

the needs of agencies from NYS, NYC, NJ, VA, MD, TX, the industry, FHWA, and PANYNJ.

Specifically, Dr. Goulias has extensive experience in the design, behavior and performance

evaluation of pavement materials, pavement condition assessment and NDT, pavement

23

performance and specifications, and field instrumentation. Dr. Goulias is currently involved in

the development and evaluation of High Performance Concrete Materials for highway and

airfield pavements, and field pavement instrumentation. The interests and expertise of Dr.

Goulias also cover Pavement Management Systems, GIS, use of spatial data and time series

analysis for highway analysis and sensor placement, high performance materials, smart materials

and composites. Dr. Goulias is involved in several national and international professional and

scientific committees and his work is published in more than 60 refereed publications,

conference proceedings and technical reports.

Dr. Ali Haghani is an expert in large scale network optimization and fleet management. In the

past 10 years his research efforts have focused on developing large scale network models and

decision support systems for transit fleet management, public service fleet management and real

time traffic network control. He has developed one of the prototype real-time traffic adaptive

control systems for the Federal Highway Administration and fleet management decision support

systems for Mass Transit Administration in Baltimore and Maryland State Highway

Administration. Dr. Haghani is the director of undergraduate and graduate studies in

transportation engineering at the University of Maryland. Dr. Haghani has served as a member of

the Editorial Advisory Board for Transportation Research and is currently a member of the Joint

Publication Committee for Highway Engineering and Urban Transportation, ASCE, and the Bus

Transit Committee of the Transportation Research Board. He is also the Chairman of the

Transportation Network Modeling Committee of the Transportation Research Board.

Dr. Ricardo A. Medina is an Assistant Professor in the Department of Civil and Environmental

Engineering at the University of Maryland, College Park. He obtained his Ph.D. degree in

Structural Engineering at Stanford University. His research interests include: nonlinear behavior

of structural materials, performance-based design, earthquake engineering, structural dynamics

and structural reliability. Dr. Medina has carried out extensive work in areas such as the

probabilistic assessment of the global collapse potential of SDOF and MDOF structures

subjected to strong ground shaking. In this work, he utilized nonlinear hysteretic models to

represent the physics of various failure modes at the structural component level and take into

account history-dependent strength deterioration and stiffness degradation. The hysteretic

models trace deterioration as a function of past loading history and the energy dissipation

24

capacity of each component. Dr. Medina has also worked on the global collapse assessment of

MDOF structural frames due to second order (P-Delta) effects.

Dr. Richard La brings together considerable scientific, engineering and practical expertise

(acquired through industry stints with Motorola and Alcatel U.S.A.) in communication networks,

especially wireless communications, which is a critical element in this project both for the sensor

side as well as on the control side.

Dr. Charles W. Schwartz’s expertise is in the areas of pavement engineering and

geomechanics. His research interests focus on analytical and numerical modeling techniques for

pavement structures, characterization of pavement materials and pavement systems, and

pavement performance modeling and forecasting. He currently is Co-Principal Investigator for

the Superpave performance models project (NCHRP Project 9-19), where he leads the effort to

develop advanced material models that track the evolution and accumulation of damage for the

multiple distress modes in asphalt concrete under combined structural loads and environmental

influences. He also led the University of Maryland portion of the AASHTO 2002 Pavement

Design Guide development effort (NCHRP 1-37A), which focused on mechanistic-empirical

methods for forecasting pavement performance. Other recent research includes investigations of

stress-dependent behavior and plasticity of unbound pavement materials, geosynthetic

reinforcement for flexible pavement structures, and neural network modeling of airfield

pavement performance. Dr. Schwartz in the past has also been active in the development and

implementation of pavement management systems for airfield and highway pavements,

including John F. Kennedy, Newark, and Laguardia International airports in the United States,

the national civil airport system in Brazil, and the Delaware state highway network.

Dr. Min Wu brings essential expertise in the areas of multimedia data transmission and security,

on both sides of the control loop: through the sensor network, where critical decisions of

processing, compression, storage, transmission and authentication must be addressed, as well as

in the dissemination of information in various forms to the end users- where data format and

security are key concerns.

EQUIPMENT AND FACILITIES

The research will be conducted primarily on the campus of the University of Maryland, in

College Park, MD. Space for computers and research assistants is available to the investigators.

The University is a world-class research institution with excellent library collections and

25

computing facilities and services. Extensive wireline and wireless connections to the Internet

are available throughout the campus, as well as through remote access.

The Clark School of Engineering at the University of Maryland is home to one of the most

vibrant compilations of research activities in the country. With major emphasis in key areas such

as communications and networking, systems engineering, rotorcraft technology, optoelectronics,

transportation systems, and space engineering, as well in electronic packaging, smart small

systems and advanced materials, the College is leading the way toward the next generations of

engineering technology. It is notable that the College and the University recognize the

transportation systems area as one of the key areas of excellence at the University.

The Maryland Transportation Initiative (MTI) at Maryland brings together under one roof the

various and diverse transportation research centers and efforts going on at the University. The

graduate program attracts excellent and highly qualified students from all over the world, who

bring excellent academic background and practical work experience to their research projects.

The Center for Advanced Transportation Technologies associated with the MTI is a leader in

promoting the development and application of advanced technologies to real-time transportation

system operations. The Technology Transfer Center helps disseminate the research findings

of the faculty and affiliated researchers.

The Department of Electrical and Computer Engineering at the University of Maryland

has almost 30 laboratories dedicated to research in electrical and computer engineering. Many of

these laboratories have affiliations with other departments, as well as other research institutes

and centers within the university. Richard La have a joint appointment with the Institute for

Systems Research on campus and will have access to the advanced laboratory facilities of the

ISR and its constituent Center for Satellite and Hybrid Communication Networks (CSHCN).

This includes state-of-the-art hardware and software development facilities available in the

Systems Engineering and Integration Laboratory (SEIL) of the ISR and the Hybrid Networks

Laboratory (HNL) of the CSHCN. These laboratories offer advanced testbeds for network

modeling, simulation and management, as well as experimental facilities for wireless

networking. Both laboratories are managed by experienced professional staff.

La and Wu are among 17 faculty members associated with the Communication and

Signal Processing Laboratory (CSPL) that supports research in signal processing and

communications at the University of Maryland. The research expertise includes areas ranging

26

from theoretical analysis, algorithm development, to design and implementation of

communication and signal processing systems. The research at CSPL has been supported by

National Science Foundation, Office of Naval Research, National Institute of Health, DARPA,

Army Research Laboratory, Maryland Industrial Partnership, Westinghouse, Watkins Johnson,

Fujitsu, Allied-Signal, and Minta Martin Foundation. The CSPL lab has a dozen of high-

performance SUN and Pentium-IV workstations, and high-speed wired and wireless networks.

The pavement faculty at the University of Maryland has instrumented and monitors three

pavement sections in a major highway artery in Maryland. Furthermore, the services of a leader

in airfield runway instrumentation (CTL Inc.), with ongoing working relation, will be used on an

as needed basis in this project. The Maryland pavement group has been involved in a variety of

studies related to pavement conditions instrumentation, condition assessment and performance

modeling.

The Protective Technology Center At Penn StateInterest in protecting structures from the effects of blast, shock, and impact began at the

Pennsylvania State University in 1989 within the Department of Civil and Environmental

Engineering. By 1995, this interest produced the first one week, university-sponsored short

course designed to provide design engineers, architects, and related professionals with the latest

progress regarding materials, construction techniques, and research to protect this nation’s

critical infrastructure from explosively-induced loading events. The Modern Protective

Structures course received national recognition in a recent Associated Press article (July 14,

2002), and a graduate level course on this topic has been introduced in the Department of Civil

and Environmental Engineering at Penn State in August 2002. In 2000, the College of

Engineering at Penn State, established the Protective Technology Center as a separate entity

tasked with providing an interface between the academic research interests of a major university

with the commercial and governmental needs for technical solutions to structural applications.

The Protective Technology Center’s mission is to research technologies that provide

structural solutions that protect lives and preserve property. This mission implies several

specific objectives:

To assist in mobilizing the U.S. scientific community in support of this national effort.

27

To establish comprehensive long and short-term R&D activities in protective technology

to insure the safety of all U.S. citizens and the protection of critical infrastructure from

terrorist attack.

To develop innovative and effective blast, shock, and impact mitigation technologies.

To promote sharing emerging technologies and research with legitimate users.

To develop and deliver technical education and training venues to present the newest

advances in blast, shock, and impact mitigation technologies.

Current Technical Activities

The largest activity at the center is in support of the US Army Engineer Research and

development Center (ERDC). The actual tasks have been defined by the program participants

and an advisory board to address the required enhanced force protection from blast, shock and

impact, as follows: Force Protection Methodology Development and its Implementation;

Explosive Load Definition; Material Behavior under Blast, Shock and Impact; Computational

Simulations and Computer Code Validation; Building Enclosure under Blast, Shock and Impact;

Structural Behavior, Design, Performance and Safety; Pre- and Post Incident Damage and

Condition Assessments in support of Evacuation and Rescue Recovery Activities; Technology

Transfer and Training. Besides continuing the above mentioned activities, various collaborative

activities with current and future sponsors and in cooperation with industrial partners are under

development.

Facilities

The 2,000 square foot structures lab is the center of PTC’s experimental research. This facility

houses an impact pendulum capable of swinging weights of up to 1,600 lbs 7.1 kN) through an

arc with a vertical drop of approximately 4 ft (1.2 m). The lab’s drop hammer has a weight

capacity of 7,000 lbs (31 kN) and a drop height of up to 23 ft (7 m). Instrumentation support for

these testing devices includes a 64 channel data acquisition system capable of recording up to 2

mega samples per second, and a high-speed camera operating at up to 10,000 frames per second.

PTC staff members participate in explosive testing at various government test facilities.

The Protective Technology Center accesses various advanced computer systems resident at

Pennsylvania State University. The computer codes available for PTC research include:

ANSYS, MSC/Nastran, ABAQUS (Implicit and Explicit), MARC, DYNA3D, LS-DYNA3D,

and AUTODYN. A family of special codes for analysis of structural systems under explosive

28

loads was developed by PTC researchers and is available. Other codes for computational fluid

dynamics (CFD), and new codes for expedient blast load simulation are under development.

Personnel

Eight faculty members from the College of Engineering (Civil and Environmental Engineering,

Aerospace Engineering, Engineering Sciences and Mechanics), and the College of Information

Science and Technology support PTC research activities. Dr. Theodor Krauthammer resides as

the PTC Director and administers a staff of six administrative and technical staff. The Center

supports the research and advanced degree requirements for approximately 15 engineering

students.

Additional information about the PTC, and contacts are available at:

http://www.ptc.psu.edu

29

    DARPA PROGNET-MD PROJECT TEAM      H.S. Mahmassani, Principal Investigator Elise Miller-Hooks and Theodor Krauthammer, co-PI's        

    MARYLAND TRANSPORTATION INITIATIVE   PROTECTIVE TECHNOLOGY CENTER      H.S. Mahmassani, Director   Theodor Krauthammer, Director           

    Materials and   Structures

Real-Time Systems and Optimization      

            8 Specialists in      Bridges   Pavements H. Mahmassani              E. Miller-Hooks   Structural materials      R. Medina D. Goulias A. Haghani          C. Fu C. Schwartz     Bomb Blast Damage Assessment      B. Ayyub                  Structural Prediction and Reliability  

    Sensors and Probability, Reasoning, Inference          Communications and Decision-Making   Probabilistic reasoning                     P. Chang B. Ayyub            R. La H. Mahmassani          M. Wu e. Miller-Hooks            R. La                                                        

Figure 1

Team Composition by Specialty Area and Institutional Affiliation

30