University of Central Florida

STARS

Electronic Theses and Dissertations, 2004-2019

2012

A Framework For Process Data Collection, Analysis, And

Visualization In Construction Projects
Reza Akhavian
University of Central Florida

Part of the Civil Engineering Commons

Find similar works at: https://stars.library.ucf.edu/etd
University of Central Florida Libraries http://library.ucf.edu

This Masters Thesis (Open Access) is brought to you for free and open access by STARS. It has been accepted for
inclusion in Electronic Theses and Dissertations, 2004-2019 by an authorized administrator of STARS. For more
information, please contact [email protected].

STARS Citation
Akhavian, Reza, "A Framework For Process Data Collection, Analysis, And Visualization In Construction
Projects" (2012). Electronic Theses and Dissertations, 2004-2019. 2177.
https://stars.library.ucf.edu/etd/2177
A FRAMEWORK FOR PROCESS DATA COLLECTION, ANALYSIS, AND
VISUALIZATION IN CONSTRUCTION PROJECTS

by

REZA AKHAVIAN
B.S. University of Tehran, 2010

A thesis submitted in partial fulfillment of the requirements

for the degree of Master of Science
in the Department of Civil, Environmental, and Construction Engineering
in the College of Engineering and Computer Sciences
at the University of Central Florida
Orlando, Florida

Summer Term
2012
© 2012 Reza Akhavian

ii
 ABSTRACT

Automated data collection, simulation and visualization can substantially enhance the

process of designing, analysis, planning, and control of many engineering processes. In

particular, managing processes that are dynamic in nature can significantly benefit from

such techniques. Construction projects are good examples of such processes where a

variety of equipment and resources constantly interact inside an evolving environment.

Management of such settings requires a platform capable of providing decision-makers

with updated information about the status of project entities and assisting site personnel

making critical decisions under uncertainty. To this end, the current practice of using

historical data or expert judgments as static inputs to create empirical formulations, bar

chart schedules, and simulation networks to study project activities, resource operations,

and the environment under which a project is taking place does not seem to offer reliable

results.

The presented research investigates the requirements and applicability of a data-driven

modeling framework capable of collecting and analyzing real time field data from

construction equipment. In the developed data collection scheme, a stream of real time

data is continuously transferred to a data analysis module to calculate the input

parameters required to create dynamic 3D visualizations of ongoing engineering

activities, and update the contents of a discrete event simulation (DES) model

representing the real engineering process. The generated data-driven simulation model is

iii
an effective tool for projecting future progress based on existing performance.

Ultimately, the developed framework can be used by project decision-makers for short-

term project planning and control since the resulting simulation and visualization are

completely based on the latest status of project entities.

iv
 ACKNOWLEDGMENTS

The present document is my final Master’s Thesis. This Thesis describes my graduate

research which began in Fall 2010 and concluded in Summer 2012.

I would like to thank many people in my life who have contributed to my academic

journey. Above all, my sincere appreciation is due to my advisor Dr. Amir H. Behzadan

for his continuous support, endless patience and unstinting encouragement during my

Master’s studies and in the course of this research. He is truly a dedicated teacher and

incredible advisor. Thank you for believing in me, giving me the opportunity to succeed,

and helping me to promote from an ordinary student to a curious and motivated

researcher.

My special thanks to my other committee members, Dr. Oloufa and Dr. Tatari who

offered guidance and support. I am always indebted to Dr. Oloufa for his motivational

lessons about construction engineering and management during the very first semesters I

began my studies in the United States.

There are some other people that I am deeply grateful to them for their presence in my

life; my lovely parents Hossein Akhavian and Maryam Batebi who were both teachers

and I have been and will be always their student. Their love and patience have been

always a source of support for me, particularly during the days I am far away from them.

I would also like to appreciate the presence of Negin Alimohammadi, my love, wonderful

v
friend, and great source of emotional support and motivation during my Master’s studies.

She never left my hand during hard days.

I am also thankful to all my friends, especially Saeed Hadian and Sina Zel Taat for their

help and friendship and all nice experience we had during these years.

vi
 TABLE OF CONTENTS

LIST OF FIGURE............................................................................................................... x

LIST OF TABLES ........................................................................................................... xiii

CHAPTER 1: INTRODUCTION ....................................................................................... 1

1.1 Thesis Statement .................................................................................................. 1

1.2 Research Motivation ............................................................................................ 1

1.3 Research Contributions ........................................................................................ 3

1.4 Research Objective and Project Tasks ................................................................. 5

1.5 Organization of the Thesis ................................................................................... 6

CHAPTER 2: LITERATURE REVIEW ............................................................................ 8

2.1 Automated Data Collection .................................................................................. 8

2.2 Simulation in Construction................................................................................. 13

2.3 Real Time Simulation......................................................................................... 14

2.4 Visualization in Construction ............................................................................. 16

CHAPTER 3: DYNAMIC DATA-DRIVEN APPLICATION SYSTEM (DDDAS) ...... 18

3.1 Overview ............................................................................................................ 18

3.2 General Applications .......................................................................................... 20

3.3 Developed Framework Based on DDDAS......................................................... 22

vii
CHAPTER 4: VISUALIZATION WITH OPENSCENEGRAPH (OSG)........................ 25

4.1 Overview ............................................................................................................ 25

4.2 Implementation in This Research ....................................................................... 31

CHAPTER 5: DEVELOPED FRAMEWORK ................................................................. 33

5.1 System Architecture ........................................................................................... 33

5.2 Real Time Data Collection ................................................................................. 34

5.2.1 Serial Port Communication in .NET Environment ..................................... 36

5.2.2 Data Collection Using LabVIEW ............................................................... 40

5.3 Automated Data Classification and Analysis ..................................................... 45

5.3.1 VI Structure for Data Analysis Module ...................................................... 49

5.4 Data-Driven Simulation ..................................................................................... 50

5.5 Pre- and post-processed Animations .................................................................. 52

5.6 What-If Analysis ................................................................................................ 55

5.7 Decision-Making and Dynamic Feedback ......................................................... 56

CHAPTER 6: Laboratory Scale Experiments and Results ............................................... 58

6.1 Preliminary Results ............................................................................................ 58

6.1.1 Experiment Tools and Peripheral Devices .................................................. 58

6.1.2 Single Object Data Collection and Visualization ....................................... 61

6.1.3 Double Object Data Collection and Visualization ...................................... 64

viii
 6.2 Comprehensive Example: Data-Driven Simulation ........................................... 69

CHAPTER 7: CONCLUSIONS AND FUTURE WORK ................................................ 78

7.1 Conclusions ........................................................................................................ 78

7.2 Future Work ....................................................................................................... 81

APPENDIX A: C++ ALGOTIRHMS FLOWCHARTS .................................................. 83

APPENDIX B: LabVIEW GRAPHICAL PROGRAMMING AND ALGORITHMS .... 90

REFERENCES ................................................................................................................. 98

ix
 LIST OF FIGURE

Figure ‎3.1: Basic Concepts of DDDAS ............................................................................ 19

Figure ‎3.2: Mandel et al. Developed DDDAS for real time modeling of Wildfire .......... 21

Figure ‎3.3: Developed DDDAS in the Presented Research .............................................. 24

Figure ‎4.1: A Sample Scene Graph Hierarchy.................................................................. 27

Figure ‎4.2. Hierarchical Scene Graph and Relationships between Different Nodes ........ 29

Figure ‎5.1: System Architecture of the Developed DDDAS Framework ........................ 34

Figure ‎5.2: Datagram Structure of the PNI TCM Prime Orientation Tracker .................. 37

Figure ‎5.3: C++ Orientation Tracker Serial Communication Class ................................. 39

Figure ‎5.4: Flowchart of 3D Orientation Tracker Serial Communication Process........... 40

Figure ‎5.5: A VI Consists of a Front Panel and a Block Diagram.................................... 42

Figure ‎5.6: Developed VISA Interface Structure ............................................................. 43

Figure ‎5.7: Simplified Layout of an Earthmoving Operation ........................................... 46

Figure ‎5.8: Activity Durations Based on the Variation of Equipment Body Orientation

with Respect to Time (RB = Raise Bucket, LT = Load Truck, LB = Lower Bucket, RTB

= Raise Truck Bed, P = Put, LTB = Lower Truck Bed) ................................................... 47

Figure ‎5.9: Single VI Containing Data Acquisition and Data Analysis Functions .......... 50

Figure ‎5.10: DES Model of a Typical Earthmoving Operation ........................................ 51

Figure ‎6.1: Overall Arrangement of Experiment Tools and Devices ............................... 59

x
Figure ‎6.2: A Prime 3D Orientation Tracker Mounted on a Model Excavator with

Definitions of Yaw, Pitch and Roll Angles ...................................................................... 60

Figure ‎6.3: Orientation Tracker Mounted on a Loader's Boom ........................................ 61

Figure ‎6.4: Front Panel of Data Collection VI for a Single Loader .................................. 62

Figure ‎6.5: Real Time Display of Loader's Boom Movements and Corresponding 3D

Animation Generated in Real Time .................................................................................. 63

Figure ‎6.6: Orientation Trackers Mounted on a Loader's Boom and a Truck’s Bed ........ 65

Figure ‎6.7: Front Panel of Data Collection VI for a Double Object Experiment ............. 66

Figure ‎6.8: Partial View of the Extensive Block Diagram Developed in this Research .. 67

Figure ‎6.9: Real Time Display of Loader's Boom and Truck’s Bed Movements and

Corresponding Animations ............................................................................................... 68

Figure ‎6.10: Experiment Layout of a Model Dam Construction Scenario ....................... 69

Figure ‎6.11: DES Model of Rock Hauling Activity ......................................................... 70

Figure ‎6.12: STROBOSCOPE Simulation Input File ...................................................... 71

Figure ‎6.13: STROBOSCOPE Simulation Output File .................................................... 72

Figure ‎6.14: Developed VI for Data Collection and Analysis for Rock Hauling Example

........................................................................................................................................... 74

Figure ‎6.15: STROBOSCOPE Simulation Input File Containing Updated Activity

Durations ........................................................................................................................... 75

Figure ‎6.16: STROBOSCOPE Simulation Output File Based on the Updated Durations 76

Figure A. 1: CreateAnimationPath() Function Flowchart ................................... 86

Figure A. 2: CreateMovingModel() Function Flowchart .......................................... 87

xi
Figure A. 3: CreateModel() Function Flowchart......................................................... 88

Figure A. 4: Main() Function Flowchart ......................................................................... 89

Figure B. 1: A Customized VI - The Upper Window is the Front Panel and the Bottom

Window is the Block Diagram .......................................................................................... 92

Figure B. 2: VISA Open Opens the Specified Port by the VISA Resource Name ............. 93

Figure B. 3: VISA Write Writes Data to the Specified Port by the VISA Resource Name 94

Figure B. 4: VISA Read Reads Data from the Specified Port by the VISA Resource Name

........................................................................................................................................... 94

Figure B. 5: VISA Close Closes the Specified Port by the VISA Resource Name ............ 94

Figure B. 6: A Series of VISA Functions and Their Connections as Used in this Research

........................................................................................................................................... 94

Figure B. 7: Requested Data Classified from Cluster of Real Time Orientation Data ..... 95

Figure B. 8: Unbundled By Name Function that Returns Cluster Elements Whose Names

Have Been Specified ......................................................................................................... 95

Figure B. 9: Greater? Function Returns True If x Is Greater than Y - This Function Was

Used to Detect Data Exceeding a Specified Threshold .................................................... 96

Figure B. 10: Tick Count Function That Returns the Value of a Timer – This Function

Was Used to Measure the Duration of Each Activity ....................................................... 96

Figure B. 11: Build Array Function to Store Activity Durations in a Numerical Array... 96

Figure B. 12: Statistics Tool Returns the Specified Statistical Characteristics of Input

Arrays ................................................................................................................................ 97

xii
 LIST OF TABLES

Table 2.1: Previous research on remote data collection applications in construction

jobsites .............................................................................................................................. 12

Table 6.1: Manufacturer’s Specifications of Prime 3D Orientation Tracker.................... 60

Table ‎6.2: Comparison between Estimated Durations and Actual Durations Based on

Real Time data .................................................................................................................. 77

xiii
 CHAPTER 1: INTRODUCTION

1.1 Thesis Statement

The efficiency of various construction tasks including the planning and control of

equipment operations can significantly increase if adequate operational data is collected

in real time, analyzed, and effectively integrated into the decision-making process. This

real time filed data stream can be used as a reliable source to modify project plans,

validate and improve existing control metrics, and update the underlying parameters of

computer models (e.g. simulation and visualization) describing the interactions between

different project resources, all in an effort to assist project personnel in predicting the

future performance given the current conditions of resources on the ground.

1.2 Research Motivation

Resource planning and control at the operations level are critical components of

managing the performance of ongoing activities in a construction site [1]. A

comprehensive operations level plan can help project decision-makers and site personnel

foresee potential problems such as spatial conflicts and resource underutilization even

before the actual operation takes place. This will also help save effort that would have

otherwise been put on reworks, resolving conflicts, and performing change orders, which

1
will ultimately translate into significant savings in project time and cost. For example,

Cox et al. [2] suggested that rework is typically responsible for 6-12% of the overall

expenditure for a typical construction project. Construction Industry Dispute Avoidance

Task Force (DART) reported that annually, more than $60 billion was spent on change

orders in the United States [3]. Also, according to the Federal Facilities Council (FFC), in

10-30% of all construction projects serious disputes are estimated to arise with a total

cost of resolution between $4-12 billion each year [4]. One of the major impediments of

effective project planning is managing a large volume of information including inputs

from alternative designs, material properties, labor productivity, equipment

specifications, and work schedules. This will become even more sophisticated when the

dynamics of the construction project creates several layers of uncertainty that can range

from internal factors (e.g. project time and cost variations, equipment breakdowns,

contractor claims) to external events (e.g. weather conditions, financial market stability).

Computer applications have thus evolved during the past several years to facilitate the

process of project planning by providing a convenient and reliable means for modeling,

simulating, and visualizing project activities [5, 6, 7, 8, 9, 10, 11]. In order to create

reliable computer models of a future construction project during the planning stage, one

needs to carefully examine every detail of the operations within that project, and identify

major events and processes that will potentially impact the outcome of each operation.

Once such events and processes are identified, attributes such as resource consumption

levels and activity durations should be determined. For a small operation, this can be

done in a relatively short period of time using existing numerical tools and statistical data

2
from past projects. However, as the size of the operation increases and with the

introduction of more resources and activities, creating a simulation model that

realistically represents the actual operation becomes a tedious if not an impossible task

[12]. This is mainly due to the fact that collecting accurate and reliable field data from

ongoing activities and resource operations, and integrating the collected data into the

planning process turns into a challenging task. In addition, the uncertainties caused by

unforeseen site conditions, equipment breakdowns, work delays, and the evolving nature

of a construction project may slow down or interrupt the progress of data collection. Even

if all such data is collected, handling a large volume of information in a single platform

can prove to be time and labor intensive. As a result, it is very likely that the modeler

uses strict rules, simplifying assumptions, and rigid design parameters inside the model to

streamline the modeling process. These may seriously impact the accuracy of the model

in representing the dynamics of the project which will ultimately be detrimental to the

reliability of the model for verification and validation purposes [13].

1.3 Research Contributions

Traditional simulation paradigms employ static data and information available from

similar projects and operate under a given set of system design parameters (e.g. activity

precedence relationships, duration distributions) [14]. In the absence of a methodology

that facilitates real time field data collection, most project decision-makers rely on readily

available project information and subjective personal judgments when evaluating

uncertainties and forecasting future project performance [12]. Recently, advances in

3
automation and information technology resulted in new approaches for collecting and

managing construction work data. In particular, automated tracking systems have been

evolved to collect necessary information about the position of construction resources for

different purposes [15, 16, 17]. Timely use of field data to determine the location and

status of resources (e.g. construction equipment and personnel) helps in describing the

context surrounding the operations and therefore is valuable for monitoring the workflow

of activities during these operations. Also, field data supports operational decisions and

helps predict the performance of a construction system based on the latest project status.

Another valuable implication of field data acquisition is the application of the collected

data in creating visual representations at different levels of detail corresponding to

various operations on a construction site. Visualizing field data has been demonstrated to

have many applications such as maintenance crew training [18], safety management [19],

and damage prevention [20]. But from the point of view of planning, monitoring, and

control, 3D visualization not only does offer a convenient tool for decision-makers to get

a real insight of what is exactly happening in a jobsite (particularly for operations that are

hard to quantify or represent in a parametric model), but also is a of substantial value for

verification and validation of the underlying simulation model(s). This is especially

important because decision-makers often do not have the time and knowledge to confirm

the accuracy and validity of simulation models and thus do not usually rely on the results

obtained from such models [10]. In addition, visualization assists in investigating events

that are hard to be quantified in a definitive manner, but yet can affect the final outcome.

Examples of such events include work zone overcrowding due to simultaneous execution

4
of different trades in building construction, safety problems, and potential for physical

collisions.

The benefits of construction field data collection, simulation, and visualization have been

investigated in isolated cases in the past. However, the potential of these three promising

techniques when integrated in a single framework that facilitates the process of short-

term planning and control of construction projects in operations level has not yet been

explored [21]. Hence, the presented research is mainly motivated by this need and is

aimed to fill this gap by investigating the requirements and applicability of an integrated

framework that uses the paradigm of dynamic data-driven simulation to address the

problem of short-term operational level planning and control. The underlying concepts

and applications of dynamic data-driven simulation, which is also referred to as dynamic

data-driven application system (DDDAS) are introduced in Chapter 3.

1.4 Research Objective and Project Tasks

The overall objective of this study is to design a framework for integrating field data

collection, data analysis, visualization and simulation for short-term decision-making in

construction projects. In order to achieve this objective, the following research tasks were

identified and successfully completed:

 Investigate the requirements and design a functional system to collect real time

data from equipment involved in different construction processes.

5
  Build data classification and analysis methods to provide orderly data and link

them to specific activities describing the status of construction equipment.

 Develop an algorithm for creating 3D pre-processed visualizations of concurrent

construction equipment activities.

 Conduct statistical analysis on data to obtain and update the probabilistic

distributions describing the duration of individual field activities within a

simulation model corresponding to the actual operations.

1.5 Organization of the Thesis

The following Chapters of this Thesis are shaped around the concepts, details, and

implementation of the research tasks listed above. This Thesis is divided into seven

Chapters. In particular:

 Chapter 1: Introduction – This Chapter contains the Thesis statement, identified

gaps that motivated this research, the novel approach that this study adopts to

address the identified gaps, and the overall objective and tasks defined and

accomplished in this project.

 Chapter 2: Literature Review – A review of previous related research and state-of-

the-art studies in the realm of automated data collection, simulation in

construction, visualization in construction, and using the advantages of real time

simulation in construction projects is presented in this Chapter.

6
 Chapter 3: Dynamic data-driven application system (DDDAS) – This Chapter

introduces the basic concept of a relatively new paradigm for data-driven

simulation and outlines its application in various field of science and engineering

and emphasis on its application in the context of presented study.

 Chapter 4: Visualization with OpenSceneGraph (OSG) – Detailed description and

technical aspects of the visualization toolkit that has been used in this research is

presented in this Chapter and it is shown that how the proposed methodology

benefits from employing these concepts.

 Chapter 5: Developed Framework – The overall system architecture of the

developed framework is introduced in this Chapter and individual components of

the system and their tasks are discussed in detail.

 Chapter 6: Laboratory Scale Experiments and Results – This Chapter

demonstrates the validity and applicability of the presented methodology by

presenting preliminary laboratory scale experiments and resulted outcomes.

 Chapter 7: Conclusions and Future Work – A discussion about the identified gaps

in knowledge and the developed research methodology for addressing these gaps

is presented in this Chapter and future research for further development of the

presented framework is described.

7
 CHAPTER 2: LITERATURE REVIEW

In Chapter 1, a general introduction to the research was presented and the motivation,

potential contributions, research objective, and project tasks were described in details.

The presented research aims to address the gaps identified in the current body of

knowledge (as described in Chapter 1) through investigating the potentials and

opportunities provided by emerging innovations in engineering instrumentation and

computation. In this Chapter, a comprehensive review of recent research efforts and

current demands in the areas of automated data collection, and visualization and

simulation within the construction engineering and management domain will be

conducted, in an effort to put the presented work into context and demonstrate its

potentials in addressing some of the longstanding challenges faced by the construction

research community.

2.1 Automated Data Collection

Collecting accurate and reliable data is one of the most critical components of every

decision support system. Data captured manually using traditional onsite data collection

techniques can be outdated, inaccurate, or missing certain pieces [22, 23, 24, 25, 26].

McCullouch indicated that field supervisory personnel spend on average 30%-50% of

their time on recording and analyzing filed generated data[23]. Saidi et al. [27] stated that

despite the recent advancements in construction measurement and sensing technologies,

having accurate and updated information about the status of construction operations

8
remains an issue in the construction industry. As a result, automated data collection and

resource location tracking techniques have received credibility over the past several

years, as they facilitate processes including but not limited to resource management,

productivity analysis, quality control, and monitoring workflow processes. To this end,

work still needs to be done in order to take advantage of such technologies when

planning activities at early stages of a project where the scope of the work and the

dynamics of the project environment are still evolving.

Automated resource (personnel, equipment, materials) tracking has been the subject of

many studies in construction and facility management [15, 28, 29, 30, 31, 32]. Resource

location tracking applications use different techniques for indoor and outdoor

environments. A variety of outdoor and indoor location tracking technologies exist with

significantly different characteristics, infrastructure, and device requirements [16]. Radio

Frequency Identification (RFID), for example, has been increasingly used for tracking

purposes in construction jobsites. RFID systems use tags and a reader which sends radio

frequency signals to read data from the tags. One of the early attempts in using of RFID

in construction industry was made by Jaselskis et al. [33]. They proposed RFID for

tracking high-valued materials on construction jobsites. Song et al. [28] used RFID to

automate the task of tracking the delivery and receipt of fabricated pipe spools in lay

down yards and under shipping portals. Since RFID readers and tags do not require line-

of-sight, the readers can detect several tags at a time, and the tags can function properly

in harsh conditions. However, the short reading range which mostly is a function of the

communication frequency can be an obstacle for the use of RFID systems in large

9
construction sites [34]. Researchers have also used the Global Positioning System (GPS)

for its capability in tracking construction labor and equipment in outdoor environments

and construction sites [16, 35, 36]. GPS is an outdoor satellite-based worldwide radio-

navigation system formed by a constellation of 24 satellites, ground control stations, and

end users [37]. To address the challenge faced by equipment operators who have limited

field view and depth perception when they control equipment remotely with video

cameras, Oloufa et al. [35] developed a system for collision detection and vehicle

tracking by using differential GPS, wireless, and web-based technologies. The most

important impediment in using GPS is that its functionality is to the most extent, limited

to outdoor environments since a clear line-of-sight between the satellites and the GPS

receiver is always needed. More recently, there have also been some attempts in

combining RFID with GPS technology [31, 38]. Jang et al. [15] introduced an Automated

Material Tracking (AMTRACK) system based on ZigBee localization technology to

overcome the drawbacks of GPS and RFID systems in terms of accuracy and cost.

Another technology that has been studied for automated tracking is Ultra Wide Band

(UWB). Teizer et al. [39] developed an UWB data collection tool for work zone safety

management and location tracking. In an indoor environment, where Global Navigation

Satellite System (GNSS) data is not available, indoor positioning technologies are used.

RFID and UWB can be used in both indoor and outdoor environments. GPS, as stated

before, has generally developed only for outdoor environments. However, another

technology called indoor GPS has recently emerged which is not satellite-based [32].

Wireless Local Area Network (WLAN) is another technology used for indoor tracking

10
and localizations [32]. Inertial Navigation Systems (INS) such as accelerometers and

other systems such as Bluetooth, Infrared, and Ultrasonic are other examples of indoor

localization technologies [40].

Another category of localization and tracking technologies is motion-based. Generally,

motion-based devices sense motion and its attributes such as velocity, acceleration, and

heading directions. For position sensing, inertial navigation systems (INS) or inertial

measurement unit (IMU) are constructed using a combination of gyroscopes,

accelerometers, and magnetometers [41]. Using IMU the current state of the target in

terms of location, speed, and heading direction can be determined by using state

estimation algorithms using the information provided by gyroscopes and accelerometers.

Behzadan et al. [42] developed an augmented reality (AR) hardware framework in which

they used orientation trackers capable of measuring compass heading using magnetic

field sensors.

In addition to the sensor-based data acquisition technologies described above, vision-

based tracking has lately started to gain credibility among researchers. In a recent study,

Brilakis et al. [43] presented an automated framework for vision based tracking using two

cameras. Although this method seems to overcome the disadvantages of existing sensor-

based techniques such as limited coverage area and dependence on preinstalled tags on

the objects, it is still much costly and requires a more involved maintenance and

calibration. Table 1 summarizes the existing tracking and localization techniques. As

Table 1 suggests and to the author’s best knowledge, the application of real time data

11
collection for the purpose of planning and monitoring of equipment motions has not yet

been investigated.

Table 2.1:

Previous research on remote data collection applications in construction jobsites

Study Application Technology

Ergen et al. 2007 Material Tracking RFID

Song et al. 2006a Material Tracking RFID

Behzadan et al. 2008 Personnel Tracking GPS

Caldas et al. 2006 Material Tracking GPS

Grau and Caldas 2009 Material Tracking RFID + GPS

Ergen et al. 2007 Material Tracking RFID + GPS

Jang et al. 2007 Material Tracking ZigBee

Teizer et al. 2007 Safety UWB

UWB/Indoor
Khoury and Kamat 2009 Tracking Mobile Users
GPS/WLAN/

Behzadan et al. 2008b Mobile AR Hardware IMU

Brilakis et al. 2010 Project Entities Vision Based

12
2.2 Simulation in Construction

Simulation is a valuable tool for effective construction planning and management mainly

due to the presence of operational and decision-making uncertainties in most construction

processes. However, a large amount of research previously conducted in construction

simulation have one thing in common: almost all of them assume that when the

simulation model is created, sufficient data with adequate level of detail is readily

available mainly in form of historical records from similar projects or expert thoughts

and judgments (which may prove to be subjective). It is clear that providing such input,

there is almost no guarantee that the generated output reliably reflects the expected

performance of project entities, since the bulk of the data do not particularly belong to

that project.

Among several existing methods for modeling construction operations, discrete-event

simulation (DES) has gained a lot of interest by researchers since almost every

construction operation can be effectively broken down to and modeled as a network of

discrete activities, each consuming resources (personnel, material, and equipment) to be

completed [44]. DES models provide an effective means to establish logical relationships

between activities within a project which compete over and make use of available and

often scarce resources. The introduction of CYCLONE [45], marked the beginning of a

new era in modern construction simulation research. CYCLONE aimed to simplify the

modeling of processes that are cyclic in nature. Subsequently, many attempts were made

to develop different simulation systems based on CYCLONE. Examples include

13
INSIGHT [6] that enabled videotaping of field operations, and extracting and analyzing

videotaped data to obtain estimated values for the productivity of the system and its

components. Further studies explored the applicability of object-oriented and modular

programming in developing simulation systems. Examples of such efforts include

MODSIM [46] capable of translating a simulation code to the C language for compilation

and linking. In another effort, STROBOSCOPE [47] an extensible programmable system

capable of modeling complex construction operations was introduced. Later, an activity-

based construction modeling and simulation method called ABC [9] was developed.

A DES system called COOPS was introduced by Liu and Ioannou [7] which used object-

oriented design for simulation. Martinez and Ioannou [44] examined DES systems based

on three characteristics: application breadth (general or special purpose), modeling

paradigm (process interaction versus activity scanning), and flexibility (i.e.

programmability). Also, a new simplified DES approach or SDESA was developed by Lu

[48] for planning construction operations which can be used as a general-purpose

construction planning tool to track the performance of individual resources and handle

cyclic or looped processes.

2.3 Real Time Simulation

Real time simulation has been explored by researchers in several engineering and

scientific fields. For example, Hunter et al. [49] developed a simulation model based on

inflow data aggregated over a short time interval to create an accurate estimate of the

evolving state of transportation systems. In another example, Tavakoli et al. [50]

14
suggested a generic simulation platform for real time DES modeling in healthcare and

manufacturing applications. Also, a yard crane dispatching algorithm based on real time

data driven simulation was proposed by Guo et al. [51] to solve the problem of yard crane

job sequencing by minimizing the average vehicle waiting time. In the construction

domain, however, despite previous work in real time data collection and processing, very

limited amount of research has been done in effectively incorporating field data into an

existing simulation model for short-term planning and control of the same operations.

Chung et al. [52] suggested using Bayesian techniques to update the distributions of input

parameters for tunnel simulation by “manually” collecting project data from a tunneling

project on a bi-weekly basis and using the collected data to improve simulation input

models. Also, Song et al. [12] described a framework of real time simulation for short-

term scheduling of heavy construction operations and developed a prototype system for

asphalt hauling and paving projects.

To this date, only a limited number of previous projects investigated the planning and

control of engineering systems through real time simulation using the latest changes in

activity patterns and interactions. In the absence of a simulation system that is not using

an accurate input data, the resulting output should be evaluated with prudence. Abourizk

et al. [53] discussed that random input tends to propagate to the output of the simulation

model. They warned of using improper molding of input data through demonstration of

the sensitivity of the output parameters as well as resource utilization to the input model

utilized.

15
2.4 Visualization in Construction

The role of visualization in construction engineering and management has been generally

limited to the design of construction products using 3D CAD modeling or the

demonstration of how an entity evolves over time using 4D CAD applications.

Visualizing the actual interactions between resources (including personnel, equipment,

and materials) that result in a constructed facility has received a very little attention [54,

55]. Almost all of the efforts in this area concentrated on visualization of “simulated”

construction operations. Schematic modeling such as DISCO, iconic animation [56], and

2D system visualizations such as PROOF [57] are some of the first generation systems

intended for visualizing simulated construction operations. More recently, Kamat and

Martinez [10] presented VITASCOPE as a general-purpose, user-extensible 3D

animation system for visualizing simulated processes in smooth, continuous, 3D virtual

worlds. Behzadan and Kamat [11] designed and implemented ARVISCOPE, an

augmented reality (AR)-based mobile visualization system that allowed dynamic

visualization of simulated operations in outdoor environments using an external scripting

language.

Confirming the veracity and validity of the simulated construction operation is a major

goal in creating post-processed visualization systems [54]. Nevertheless, verification and

validation of the simulation model can be conveniently performed if a similar, yet pre-

processed animation representing the actual ongoing activities exists. Having both pre-

16
and post-processed animations in a similar visualization environment side by side,

facilitates the comparison between the real world systems and the model.

17
 CHAPTER 3: DYNAMIC DATA-DRIVEN APPLICATION SYSTEM
(DDDAS)

3.1 Overview

As described in previous Chapters, a major requirement of a robust decision support

system capable of offering real time analysis of concurrent construction operations is the

ability to provide decision-makers with a reliable basis to predict upcoming system

performance by using the incoming data streams to simulate the actual operations. To

achieve this, the concept of a relatively new simulation paradigm often referred to as

dynamic data-driven application system (DDDAS) and its potentials in the realm of

construction engineering and management was investigated in this research.

A DDDAS model is sought to dynamically measure site data in form of a new

information layer, integrate the collected data with the corresponding simulation model to

constantly adapt the model to the dynamics of the construction system, and constantly

update it based on the latest collected operational data [58]. Although the dynamic nature

of many complex systems such as those in construction requires simultaneous injection of

collected data into the simulation model in response to the evolving conditions, many

computational models used to date only allow fixed data inputs while the simulation is

launched [14].

Initially, DDDAS was conceived by the National Science Foundation (NSF) in 2000

following two catastrophic events. The first was the missed prediction of the track and

18
magnitude of a storm that blanketed a number of cities from South Carolina to New

England in January 2000, and the other was the failure of a simulation model to predict

the propagation and behaviors of a fire near Los Alamos National Laboratory in May

2000 mainly due to the changing nature of fire and consequently, the inability of

emergency response agencies to take appropriate actions to limit its propagation [59].

Scientists believed that such miscalculations were due to computer simulation models

that were unable to incorporate real time changing conditions on the ground [59].

Recently, advances in computational technologies for data collection, analysis, and

modeling provided the necessary tools for accurate measurement and injection of

necessary data into corresponding simulation models and enabled the development of the

DDDAS. Figure 3.1 is a schematic diagram showing the basic components of a DDDAS

(as introduced by the NSF) consisting of the following modules: data acquisition tools,

simulation model, dynamic data control and acquisition, and visualization and human

interface.

Figure ‎3.1: Basic Concepts of DDDAS

19
Data acquisition tools refer to field equipment used for remote data collection such as

wireless sensors and instruments. Simulation model represents those models that need to

be updated based on the stream of the incoming data. Dynamic data control and

acquisition includes algorithms for data analysis used to prepare data for representation

and input modeling. Finally, visualization and human interface refers to the human expert

interaction to steer the model (if needed) and determine answers to critical decision-

making problems based on the simulation results. These components and their

interactions, as stated before, symbolize a rudimentary representation of the DDDAS

concept and most of the platforms, including what was developed in this research are

built upon this basic premise.

3.2 General Applications

As an emerging and promising area of research, DDDAS is gaining credibility among

scientists and researchers in various fields of study while posing challenges in

mathematical algorithms, systems software and data collection. Nevertheless, engineering

problem solving in general and construction engineering and management in particular,

are yet to benefit from the opportunity offered through employing this concept.

In a research aimed at forecasting the wildfire behavior, Mandel et al. [60] proposed a

DDDAS that included coupled weather and fire numerical models, an automated data

acquisition and control module, visualization and user interface module, and a

communication infrastructure. In their developed system, data acquisition and control

module directs data to the numerical models where multiple simulations are running.

20
Synchronously, the simulation inputs are adjusted based on the actual measurements of

the field. Also, simulation results are presented through visualization and user interface

module to the user in order to determine alternative firefighting scenarios. In this

example, data collection was performed using wireless network sensors and cameras

mounted on airplanes. Personal digital assistant (PDA) devices were also used as

convenient visualization and user interface tools while numerical model ran on a remote

supercomputer. Figure 3.2 shows how this particular application has been built upon the

basic DDDAS concept previously shown in Figure 3.1.

Figure ‎3.2: Mandel et al. Developed DDDAS for real time modeling of Wildfire

In another research, Douglas et al. [61] investigated the application of a DDDAS in an

environmental engineering set up. They considered the case of a contaminant spill

occurring near a clear water aquifer. Sensors were used to measure where the

contaminant was, how and in what direction it was moving, and to monitor the

environmental impacts of the spill. Numerical simulation procedures for multi-scale

interpolation were used in order to map sensor data and to continuously update the

21
simulation model. The study demonstrated that frequent updating of the sensor data in the

simulation considerably improved the prediction results.

Gaylor et al. [62] indicated that in case of a crisis, management should make decisions in

order to react to dynamic uncertain conditions. In this regard, having access to real time

data in a format, that can be readily understood and acted upon, is critical. Therefore,

they applied the concept of DDDAS to support emergency medical treatment decisions in

crisis conditions. Their complex dynamic environment fed and responded to a stream of

real time data coming including positional data coming from GPS trackers mounted on

ambulances and vital signs sensors mounted on patient body.

The NSF has also proposed some applications in workshops held for introducing

DDDAS. An interesting DDDAS application is traffic light control, since there are

always two significant variants: whether the plan is to minimize or to maximize the

number of red lights encountered. As stated by NSF 2000 [59], the ultimate goal should

be to continuously optimize the timing of the traffic lights. Using DDDAS and based on a

sophisticated model, data generated by sensors embedded under streets and also other

factors such as weather conditions can assist in predicting and optimizing the flow of

vehicle movements [59].

3.3 Developed Framework Based on DDDAS

Unlike several other scientific fields, the idea of DDDAS has been given very little

attention in engineering simulation in general, and has not been widely applied to

22
construction research in particular. DDDAS enables a more accurate prediction of how a

dynamic construction system will behave in the future based on the current status of its

constituents (i.e. resources). Therefore, construction projects can benefit from this novel

paradigm if necessary infrastructure, algorithms, and tools to launch robust DDDAS

platforms are effectively designed and implemented.

Traditionally, there has been a major disconnect between DES modeling (which is mainly

conducted at the planning stage) and the actual site dynamics (during the construction

phase). Incorporating the concept of DDDAS into the modeling process can help

significantly improve conventional DES modeling. For example, more realistic activity

parameters (e.g. probabilistic duration distributions, dependencies) can be obtained by

measuring data collected from different pieces of equipment involved in that activity. In

short, DDDAS facilitates the process of tailoring an existing DES model to better meet

the evolving conditions of the real system using the latest data as input to the

corresponding simulation model.

The DDDAS technique designed and implemented in this research captures sensor-based

real time data from resources on a jobsite, classifies and analyzes the collected data to a

meaningful format for the following modules, incorporates the analyzed data to update

the corresponding DES model, and creates an exact dynamic 3D visualization of the

ongoing operations using the collected data, all in an effort to assist project decision-

makers in short-term operations planning and control [63]. Figure 3.3 illustrates a

simplified diagram of the developed DDDAS in this research.

23
 Figure ‎3.3: Developed DDDAS in the Presented Research

As shown in this Figure, the framework built upon the general concept of DDDAS. Real

time collected data from ongoing construction operation move through a data analysis

module to provide required information for updating the data-driven simulation model.

Also, a visualization system providing a concurrent 3D animation of ongoing activities

serves as the human interface module. Detailed description and system architecture of the

developed framework can be found in Chapter 5.

24
CHAPTER 4: VISUALIZATION WITH OPENSCENEGRAPH (OSG)

4.1 Overview

In this research, OpenSceneGraph (OSG) which is built upon the industry standard

OpenGL graphics library is used inside the .NET environment to create pre-processed

animations of ongoing equipment activities and to link each and every object motion

inside the animation to the collected field data that represent the actual motion of that

object. This Chapter provides technical details about the algorithms developed to create

3D animations using CAD models of construction equipment.

A scene assembled from discrete components that can be dynamically manipulated,

provides essential means for creating a contextual animation. To facilitate the creation

and manipulation (i.e. positioning, orienting, and scaling) of objects in an assembled

scenes, the concept of scene graphs were implemented in this research. Generally, a scene

graph is a hierarchical organization of shapes, groups of shapes, and groups of groups

that together construct a scene [64]. Computer graphics implementations build upon the

concept of scene graphs release the end user from implementing and optimizing low level

graphical programming and complexities involved in rendering process of 3D objects in a

scene [41]. The scene graph application programming interface (API) provides a means

for constructing scenes that follow a hierarchical data structure of objects.

25
OSG is a collection of open-source libraries that provide scene management and graphics

rendering optimization functionality to applications. It has been written in ANSI C++ and

uses the industry standard OpenGL low-level graphics API [65]. Although there are a few

other scene graph-based libraries such as Performer, Open Inventor, and Java3D, this

research used OSG due to the fact that it is capable of reading various image file formats

which supports the prospect of designing a more generalizable visualization platform. At

the same time, OSG provides the functionalities required to describe a complex scene

using an object-oriented representation which releases the user from implementing and

optimizing low level graphical programming and facilitates rapid development of graphic

applications.

In OSG terminology, a node is an object that can be part of or entirely comprise a scene

graph. Each node as a collection of one or more values and methods compresses what is

required to be drawn. Root node is the highest level node to which all the elements of a

scene graph (directly or indirectly) are connected [66]. Each scene graph comprised of

nodes in a graph structure that are connected together via individual child-parent

relationships. The edges that connect the nodes describe a meaningful relationship that

exists between them. The root node is usually connected to intermediate grouping nodes

called internal or group nodes. These nodes commonly are responsible for 3D

transformations performing positioning (translation), orientation (rotation) and size

(scaling). Leaf nodes are the lowest level nodes that contain the geometrical description

of the components and are located at the terminus of a branch [67]. Figure 4.1 shows a

sample scene graph in which Jobsite is the root node. Scene sub-graphs are created and

26
attached to the root node to complete the scene structure by encapsulating the entire

jobsite. In Figure 4.1, sub-graphs Truck, Excavator and Terrain are all child nodes of

Jobsite. Also, nodes Excavator and Truck have their own child nodes at the lowest level

of the hierarchy.

Figure ‎4.1: A Sample Scene Graph Hierarchy

Using transformation nodes, each geometrical model is created in its own local

coordinate frame, stored as a leaf node in the scene graph, and appropriate placement of

the model in terms of position and orientation will be made inside the coordinate frame of

27
its parent node. Scene graph developers can manipulate the translation, rotation, and

(scale) of different nodes using transformation nodes.

The overall transformation of a child object relative to its parent node is obtained by

multiplying the individual matrices as follows:

1 0 0 Tx  1 0 0 0   Cos  0 Sin  0   Cos   Sin  0 0 S x 0 0 0

     
0 1 0 Ty 0 Cos   Sin  0 0 1 0 0 Sin  Cos  0 0 0 Sy 0 0
      
Parent
TChild
0 0 1 
Tz 0  Sin  Cos   
0  Sin  0 Cos  0  0 0 1 0  0 0 Sz 0
     
0 0 0 1  0 0 0 1 0 0 0 1 0 0 0 1 0 0 0 1

Where the first matrix shows the transformation of the child note with respect to its

parent node, the second, third, and fourth matrices, are the rotation about the local X, Y,

and Z axes, respectively, and the fifth matrix is the scale matrix. Considering a scene

consisting of a loader and a truck, Figure 4.2 shows the hierarchical scene graph and

relationships between different nodes.

28
 Figure ‎4.2. Hierarchical Scene Graph and Relationships between Different Nodes

Using the concept of scene graphs, if the angle of rotation of a child node about the X

axis of its parent node is γ, the default value for this angular motion can be set to zero to

represent the initial rotation matrix of the child node relative to the parent node, as

follows,

1 0 0 0 1 0 0 0
   
0 cos   sin  0 0 1 0 0
  

 
 0 
0 sin  cos  0 0 0 1 0
   
0 0 0 1 0 0 0 1

If a motion sensor capable of detecting angular motions is connected to the real object

being represented by the child node in the scene graph, as soon as the rotation angle

about the local X axis (also called the pitch angle) changes due to a change in the real

29
object’s orientation, the sensor determines the change, the collected value is used to

update the value of γ, and consequently the above rotation matrix is updated. For

example, the truck bed shown in Figure 4.2 is rotated upward by 45˚ from its initial

orientation. When this change is detected, the new pitch angle is used to update the

corresponding rotation matrix as follows,

1 0 0 0 1 0 0 0
   
0 cos   sin  0 0 cos 45   sin 45  0
     
 45  
0 sin  cos  0 0 sin 45  cos 45  0
   
0 0 0 1 0 0 0 1

This new rotation matrix will be then used to update the overall transformation matrix of

Parent
the child (bed) node relative to its parent (truck) node ( T Child ). The animation is updated

in each frame according to the overall transformation matrices of all the objects exist in

the scene.

One of the important requirements of any visualization system is providing a suitable

view of the described scene. OSG provides several utilities to arrange desired viewpoints

from which the viewer can watch the scene. The position and orientation of a viewpoint

can be manipulated while the scene is being displayed to achieve the desired view of a

scene graph or different views of the same scene graph. Moreover, it is possible to set

several viewpoints with different positions and angles to have various views of the same

scene graph as if depicting a single scene with different cameras installed in distinct

30
places [68]. In OSG, viewpoint definition is independent from the actual scene graph

representation and as such, there is no nodal representation for viewpoints.

Creating an animation of the scene objects requires that a dynamic relationship between

scene graph components is first established. Such relationship can be obtained through

dynamically manipulating the values of the transformation matrices in the scene graph.

Although the increments by which values in a transformation matrix change is discrete

over time, a realistic animation showing exact movements of real world objects need a

smooth transition between discrete points with the passage of animation time. OSG

provides complex mechanisms to achieve this goal through constant monitoring and

updating of all moving objects using frame updating algorithms [66]. Technical details of

these algorithms are beyond the scope of this study but can be found in [64, 67].

4.2 Implementation in This Research

Since OSG is a free open source toolkit, it allows access to code internals, thus providing

the opportunity to manipulate and modify the original content of the code to supplement

the rest of the framework developed in the .NET environment for the specific purposes of

this research.

Each of the articulated components depicted in the scene consists of separate parts

(nodes) created within different modeling packages such as 3D Studio™ (.3ds),

AutoCAD™ (.dxf), MicroStation™ (.dwg), and VRML (.wrl) that are stored in the user

computer. By connecting these nodes through assigning a special coordination relative to

31
the main origin of the scene, and also creating a meaningful child-parent relationship

between separate parts, each single component of the scene is created as a standalone

model that can be moved inside the animation as necessary. Also, the animation speed

which is the rate of dynamic increase/decrease of the values of transformation matrices,

and a desired coordination for the viewpoint should be specified.

Using the real time positional and orientation data from the sensors mounted on the target

objects (in real world), the developed algorithm stores the data in form of vectors as the

animation path for each solid, yet articulated entity in the scene. Thus, the animated scene

is capable of showing the actual movements of every real object using real time data

representing the translation and/or orientation of that object’s articulated parts, or the

object as a whole.

32
 CHAPTER 5: DEVELOPED FRAMEWORK

5.1 System Architecture

In Chapters 3 and 4, the technologies, concepts, and tools necessary to address the

challenges described in Chapter 1 were discussed in detail. The need for the presented

research was further justified by an extensive review of previously conducted studies in

Chapter 2. This Chapter outlines a detailed account of the individual components of the

framework and their interconnection in the context of the developed research

methodology.

Figure 5.1 depicts the higher level system architecture of the developed framework in

which the relationship between major building elements, as well as an overall view of

how raw operational data flows through the system, is eventually transformed into a

meaningful format, and used in different processes are illustrated. As previously

described in Chapter 3, the framework is built around the concept of dynamic data-driven

application system (DDDAS) and thus contains major components (modules) that were

previously illustrated in Figure 3.3. The following Subsections provide more details

about these components.

33
 Figure ‎5.1: System Architecture of the Developed DDDAS Framework

5.2 Real Time Data Collection

Banks [14] summed up the simulation environment from a data collection point of view

by indicating that data are rarely readily available and data collection is one of the most

important and difficult problems in simulation modeling. Given a dynamic simulation

modeling system, the problem could even get more complicated since the system

requires real time field data collection and integration. As a result, data acquisition is one

of the most challenging and computing intensive parts of a DDDAS given that it is

almost impossible to manually collect real time data in large projects. Depending on the

extent and complexity of a project, designing and implementing a reliable means to

acquire, communicate, and synchronize data from multiple sources may itself be a major

challenge. Real time data is used not only for updating and fine-tuning the model with the

34
latest changes occurring in the real system, but it also serves as the basis for model

validation and verification. Since the model needs to be continually updated, an

uninterrupted flow of input data is needed to reflect the latest changes in the status of

activities and resources. Therefore, developing and implementing a robust and automated

data collection infrastructure including sensing and communication technologies is

necessary.

In many construction projects, resources are in constant motion. Examples include dump

trucks transferring soil from a cut area to a fill location, crews laying reinforced concrete

rebars on a floor slab, and a tower crane lifting steel sections from a flatbed truck. As a

result, from a modeling perspective, capturing these changes in resource (e.g. equipment,

personnel, and material) positions is necessary. In addition to the positional data, most

construction equipment (e.g. cranes, excavators, shovels, loaders) have hinged moving

parts and thus, collecting the angles of orientation for these parts is also essential in order

to describe their motions. Such data can be acquired using orientation sensors that capture

three angles of rotation (i.e. yaw, pitch, and roll). In the presented research, orientation

data are captured and transmitted to the system in order to simulate and animate the body

configuration of construction equipment in real time.

In the course of this study, the data collection procedure was developed in two different

environments. Since both the manufacturer sample algorithms for the data collection

device and the open source code visualization toolkit, OpenSceneGraph (OSG), were

written in a .NET environment, a data collection system was initially designed by

35
creating an object-oriented platform in .NET environment. Later, due to the flexibility of

LabVIEW graphical programming for more sophisticated data analysis and processing

required in this research, a more efficient data collection procedure was developed in

LabVIEW using almost the same principles and algorithms originally created in .NET

environment. In this Section, first the overall functions and classes of the developed data

collection system in C++ are described. Subsequently, a general description of

implementing these functions in the graphical programming environment of LabVIEW

will be presented.

5.2.1 Serial Port Communication in .NET Environment

Serial is a standard device communication protocol used for transferring data to or from a

peripheral devices via computer serial ports [69]. In order to communicate with the data

collection devices used in this research, a serial port communication algorithm was

developed. A major factor in designing this algorithm was generalizability which in the

context of this research, is defined as the ability of the framework to communicate with a

variety of data collection devices without the need to significantly modify the

communication algorithms. Although the developed algorithm is to certain extents,

unique and has been tested with the PNI TCM Prime 3D orientation tracker, it benefits

from a generic structure that can be easily used to communicate with other data collection

devices that transmit data using the RS-232 protocol. RS-232 is a specification for serial

communication and is one of the most popular for sensor connections [69]. Since the

collected tracker data is in a binary format, the developed serial port communication

36
algorithm contains methods to decode the transmitted data and convert them to a

computer interpretable format.

The developed algorithm uses the advantages of object oriented programing in Microsoft

Visual C++ .NET environment. Using serial communication libraries, the initial

communication with the port is established, the port is opened, data (i.e. three orientation

angles) is received through the port, and the port is closed at the end of the experiment.

The orientation data coming through different brands of orientation trackers follow

different data transmission standards. The orientation tracker sensor, PNI TCM Prime

module, utilizes a binary data transmission protocol to obtain and extract the tracker data

that is transmitted over an RS-232 interface. Each data packet contains a component

called Frame Type ID that describes the content of the packet. Based on this ID, the

packet may contain each of the 3D rotational angles as well as the current temperature

(ranging from -40 ºC to +85 ºC). These values are stored in the packet Payload [70]. The

datagram structure of the PNI TCM Prime module is shown in Figure 5.2.

Figure ‎5.2: Datagram Structure of the PNI TCM Prime Orientation Tracker

37
Using the binary data provides the system with the advantage of fast data transmission.

However, this will in turn make the communication very sensitive to data corruption. As

a result, a mathematical transformation method called the Cyclic Redundancy Check

(CRC) is used to separate useful and corrupted binary data packets. CRC is applied to a

series of bytes and produces an integer result that can be used for error detection. After

data is received from the orientation tracker, the tracking application computes the CRC

value using the existing contents of the data packet and compares this value to the one

originally calculated when the packet was being constructed prior to transmission. If the

two values are not identical, the packet is considered as corrupted and will be disregarded

and the application waits for the next data packet. If the two values are equal, the data is

safe to be used and extracted into its components. Using a set of binary data manipulation

statements provided in the application programming interface (API) of the tracker device,

the numerical values for each of the orientation angles are obtained. The main

functionalities of the managed C++ class developed for acquiring orientation tracker data

through a serial port is shown in Figure 5.3.

38
 Figure ‎5.3: C++ Orientation Tracker Serial Communication Class

The PRIME::Initialize() function is called first to open the serial port and set up the port

properties (e.g. baud rate, data bits). Then, PRIME::Control() extracts the Payload piece

by piece. The number of requested angles (up to three) should be defined in this function.

Based on the number of requested data pieces, this function will be called consecutive

times and each time sends a control command to the tracker. In response, the tracker

sends a single packet containing binary values of the requested angles. For example, if all

three orientation angles (yaw, pitch, and roll) are required, the function will be called

three times and in return, the tracker sends binary values of three angles. Next,

PRIME::ReceivePacket() is called to receive the binary data packet. This is followed by a

call to PRIME::CRC() to check if the received data is error-free. If the data is not

corrupted, the contents of the packet will be extracted by PRIME::ParsePrimePacket().

This function stores the numerical value of the required angles in numerical variables

which will later be used to construct and display the real time animation of moving parts.

Finally, the PRIME::Shutdown() class will close the port. The flowchart in Figure 5.4

shows major steps in acquiring orientation data using the PRIME class introduced in

39
Figure 5.3. More information about all other functions and classes developed using

Microsoft Visual C++ can be found in Appendix A.

Figure ‎5.4: Flowchart of 3D Orientation Tracker Serial Communication Process

5.2.2 Data Collection Using LabVIEW

The presented methodology for data collection takes advantages of LabVIEW graphical

programming environment developed by the National Instruments (NI). This essentially

40
enables the creation of a standalone data collection and analysis framework that uses the

same functionalities employed in C++ but in a more efficient manner. LabVIEW, in

essence, is a system design platform that enables automating data collection and

measurements supporting a wide variety of sensors. As a powerful data collection tool,

LabVIEW abstracts much of the administrative complexity of computer programming

such as memory allocation and language syntax [71]. It was used in this research in order

to create a single platform that provides more control and flexibility as far as data

collection and analysis, and displaying the results in a highly interactive (i.e. visual)

environment are concerned. Figure 5.5 shows a sample snapshot of the LabVIEW

graphical environment. Each program written in LabVIEW is called a Virtual Instrument

(VI) which consists of a graphical user interface (i.e. Front Panel) and a graphical code

(i.e. Block Diagram). Each node in a Block Diagram performs a specific task and is

connected to other nodes via wires. More information about LabVIEW graphical

programming can be found in Appendix B.

41
Figure ‎5.5: A VI Consists of a Front Panel and a Block Diagram

42
In this research, a real time data acquisition VI was designed and implemented to

customize and append an instrument driver for the PNI TCM orientation tracker.. To

create an interface between the instrument driver and the data collection device, the NI

Virtual Instrumentation Software Architecture (VISA) API was used for serial

communication. VISA, basically provides users with the ability to open, configure (i.e.

setting baud rate, flow control, parity), write to and read from, and close any type of

interfaces such as GPIB, TCP/IP, Ethernet/LAN, IEEE 1394, USB, and serial, and handle

errors in a fast and easier way in comparison with the same functions developed in a text-

based programming environment (e.g. C++). Figure 5.6 shows a rudimentary structure of

the VISA implemented in the developed framework.

Figure ‎5.6: Developed VISA Interface Structure

As shown in this Figure, VISA Resource Name passes session information between

instrument driver and SubVIs and is a unique identifier reference to the data collection

device (e.g. COM1, COM2). VISA Open essentially opens a session to communicate with

43
the device specified by the VISA Resource Name and returns a session identifier that can

be used to invoke operations on that device. This is equivalent to the function

PRIME::Initialize() in the Figure 5.3. Serial Configuration, as stated before, sets the port

configuration parameters specific to the device such as baud rate which is a measurement

for communication speed equal to 38,400 HZ for the PNI TCM module. VISA Write has a

performance similar to PRIME::Control() in the Figure 5.3. It extracts the Payload and

requests needed angle measurements. Subsequently, VISA Read reads the requested data

based on what was defined in VISA Write. The CRC will be performed to detect

corrupted data packets by calculating the CRC-16 of the output string from the tracker

and comparing it to the checksum at the end of the output string. In essence, this

procedure is identical to what was described in Subection 5.2.1. Finally, similar to

function PRIME::Shutdown(), VISA Close shuts down the port and terminates the

software connection to the device.

The advantage of real time automated data collection is that it enables the simulation

model to update itself in response to changes in the project environment. This can be

achieved by continually collecting time-stamped data. However, before the raw data

stream enters the simulation model or is used as input for visualization, it should be

classified, analyzed, and converted to a format that defines the state and the context of the

entity for which the data is collected. As such anda s shown in Figure 5.1, the raw data

collected using either the .NET or LabVIEW operational environments is passed onto the

data classification and analysis module of the developed DDDAS framework. The

following Subsection provides more details about this module.

44
5.3 Automated Data Classification and Analysis

One of the major challenges in collecting a large volume of heterogeneous information is

that unnecessary data may also be inevitably collected. For example, in order to

mathematically describe the motion of a loader’s boom within the context of an

earthmoving operation, a 3D orientation tracker mounted on the boom would capture

three angular values namely yaw, pitch, and roll. However, given that the boom must be

raised or lowered to load or unload a truck, the main piece of information needed to

determine the start and end times of load or unload activities is the pitch angle. As such,

potential trembles resulting in small changes in the roll angle and also possible motions

such as sidewise movements and maneuvering of the loader leading to a change in the

yaw value are to the most extent, redundant as far as detection the beginning and end of

load and unload activities for the simulation model and having a smooth animation for

visualization are concerned. Therefore, collected data must be carefully classified so that

only relevant and useful information is passed onto the next steps.

Classified data also needs to be transformed into a proper format interpretable by the

simulation model. One such format is a numerical representation of activity durations

using probabilistic distributions. Since discrete events mark the beginning and end of

each activity, identifying the duration of individual activities can be achieved by

detecting time-stamped events corresponding to the beginning and end of that activity.

Therefore, activity durations can be derived from the pool of classified collected raw data

and suitable probability distributions will be then fit to the calculated duration values. In

45
the earthmoving example described above, the angle of the boom and the truck bed

relative to the horizontal line can be used to identify the start and end points of load and

unload activities and determine activity durations. For example, in the operation depicted

in Figure 5.7, activity durations can be calculated by comparing the time stamps

corresponding to when each event (i.e. raise boom, load truck , lower boom, haul, raise

bed to dump, lower bed, return) occurs based on the orientation data (i.e. angles) received

from sensors mounted on the equipment.

Figure ‎5.7: Simplified Layout of an Earthmoving Operation

An example of how a series of time-stamped data can be used to extract certain activities

and their durations is illustrated in Figure 5.8.

46
 Figure ‎5.8: Activity Durations Based on the Variation of Equipment Body Orientation
with Respect to Time
(RB = Raise Bucket, LT = Load Truck, LB = Lower Bucket, RTB = Raise Truck Bed, P = Put, LTB =
Lower Truck Bed)

In this Figure, the first diagram shows changes of angle α (loader boom angle relative to

the horizontal line) and the second diagram shows angle β (truck bed angle relative to the

horizontal line) over time. Considering angular variation histograms displayed in these

two diagrams, a timeline representing the duration of each activity can be generated. For

example, an increasing angle α and a constant angle β (close to zero) indicate that the

loader is raising its boom while the truck is waiting to be loaded (RB in Figure 5.8). A

near constantan angle α (close to its peak value) and a constant angle β (close to zero)

indicate that the loader is putting soil into the truck (LT in Figure 5.8). A decreasing

47
angle α and a constant angle β (close to zero) indicate that the loader is lowering its boom

while the truck is preparing to move (LB in Figure 5.8). An instance of “Load” activity is

completed when all three (RB, LT, and LB) processes are completed.

A similar analysis can be done to isolate instances of “Haul”, “Dump”, and “Return”

activities. For instance, given that angle α is constant (at a value close to zero), if angle β

is increasing from zero, the truck bed is being raised (RTB in Figure 5.8), if angle β is

almost constant (close to its peak value), soil is being dumped (P in Figure 5.8), and if

angle β is decreasing, the truck bed is being lowered (LTB in Figure 5.8). These three

processes, put together, will constitute an instance of “Dump” activity. Since histogram

data is time-stamped, duration values can be easily determined for all such instances.

Mathematical models will then be applied to a well-populated pool of these calculated

durations to determine a distribution function that best represents the duration of that

activity. This distribution function is then used to describe the duration of that activity in

the corresponding DES model [63].

It is worth mentioning that given the unavailability of GPS to obtain time-stamped

positional data in an indoor environment (e.g. laboratory setting where the components of

this framework was tested), a number of simplifying assumptions had to be made when

developing the methodology for extracting the duration of activities. For example, it was

assumed that the haul activity would not start until the loader lowers its boom and would

not finish until the truck raises its bed. Likewise, return activity starts when truck’s β

angle reaches zero and finishes at the beginning of the load activity, when the loader

48
starts raising its boom. It is clear that incorporating positional data into the proposed

algorithms for calculating activity durations not only does eliminate the need to consider

these and similar simplifying assumptions, but also enhances the accuracy of the

algorithm. As such, future work in this research will include activities specifically

targeting this need.

The classification and analysis module accepts input from data collection devices, outputs

the classified data for pre-processed animation, and also analyzes data to be passed onto

the simulation model. This guarantees that only relevant data is used and that the

simulation model is not only the receiving end of the process but also can steer the data

collection process by requesting additional field data to be collected, if necessary. The

classification and analysis process also includes statistical analysis algorithms to

categorize the activities based on the trend of the collected data and to remove the

outliers and eliminate the non-relevant data. The next Subsection describes the developed

system for implementing such algorithms.

5.3.1 VI Structure for Data Analysis Module

In order to create a standalone platform consisting of both data collection and data

analysis modules, all mathematical and logical functions for data classification, extraction

of activity durations, and statistical analysis were appended to the same VI. Figure 5.9

shows the VI structure.

49
 Figure ‎5.9: Single VI Containing Data Acquisition and Data Analysis Functions

A cluster of real time orientation data enters the VISA interface and undergoes steps

depicted both in Figure 5.6 and by a dashed outline in Figure 5.9. This is followed by raw

data being classified and time-stamped. The time-stamped data will then be used to

calculate activity durations using several mathematical and logical commands built into

the VI. Statistical analysis will be also performed on the well-populated pool of duration

to calculate mean, standard deviation and other parameters required to describe activities

in a data-driven simulation model.

5.4 Data-Driven Simulation

As far as the system architecture illustrated in Figure 5.1 is concerned, once the data is

available after the classification and analysis step, input parameters for simulation model

50
are determined. Construction operations can be broken down into and modeled as a

system of discrete activities which makes DES a viable method for simulating such

operations. One of the most commonly used DES systems is STROBOSCOPE [47].

STROBOSCOPE, initially designed for construction operations, is an open-design

programming language that enables users to make complex dynamic decisions and thus,

control the simulation at run-time. The advantage of STROBOSCOPE over many other

existing DES modeling platforms is that it considers the diversity of resources and their

characterizations. In addition, it has been built upon the concept of traditional activity

cycle diagram (ACD) which makes it suitable for modeling a large group of construction

operations that are cyclic in nature. STROBOSCOPE models are based on a graphical

network of interconnected modeling elements. A DES model of a sample earthmoving

operation is illustrated in Figure 5.10.

Figure ‎5.10: DES Model of a Typical Earthmoving Operation

In this Figure, SoilInPlace, LoadersWait, TrucksWait, and MovedSoil are queues where

resources wait before being drawn to activities (if needed). Also, Load is called a combi

51
activity since it immediately follows a queue, and Haul, Unload, and Return are normal

activities. In order for a STROBOSCOPE model to describe a real system, attributes such

as activity durations, number of entities, and resource capacities must be known. In the

absence of collected field data, assumptions and personal judgment is normally used to

quantify such parameters. As previously stated, one of the main motivations behind this

research was to investigate if further improvements can be made to the existing approach

of assigning values to simulation parameters by designing a methodology that

incorporates field data to obtain more realistic simulation parameters.

STROBOSCOPE models consist of a series of programming statements written in a

script input file. All parameters pertinent to the characteristics of each model element

should be defined through those programming statements. Therefore, once the

appropriate simulation parameters are determined from the collected data, the simulation

script is opened and updated based on the calculated parameters.

5.5 Pre- and post-processed Animations

Chapter 4 illustrated a detailed description about OSG, the visualization toolkit used in

this study. Following the data flow illustrated in Figure 5.1, as soon as appropriate field

data is collected and classified, a concurrent 3D dynamic animation of ongoing activities

is created. This pre-processed (i.e. generated before data is fed into the simulation model)

animation can assist in detecting potential conflicts and enhancing safety and monitoring

of the project. The other benefit of this animation is that unlike many existing site

monitoring systems which mainly rely on video streaming, finding the best spots to

52
install cameras such that every action can be monitored with a free line-of-sight is no

longer an issue. This is due to the fact that once the animation is rendered on the screen,

the user has complete control over the viewpoints and can change their locations and

directions of look, if necessary. For example, the user can zoom in an out or navigate

around the animated scene to gain a better visual perspective of certain parts of the

operation since as stated in Chapter 4, OSG provides the opportunity to change the

viewpoint to observe the scene from any desired angle.

In addition to the pre-processed animation, the results of the DES model can be used to

create a post-processed animation. As demonstrated in literature, providing visualized

output of a simulation model is preferred by many construction planners and analysts

since very often, making decisions solely based on the textual output of conventional

simulation systems is time consuming and prone to unwanted biases and mistakes [54,

72]. However, in addition to the benefits that general visualization of simulation models

has, providing decision-makers with two identical animations, one based on the exact real

movements occurring on the jobsite and the other based on the output of the updated

simulation model with the latest data obtained from the field provides an extremely

convenient way to evaluate and compare different scenarios with the concurrent filed

configuration and make more realistic decisions. For example, since each construction

project is unique in terms of requirements and usage of its working space, having a real

time data from the project and evaluating different scenarios based on the transformation,

requirements, and limitations of the working space (e.g. maneuverability issues for

loading and dumping activates in earthmoving operations, visibility problems for the

53
crane operator in steel girder erection, safety problems and detecting potential collision,

overcrowding in particular work zones), prevent decision-makers from making general

assumptions based on historical data or their expert eye on the work. Hence, displaying

the pre-processed animation side-by-side the simulation-based post-processed animation

enables decision-makers to see first-hand how current trends on the jobsite (reflected in

the pre-processed animation) and the expected performance of resources (as displayed in

the post-processed animation) are related, and hence effectively serves this purpose.

Finally, another major advantage of having pre- and post-processed animations is that

comparing the two animations greatly facilitates the validation and verification of the

simulation model. In this case, the modeler can intuitively make sure whether the model

contains any modeling flaws or whether it performs as intended (i.e. verification). Also, it

can be visually determined by people who are not construction experts whether the

simulation model accurately represents the real word.

Creating realistic post-processed 3D animations of a simulated construction processes is a

complicated task that has been previously studied by a number of researchers. A recent

example of a post-processed 3D visualization platform is VITASCOPE. VITASCOPE is

a general purpose 3D animation system for visualizing simulated processes modeled in

simulation tools such as STROBOSCOPE, capable of writing formatted output during a

simulation run. Based on the logged simulation model runtime data, VITASCOPE

graphically illustrates modeled operations by processing sequential, time-ordered

animation commands in an ASCII text file [10]. While VITASCOPE is a great tool for

54
creating post-processed (simulation-based) animations of construction activities, the

existing visualization capabilities of the framework developed in this research enable the

generation of a pre-processed animation using the same OSG-based environment and 3D

CAD models of construction equipment. As described earlier, these two animation

streams, when simultaneously displayed, can facilitate the process of validation and

verification of the simulation model while providing a means to intuitively compare

different scenarios tried in the simulation model.

It is worth mentioning that since the designed framework is intended to function in an

automated manner, and since the OSG visualization toolkit is written and extended using

the C++ programming environment, a middleware for linking LabVIEW (i.e. containing

data collection and analysis functionalities) to Microsoft Visual C++ (i.e. containing OSG

visualization platform) had to be designed. To this end, ActiveX Automation technology

was used in this research. ActiveX has an interface that allows individual programs to be

linked together to suit for specific computing needs [73].

5.6 What-If Analysis

Another building block of the framework as shown in Figure 5.1 is the “What-If

Analysis” module. In order for a construction engineer to make necessary decisions

regarding the complex processes, different scenarios need to be assessed and the cost and

time associated with each scenario must be determined. For example, a decision

regarding equipment fleet to be used in an earthmoving operation could be the one

associated with the minimized expected cost [47]. Considering all possible configurations

55
in terms of crew sizes, number of equipment and their arrangements, operations logic,

and construction methods, a decision-maker may end up having to choose from several

combinations to perform a certain task. Using simple methods such as subjective

mathematical comparisons or more complex optimization models, the engineer can then

determine the best configuration that satisfies the predefined criteria (e.g. objective

function, time, cost).

5.7 Decision-Making and Dynamic Feedback

The last component of the system architecture presented in Figure 5.1 is Decision-

Making and Dynamic Feedback. The developed algorithms for data collection,

classification and analysis, simulation and visualization, will be best used in the presence

of a human decision-maker or a team of decision-makers who will be ultimately

responsible for making the required modifications to the target construction process. As

stated earlier, presented data to the user contains two juxtaposed animations; one

identical to the actual process taking place in the jobsite (pre-processed), and the other,

resulted from simulating alternative scenarios (post-processed). Also, the user is provided

with the results of the simulation model and the output of the what-if analysis in order to

decide which alternative solution is the most appropriate. Therefore, not only by

intuitively watching side-by-side animations, but also through intelligently interpreting

performance attributes (e.g. productivity rates) from the simulation output and various

alternative scenarios, the decision-maker(s) will have the ability to further adjust future

processes. Ultimately, and due to the dynamic nature of construction projects, the cycle

56
presented in Figure 5.1 will repeat to reflect any further changes occurring in the process.

In other words, the next phase of data collection starts after expert modifications are

applied to the construction resources, and activities and a new set of data will be

classified, analyzed, simulated, and visualized. This guarantees that through continues

data collection from the equipment involved in a construction process, at any given time,

the system will be functioning at its best performance level.

57
 CHAPTER 6: LABORATORY SCALE EXPERIMENTS AND
RESULTS

Chapter 5 outlined the individual components of the developed framework and their

relationships in the context of the overall system. In this Chapter, results of preliminary

proof-of-concepts experiments conducted in the Decision Support, Information

Management, and Automation Laboratory (DESIMAL) at the University of Central

Florida are provided to demonstrate the validity and applicability of the developed

methodology and algorithms for data collection, data analysis, visualization, and data-

driven simulation. In particular, the validation phase included a number of laboratory-

scale equipment operations scenarios designed and implemented to test certain aspects of

the developed framework. In addition, a comprehensive experiment was carried out in

which the robustness, applicability, and overall functionality of the framework in terms of

data collection and classification capabilities, ability to generate realistic pre-processed

animations, and effectiveness to create data-driven simulation was validated. The

following Subsections provide more insight about the details and outcomes of each of the

validation experiments.

6.1 Preliminary Results

6.1.1 Experiment Tools and Peripheral Devices

Preliminary experiments were performed on a laboratory-scale Construction Equipment

Automation Platform (CEAP) using remotely-controlled model construction equipment.

58
A NetCam XL IP-addressable camera and a DellTM Precision T1500 desktop system were

also deployed. The camera was used to demonstrate the correctness and precision of the

pre-processed visualization and the desktop system was the main computing platform.

Figure 6.1 illustrates the overall arrangement of the tools and peripheral devices namely

the CEAP, model construction equipment, IP-addressable camera, and the computer

system.

Figure ‎6.1: Overall Arrangement of Experiment Tools and Devices

In order to collect equipment motion data several PNI TCM 3D orientation trackers were

used. These modules were mounted on model construction equipment to capture and

59
transmit three angular values namely yaw (heading), pitch (tilt), and roll. Figure 6.2

shows a PNI TCM 3D orientation tracker mounted on a model excavator with definitions

of yaw, pitch, and roll angles. Also manufacturer’s specifications of this orientation

tracker are listed in Table 6.1.

Figure ‎6.2: A Prime 3D Orientation Tracker Mounted on a Model Excavator with

Definitions of Yaw, Pitch and Roll Angles

Table 6.1: Manufacturer’s Specifications of Prime 3D Orientation Tracker

Angle Parameter Value

Range 360°
Heading Accuracy (tilt ≤ 45°) 1° rms
Resolution 0.1°
Pitch ±90°
Range
Roll ±180°
Tilt (Pitch/Roll)
Accuracy 1° rms
Resolution 0.1°

60
6.1.2 Single Object Data Collection and Visualization

Initially, a series of small-scale validation tests were conducted using data collected from

only one model construction equipment. Later, data collection, data analysis, and

visualization algorithms were modified to enable data capturing and processing from

multiple objects.

The first in a series of these experiments was conducted using an orientation tracker

mounted on a model loader [74]. Figure 6.3 shows the loader on the CEAP and the

orientation tracker attached to the boom of the loader.

Figure ‎6.3: Orientation Tracker Mounted on a Loader's Boom

The first step in conducting each experiment was to collect and classify equipment

motion data to provide necessary input for visualization and data-driven simulation. As

stated in Chapter 5, data collection and analysis is performed using LabVIEW. Figure 6.4

shows the Front Panel (i.e. user interface) of the data collection system for the validation

experiment using a single model loader. As shown in this Figure, the interface of the

developed VI enables a user to specify a communication port to receive data from the

61
orientation tracker, start and stop the data collection task, and view the numerical values

of the collected orientation angles.

Figure ‎6.4: Front Panel of Data Collection VI for a Single Loader

Once the VI is launched the data collection task begins. This is followed by a continuous

stream of real time classified angular data displayed in the three indicators designed to

show yaw, pitch, and roll values. These indicators are marked as YawLoader,

PitchLoader, and RollLoader in Figure 6.4, respectively. As soon as the user switches off

the data collection using the “Stop Data Collection” button the data stream stops. By

clicking on “Start Animation”, an animation of the exact same movements of the loader’s

boom will appear on the screen. Figure 6.5 shows snapshots from this test. In this Figure,

several frames of the live video streams of the real system captured using the IP-

62
addressable camera are displayed next to the corresponding 3D animation frames

generated by the developed framework in real time.

Figure ‎6.5: Real Time Display of Loader's Boom Movements and Corresponding 3D
Animation Generated in Real Time

63
6.1.3 Double Object Data Collection and Visualization

In order to validate the generalizability of the overall framework and to demonstrate that

the developed methods will properly function in situations where operational data from

more than one piece of equipment has to be collected, the data collection algorithms in

LabVIEW as well as the visualization methods (created in .NET using OpenSceneGraph)

were slightly modified. In doing so, the major issue that was successfully addressed was

to update the processes inside the data analysis module to be able to identify individual

activities from a large pool of raw motion data collected from several pieces of

equipment using multiple data collection devices (i.e. 3D orientation trackers), determine

the logical relationships and interactions between a group of equipment, and

consequently extract activity durations [63]. To validate the newly developed methods, a

laboratory-scale experiment was set up where operational data was collected from two

models, and the collected data was processed to generate a live 3D animation as well as

the calculate the main input parameters needed by the data-driven simulation module to

describe equipment activities in a meaningful format. In this experiment, two orientation

trackers were mounted on a model loader and a model truck. Figure 6.6 shows these

equipment placed on the CEAP while the orientation sensors are mounted on the loader’s

boom and the truck’s bed.

64
 Figure ‎6.6: Orientation Trackers Mounted on a Loader's Boom and a Truck’s Bed

Similar to the validation experiment using only one object, a VI was created and

implemented for data collection. However, this time not only the three angular data are

shown, but also two diagrams containing a series of time-stamped data to extract activity

durations are illustrated in the Front Panel. Each of these histograms shows the how

incoming data collected from the orientation tracker changes over time. For example, the

trend of data corresponding to the loader’s boom indicates that the boom is first lowered

from its initial state, raised and remained in a steady state for some time, lowered again

and remained in a steady state for a while, and finally raised. By observing this data

trend, one can conclude that the loader was involved in a cycle of digging soil (boom

down position) followed by loading a truck (boom up position). Based on the collected

data and using the developed algorithms for detecting individual activities from a series

of angular data, mean and standard deviation of durations were calculated and displayed.

Figure 6.7 shows the Front Panel used in this experiment.

65
 Figure ‎6.7: Front Panel of Data Collection VI for a Double Object Experiment

Similar to the scenario in which only one object was used, a real time stream of motion

data is captured and displayed in the specific indicators on the Front Panel.

Simultaneously, each activity is detected by the VI based on the existing data trends and

activity duration is calculated using mathematical algorithms inside the corresponding

Block Diagram. Figure 6.8 illustrates a portion of the extensive Block Diagram

developed for data collection and analysis purposes.

66
 Figure ‎6.8: Partial View of the Extensive Block Diagram Developed in this Research

Individual activity durations are calculated as long as equipment motion data is

streaming. The calculated values are used to populate numerical arrays. The content of

each numerical array corresponding to a certain activity (e.g. load, dump) is evaluated in

real time using statistical methods to determine the mean and standard deviation of the

probabilistic normal duration that best fits all values.

Once the data collection is stopped by the user, and the start animation command is

triggered, a 3D animation showing the exact same equipment movements appears on the

screen. Figure 6.9 shows snapshots from the live video stream of the real system as well

as the corresponding 3D animation created in real time.

67
Figure ‎6.9: Real Time Display of Loader's Boom and Truck’s Bed Movements and
Corresponding Animations

68
6.2 Comprehensive Example: Data-Driven Simulation

In order to demonstrate the ability of the developed framework in supporting the prospect

of data-driven simulation by collecting, processing, and integrating real time operational

data with simulation modeling, a simplified yet comprehensive operational scenario was

designed and carried out. In this experiment, the goal was to move 200 pieces of model

rocks from a loading area (i.e. Area #1) to a dumping site (i.e. Area 2) for a dam

construction project. A model loader was used to load a model truck. The truck would

haul the rocks from the loading area to the dumping site. It was assumed that pieces of

rock are so big and heavy that each truck can carry only one rock in each hauling cycle.

In order to collect field data, two orientation trackers were mounted on the model

equipment; one on the loader’s boom, and the other on the truck’s bed. Figure 6.10 shows

the layout of the experiment conducted on the CEAP.

Figure ‎6.10: Experiment Layout of a Model Dam Construction Scenario

69
Figure 6.11 shows the DES network of this operation. In this Figure, RocksToMove,

LoadersWait, TrucksWait, and MovedRocks are queues, Load is a combi activity (i.e. it

immediately follows a queue), and Haul, Dump, and Return are normal activities. Also,

all network elements (i.e. activities and queues) are connected by links. Each link has a

specific name and can carry a certain type of resource (i.e. Rock, Loader, Truck) from

one element to the other. For example, RK2 is defined as a link connecting Load and

Haul activities which carries the Rock resource.

Figure ‎6.11: DES Model of Rock Hauling Activity

As stated before, during the planning stages of a project, simulation modelers generally

rely on expert judgments or field reports from similar past projects to determine model

parameters such as activity durations. Following the same logic and as shown in Figure

6.12, a DES script was initially created in STROBOSCOPE for the dam construction

70
scenario where activity durations were approximated based on the overall arrangement of

resources and considering the motion speed of model equipment.

Figure ‎6.12: STROBOSCOPE Simulation Input File

71
In this Figure, statements used to describe activity durations inside the simulation script

are highlighted. In addition, necessary statements were added to assess and report the

total completion time of the project. The output of this simulation model is shown in

Figure 6.13. In this Figure, average waiting time of resources inside their corresponding

queues is highlighted. Since the simulation parameter (i.e. activity durations) were

approximated in the first place, the resulting waiting times may or may not represent the

actual idle time of resources during the course of the real world project.

Figure ‎6.13: STROBOSCOPE Simulation Output File

72
Hence, it was decided to incorporate real time operational data collected from the model

equipment into the DES model to create a more accurate and realistic output that better

serve the decision-making process. To do so, data was collected from several complete

operational cycles including Load, Haul, Dump, and Return activities. The collected data

was further processed to determine and display the statistical mean and the standard

deviation of each activity as identified in the corresponding VI by establishing rules

relating equipment motions to the beginning and end events of individual activities.

Figure 6.14 shows the VI and the results obtained for activity durations.

These statistical parameters were used to replace the approximate duration values by

assigning more realistic Normal distributions to individual activity durations and update

the DES model. The revised STROBOSCOPE simulation script is shown in Figure 6.15

where newly calculated activity durations are highlighted. The updated simulation model

was then run and results were collected as illustrated in Figure 6.16.

73
Figure ‎6.14: Developed VI for Data Collection and Analysis for Rock Hauling Example

74
Figure ‎6.15: STROBOSCOPE Simulation Input File Containing Updated Activity
Durations

75
Figure ‎6.16: STROBOSCOPE Simulation Output File Based on the Updated Durations

Comparing the output of the revised simulation model (Figure 6.16) with that of the

original model (Figure 6.13), it is clearly seen that incorporating field data into the

simulation modelling has resulted the average waiting time of the loader to significantly

decrease from 72.21 seconds to 44.77 seconds. Also, the overall project completion time

is noticeably improved as a result of using real equipment data to update activity

durations. These improvements can potentially affect the outcome of the planning of

projects tasks scheduled for the immediate future tasks as far as resource arrangements

76
and combinations are concerned. Table 6.2 summarizes the results of this comparative

validation example.

Table ‎6.2:
Comparison between Estimated Durations and Actual Durations Based on Real Time
data

Approximated Duration Data-Driven Duration

Simulation Element
(sec.) (sec.)
Load N[40,5] N[55,3.63]
Haul N[35,5] N[25.57,1.96]
Dump N[3,0.5] N[8.6,1.15]
Return N[35,5] N[22.52, 2.18]
Loader’s Avg. Idle 72.21 44.77

77
 CHAPTER 7: CONCLUSIONS AND FUTURE WORK

7.1 Conclusions

Operations level planning and control is one of the most critical components of managing

ongoing activities in a construction site. Proper resource planning and control can

guarantee that the best possible arrangement of resources are deployed which will in turn,

result in substantial savings in project completion time and cost. To this end, simulation

modeling as a powerful tool for analyzing complex construction operations has gained

significant credibility during the past several years. Commonly, many simulation

paradigms use static or historical data to create computer interpretable representations of

real engineering systems. The suitability of this approach for modeling construction

operations, however, has always been a challenge since most construction projects are

unique in nature, and every project is different in design, specifications, methods, and

standards. Therefore, there is a significant need for a methodology that not only does

enable the modeling of main entities and logical relationships in a real system, but also

allows that real time changes be incorporated into the simulation model.

The major requirement of a modeling platform capable of precisely representing the real

world construction system is a data collection scheme capable of providing the simulation

model with the latest information about the status of underlying processes and project

entities. Given the dynamic nature and complexity of many construction processes,

manually gathering the information necessary to create the corresponding simulation

78
model is a tedious if not an impossible task and thus, it is necessary to employ an

automated system for collecting required data and convert them to a format

understandable by and useful for the simulation model.

This Thesis document reported on a study conducted to investigate the requirements and

applicability of a data-driven decision support system based on the relatively new

simulation paradigm of dynamic data-driven application system (DDDAS). This

paradigm was integrated with the traditional discrete-event simulation (DES) modeling to

create a single decision-making framework for short-term scheduling and system control.

The framework is capable of automatically collecting real time operational data from

construction equipment and subsequently sorting, analyzing, and using them to create

real time 3D animations of the concurrent construction processes, and also updating the

simulation model describing the real operations based on the latest trends in the data

stream collected from the construction jobsite.

The developed methodology was validated inside a .NET object-oriented environment

along with a graphical programming and data collection platform, LabVIEW. To validate

the functionality and robustness of the developed algorithms, 3D orientation trackers

were used to collect motion data from moving parts of model construction equipment,

and the collected data was analyzed and transformed into a format meaningful for the

decision-making process. All preliminary experiments were performed in an indoor

laboratory setting at the University of Central Florida.

79
The following summarizes the main milestones of this research that have been

successfully achieved:

 A data collection platform was developed in LabVIEW for collecting angular data

from 3D orientation trackers that transmitted data over an RS-232 serial port

interface.

 Data classification and analysis algorithms were developed in LabVIEW for real

time analysis of raw data and to convert them to proper format for use as input by

the simulation model and visualization system.

 Dynamic concurrent animations of ongoing activities in real system were created

inside the .NET environment using the OpenSceneGraph visualization toolkit.

 Necessary communication interfaces were created to facilitate data

interoperability between the data collection and analysis module, the object-

oriented programming platform used for visualization, and the discrete event

simulation used to model ongoing construction activities.

 Laboratory-scale validation experiments were successfully conducted and results

were documented to demonstrate the applicability and reliability of the developed

data-driven decision support framework.

80
7.2 Future Work

The presented research is part of a much larger ongoing project which aims to facilitate

the integration of real time operational data into the construction decision-making

process. The next step in developing the current system will contain communication

methods to capture Real Time Kinematics (RTK) GPS data for location tracking of

construction equipment and also the deployment of more efficient orientation trackers

that can adequately handle specific conditions of the jobsite in terms of communication

range, accuracy, and ambient noise. In addition to spatio-temporal data (i.e. position,

orientation), payload information is another potential source of data that can be collected

and used to determine the state of equipment involved in operations such as earthmoving

or steel erection where material is transported from one location to another. There are

also other types of data that are not necessarily related to construction resources but can

potentially affect the progress of field activities. Examples include weather-related (e.g.

temperature, humidity) data and soil and topography data. To this end, future work in this

research will include the design and implementation of robust algorithms to collect,

process, and fuse such multiple-source heterogeneous data [75, 76].

Also, there is a need to examine the developed pre-processed visualization module in the

presence of a post-processed (simulation-based) 3D visualization platform to highlight

the advantages and identify potential shortcomings of the current framework. In addition,

work needs to be done to improve the mathematical efficiency and statistical accuracy of

the framework in order to more effectively handle, fuse, and process large volumes of

81
raw incoming data especially when multiple heterogeneous data collection devices are

used.

Automating and optimizing equipment operations is another potential area for future

work in this research. To achieve this, machine learning methods will be investigated to

develop a self-learning system capable of observing activities that involve resource (i.e.

equipment, material, personnel) interactions, extracting information by identifying data

trends and cyclic motions, and subsequently generating knowledge-based action plans to

streamline process flows on the jobsite.

82
APPENDIX A: C++ ALGOTIRHMS FLOWCHARTS

83
As stated in Chapter 5, the computing platform developed in this research takes

advantage of a .NET object-oriented design as well as a graphical user interface (GUI)

developed in LabVIEW. Platform interoperability features that facilitate the

communication of information between the .NET environment and the LabVIEW

interface are provided using the ActiveX automation interface.

In this Appendix, a detailed description of the .NET functionalities is presented by using

flowcharts that describe how different programming modules communicate and what

type of data is transferred between these modules The illustrated flowcharts are only

intended to supplement the discussion of the topics introduced in previous Chapters and

to help interested readers gain a better understanding of the data flow in the developed

platform.

There are four major C++ functions used inside the .NET environment. These functions

include,

CreateAnimationPath()

This function plays the most critical role since it facilitates communication between C++

and LabVIEW to capture and store angular data as vectors and create and return an

animation path.

CreateMovingModel()

This function imports 3D CAD files of articulated parts of model construction equipment

and defines the parent-child hierarchical relationships between different nodes

84
constructing each object in the scene. This function creates and returns an intermediate or

group node including all objects called model.

CreateModel()

This function defines the origin of the coordination system used to create the

visualization scene. It also attaches the group node model to the root node and returns

root as the highest point of the hierarchy.

Main()

This function initializes a LabVIEW interface, tilts the scene to arrive at the desired

viewpoint, set the scene to render, and finally runs the animation.

Figures a.1 through A.4 illustrates detailed flowcharts of the above function.

85
Figure A. 1: CreateAnimationPath() Function Flowchart

86
Figure A. 2: CreateMovingModel() Function Flowchart

87
Figure A. 3: CreateModel() Function Flowchart

88
Figure A. 4: Main() Function Flowchart

89
APPENDIX B: LabVIEW GRAPHICAL PROGRAMMING AND
ALGORITHMS

90
LabVIEW1 (i.e. Laboratory Virtual Instrument Engineering Workbench) is a product of

National Instrument (NI) and is a platform for designing engineering and scientific

measurement and control systems. LabVIEW uses graphical programming (G) as a data

flow language in which nodes, as operations or functions, operate on data received

through “wires”. This approach provides an efficient way of handling and processing data

especially when compared to most text-based programming languages which operate

based on a sequential line by line manner. LabVIEW has built-in tools designed

specifically for data collection, analysis, and presentation.

LabVIEW programs are called virtual instruments (VIs). Each VI has two windows: the

user interface which is called the Front Panel, and the graphical code called the Block

Diagram. The Front Panel provides users with interactive controls such as buttons, gages,

graphs, and tables as well as tools to save data files or automatically generating reports.

The Block Diagram, on the other hand, consists of icons and nodes that are connected

together via wires. Figure B.1 shows a customized Front Panel and the corresponding

Block Diagram.

1
LabVIEW is a registered trademark of National Instruments (NI).

91
Figure B. 1: A Customized VI - The Upper Window is the Front Panel and the Bottom
Window is the Block Diagram

92
Algorithms for data collection and data analysis components of the presented framework

were developed using LabVIEW. To this end, a Plug and Play (P&P) instrument driver

initially developed by the NI was modified, customized and appended to meet the

required functionalities needed in this research. An instrument driver is a library of VIs

that controls a programmable instrument. NI instrument drivers are provided as open-

source well-documented libraries and can be customized by the end user to perform

specific tasks. In this research, an instrument driver was used for communication with

orientation trackers employed for data collection via RS-232 protocol for serial

communication.

As stated in Chapter 5, Virtual Instrumentation Software Architecture (VISA), a standard

I/O language and an application programming interface (API) for sensor programming

was used in this research. VISA basically facilitates port communication by providing

needed operations such as opening, writing to, reading from, and closing a port. Figures

B.2 through B.5 show special nodes in LabVIEW for each of the indicated tasks.

Figure B. 2: VISA Open Opens the Specified Port by the VISA Resource Name

93
 Figure B. 3: VISA Write Writes Data to the Specified Port by the VISA Resource Name

Figure B. 4: VISA Read Reads Data from the Specified Port by the VISA Resource Name

Figure B. 5: VISA Close Closes the Specified Port by the VISA Resource Name

Also, Figure B.6 illustrates a simplified layout of how these VISA functions are

connected to each other via wires in the developed data collection system.

Figure B. 6: A Series of VISA Functions and Their Connections as Used in this Research

94
Data classification and analysis algorithms were also designed in the same VI. Figures

B.7 through B.12 show built-in functions that were used to configure a relatively

sophisticated graphical code capable of real time extraction of activity durations from

angular raw data. Necessary information has been provided in each Figure caption.

Figure B. 7: Requested Data Classified from Cluster of Real Time Orientation Data

Figure B. 8: Unbundled By Name Function that Returns Cluster Elements Whose Names
Have Been Specified

95
Figure B. 9: Greater? Function Returns True If x Is Greater than Y - This Function Was
Used to Detect Data Exceeding a Specified Threshold

Figure B. 10: Tick Count Function That Returns the Value of a Timer – This Function
Was Used to Measure the Duration of Each Activity

Figure B. 11: Build Array Function to Store Activity Durations in a Numerical Array

96
 Figure B. 12: Statistics Tool Returns the Specified Statistical Characteristics of Input
Arrays

In addition to the VI elements described above, there are a number of other functions,

tools, and controls that were also developed and used for constructing the VI. For

example Case Structures, While Loops, or PointByPoint Analysis functions that each one

of which perform specific tasks under different conditions. The descriptions and technical

details of these elements are, however, beyond the scope of this document. Interested

readers are encouraged to contact the author or the DESIMAL research group at the

University of Central Florida for more information.

97
 REFERENCES

[1] D.W. Halpin, L.S. Riggs, Planning and analysis of construction operations, Wiley,
New York, 1992.

[2] R.F. Cox, R.R.A. Issa, D. Aherns, Management's perception of key performance
indicators for construction, J. Constr. Eng. Manage. 129 (2) (2003) 142-151.

[3] W. Ibbs, Quantitative impacts of project change: size issues, J. Constr. Eng. Manage.
123 (3) (1997) 08-11.

[4] NRC, National Research Council (NRC), Federal Facilities Council Technical Report
No. 149 Reducing Construction Costs: Uses of Best Dispute Resolution Practices by
Project Owners, N.A. Press, Washington, DC, 2007.

[5] D.W. Halpin, CYCLONE - A Method for Modeling Job Site Processes, ASCE
Journal of Construction Division 103 (3) (1997) 489-499.

[6] B.C. Paulson, W.T. Chan, C.C. Koo, Construction operation simulation by
microcomputer, J. Constr. Eng. Manage. 113 (302) (1987) 302-314.

[7] Y.L. Liu, P.G. Ioannou, Graphical Object-Oriented Discrete-Event Simulation

System, in: Proceeding of the 1992 Winter Simulation Conference (WSC), Association
for Computing Machinery (ACM), Arlington, VA, 1992, pp. 1285-1291.

[8] J. Martinez, P.G. Ioannou, General purpose simulation with stroboscope, in:
Proceeding of the 1994 Winter Simulation Conference (WSC), Association for
Computing Machinery (ACM), Lake Buena Vista, FL, 1994, pp. 1159-1166.

[9] J. Shi, Activity-Based construction ABC modeling and simulation method, J. Constr.
Eng. Manage. 12 (354) (1999) 354-360.

[10] V.R. Kamat, J.C. Martinez, General-purpose 3D animation with VITASCOPE, in:
Proceeding of the 2004 Winter Simulation Conference, Washington, DC, 2004.

98
[11] A.H. Behzadan, V.R. Kamat, Automated generation of operations level construction
animations in outdoor augmented reality, J. Comput. Civil Eng. 23 (6) (2009) 405-417.

[12] L. Song, A framework for real-time simulation of heavy construction operations, in:
Proceeding of 2008 Winter Simulation Conference (WSC), Miami, FL, 2008, pp. 2387-
2395.

[13] G. Adkins, U.W. Pooch, Computer Simulation: a Tutorial, Computer 10 (4) (1977)
12-17.

[14] J. Banks, Handbook of simulation : principles, methodology, advances, applications,

and practice, Wiley, New York, 1998.

[15] W.S. Jang, M.J. Skibniewski, Wireless sensor technologies for automated tracking
and monitoring of construction materials utilizing Zigbee networks, in: Proceeding of
ASCE Construction Research Congress, Grand Bahamas Island, 2007.

[16] A.H. Behzadan, Z. Aziz, C.J. Anumba, V.R. Kamat, Ubiquitous location tracking for
context-specific information delivery on construction sites, Autom. Constr. 17 (6) (2008)
737-748.

[17] D. Grau, C. Caldas, P. Goodrum, C. Haas, Assessing the impact of materials

tracking technologies on construction craft productivity, Autom. Constr. 18 (7) (2009)
903-911.

[18] J. Messner, M. Horman, D. Riley, An Interactive Visualization Environment for

Construction Engineering Education, in: Proceeding of Construction Research Congress
2005: Broadening Perspectives San Diego, CA, 2005, pp. 1279-1288.

[19] S. Han, F. Peña-Mora, M. Golparvar-Fard, S.Roh, Application of a visualization

technique for safety management, in: Proceeding of the 2009 ASCE International
Workshop on Computing in Civil Engineering, Austin, Texas, 2009, pp. 543-551.

[20] A.H. Behzadan, V.R. Kamat, Interactive augmented reality visualization for
improved damage prevention and maintenance of underground infrastructure, in:

99
Proceeding of the 2009 Construction Research Congress, Seattle, WA, 2009, pp. 1214-
1222.

[21] R. Akhavian, A.H. Behzadan, Dynamic Data Driven Simulation of Ongoing

Construction Processes, in: Proceeding of 3rd International and 9th Construction
Specialty Conference, Ottawa, Canada, 2011, pp. CN-154 (151-110).

[22] I. Davidson, M. Skibniewski, Simulation of automated data collection in buildings,

J. Comput. Civil Eng. 9 (1) (1995) 9-20.

[23] B. Mccullouch, Automating field data collection in construction organizations, in:

Proceeding of the 1997 ASCE Construction Congress Mimleapolis, Mimlesota, 1997, pp.
957-963.

[24] G.S. Cheok, R.R. Lipman, C. Witzgall, J. Bernal, W.C. Stone, NIST Construction
Automation Program Rep. No: 4, Non-Intrusive Scanning Technology for Construction
Status Determination, Gaithersburg, Md, 2000.

[25] R. Navon, Automated project performance control of construction projects, Autom.

Constr. 14 (4) (2005) 467-476.

[26] S. Taneja, A. Akcamete, B. Akinci, J.H. Garrett, E.W. East, L. Soibelman, Analysis
of Three Indoor Localization Technologies to Support Facility Management Field
Activities, in: Proceeding of International Conference on Computing in Civil and
Building Engineering, Nottingham, UK, 2010.

[27] K.S. Saidi, A.M. Lytle, W.C. Stone, Report of the NIST Workshop on Data
Exchange Standards at the Construction Job Site, in: Proceeding of 20th International
Symposium on Automation and Robotics in Construction (ISARC), Eindhoven, The
Netherlands, 2003, pp. 617-622.

[28] J. Song, C.T. Haas, C. Caldas, E. Ergen, B. Akinci, Automating the task of tracking
the delivery and receipt of fabricated pipe spools in industrial projects, Autom. Constr. 15
(2) (2006) 166-177.

[29] J. Song, C.T. Haas, C. Caldas, Tracking the location of materials on construction job
sites, J. Constr. Eng. Manage. 132 (9) (2006) 911-918.

100
[30] J. Teizer, M. Venugopal, A. Walia, Ultrawideband for Automated Real-Time Three-
Dimensional Location Sensing for Workforce, Equipment, and Material Positioning and
Tracking, Journal of Transportation Research Record 2081 (6) (2008) 56-64.

[31] D.T. Grau, C.H. Caldas, Methodology for automating the identification and
localization of construction components on industrial projects, J. Comput. Civil Eng. 23
(1) (2009) 3-13.

[32] H.M. Khoury, V.R. Kamat, Indoor User Localization for Context-Aware
Information Retrieval in Construction Projects, Autom. Constr. 18 (4) (2009) 444-457.

[33] E.J. Jaselskis, M.R. Anderson, C.T. Jahren, Y. Rodrigues, S. Njos, Radio Frequency
Identification Applications in Construction Industry, J. Constr. Eng. Manage. 121 (2)
(1995) 189-196.

[34] G.M. Jog, I.K. Brilakis, D.C. Angelides, Testing in harsh conditions: Tracking
resources on construction sites with machine vision, Autom. Constr. 20 (4) (2011) 328-
337.

[35] A.A. Oloufa, M. Ikeda, H. Oda, Situational awareness of construction equipment

using GPS, wireless and web technologies, Autom. Constr. 12 (6) (2003) 737-748.

[36] C.H. Caldas, D.T. Grau, C.T. Haas, Using global positioning system to improve
materials-locating processes on industrial projects, J. Constr. Eng. Manage. 132 (7)
(2006) 741-749.

[37] U.S. Coast Guard Navigation Center,

http://www.navcen.uscg.gov/?pageName=GPS, (accessed Sep. 27, 2011).

[38] E. Ergen, B. Akinci, B. East, J. Kirby, Tracking Components and Maintenance

History within a Facility Utilizing Radio Frequency Identification Technology, J.
Comput. Civil Eng. 21 (1) (2007) 11–20.

[39] J. Teizer, D. Lao, M. Sofer, Rapid Automated Monitoring of Construction Site

Activities using Ultra-Wideband, in: Proceeding of 24th International Symposium on
Automation and Robotics in Construction, Cochin, Kerala, India, 2007, pp. 23-28.

101
[40] S. Razavi, O. Moselhi, Indoor Construction Location Sensing using Low Cost
Passive RFID Tags, in: Proceeding of 3rd International and 9th Construction Specialty
Conference, Ottawa, Canada, 2011, pp. CN1-10.

[41] A.H. Behzadan, V. Kamat, Visualization of construction graphics in outdoor

augmented reality, in: Proceeding of the 2005 Winter Simulation Conference, 2005.

[42] A.H. Behzadan, V. Kamat, General-purpose modular hardware and software

framework for mobile outdoor augmented reality applications in engineering, Adv Eng
Inform 22 (1) (2008) 90-105.

[43] I. Brilakis, M.W. Park, G. Jog, Automated vision tracking of project related entities,
Adv Eng Inform 25 (4) (2011) 713-724.

[44] J.C. Martinez, P.G. Ioannou, General-purpose systems for effective construction
simulation, J. Constr. Eng. Manage. 125 (4) (1999) 265-276.

[45] D.W. Halpin, An Investigation of the Use of Simulation Networks for Modeling
Construction Operations, Ph.D Dissertation, University of Illinois at Urbana-Champaign,
Illinois, 1973.

[46] A.A. Oloufa, Modeling operational activities in object-oriented simulation, J.

Comput. Civil Eng. 7 (1) (1993) 94-106.

[47] J.C. Martinez, Stroboscope : state and resource based simulation of construction
processes, 1996.

[48] M. Lu, Simplified discrete-event simulation approach for construction simulation, J.

Constr. Eng. Manage. 129 (5) (2003) 537-546.

[49] M.P. Hunter, R.M. Fujimoto, W. Suh, An investigation of real-time dynamic data
driven transportation simulation, in: Proceeding of the 2006 Winter Simulation
Conference, Monterey, CA, 2006, pp. 1414-1421.

102
[50] S. Tavakoli, A. Mousavi, A. Komashie, A generic framework for real-time discrete
event simulation (DES) modelling, in: Proceeding of 2008 Winter Simulation Conference
(WSC), Miami, FL, 2008, pp. 1931-1938.

[51] X. Guo, S.Y. Huang, W.J. Hsu, M.Y.H. Low, Yard crane dispatching based on real
time data driven Simulation for container Terminals, in: Proceeding of the 2008 Winter
Simulation Conference (WSC), Miami, FL, 2008, pp. 2648-2655.

[52] T.H. Chung, Y. Mohamed, S.M. AbouRizk, Bayesian updating application into
simulation in the North Edmonton Sanitary Trunk tunnel project, J. Constr. Eng. Manage.
132 (8) (2006) 882-894.

[53] S. AbouRizk, A. Hizaji, D.W. halpin, Effect of input modeling on simulation output,
in: Proceeding of Computing in Civil Engineering (New York), 1989.

[54] V.R. Kamat, Visualizing simulated construction operations in 3D, J. Comput. Civil
Eng. 15 (4) (2001) 329-327.

[55] Y. Kui, The ASDMCon project: the challenge of detecting defects on construction
sites, in: Proceeding of the Third International Symposium on 3D Data Processing,
Visualization, and Transmission (3DPVT'06), 2006.

[56] J.J. Shi, Iconic animation of construction simulation, in: Proceeding of the1999
Winter Conference on Simulation, 1999, pp. 992-997.

[57] J.O. Henriksen, General-purpose concurrent and post-processed animation with

Proof, in: Proceeding of the1999 Winter Conference on Simulation, 1999, pp. 176-181.

[58] F. Darema, Dynamic data driven applications systems: new capabilities for
application simulations and measurements, in: Proceeding of Computational Science --
ICCS 2005, Atlanta, USA, 2005, pp. 610-615.

[59] National Science Foundation (NSF), Dynamic Data Driven Application Systems
(DDDAS), 2000 NSF Sponsered Workshop Report,
http://www.nsf.gov/cise/cns/dddas/dd_das_work_shop_rprt.pdf, (accessed Sep. 29,
2011).

103
[60] J. Mandel, Towards a dynamic data driven application system for wildfire
simulation, in: Proceeding of Computational Science - ICCS 2005. 5th International
Conference. , 2005.

[61] C.C. Douglas, Y. Efendiev, A dynamic data-driven application simulation

framework for contaminant transport problems, Journal of Computers & mathematics
with applications 51 (11) (2006) 1633-1646.

[62] M. Gaynor, A dynamic, data-driven, decision support system for emergency medical
services, in: Proceeding of Computational Science - ICCS 2005. 5th International
Conference., 2005.

[63] R. Akhavian, A.H. Behzadan, An Integrated Data Collection and Analysis

Framework for Remote Monitoring and Planning of Construction Operations, Adv Eng
Inform (26) (2012) (in press).

[64] D.R. Nadeau, Visualizing stars and emission nebulas, Computer graphics forum 20
(1) (2001) 27-33.

[65] P. Martz, OpenSceneGraph Quick Start Guide, Pmartz, 2007.

[66] V.R. Kamat, J.C. Martinez, Scene graph and frame update algorithms for smooth
and scalable 3D visualization of simulated construction operations, Journal of Computer-
aided civil and infrastructure engineering 17 (4) (2002) 228-245.

[67] Woolford, D. (2003), Understanding and Using Scene Graphs,

http://www.itee.uq.edu.au/~comp4201/scene_graphs_dsaw.pdf, (accessed Mar. 13,
2012).

[68] OpenSceneGraph (OSG) Website, http://www.openscenegraph.org/, (accessed Mar.

13, 2012).

[69] J.G. Proakis, M. Salehi, Digital communications, 5th ed., McGraw-Hill, Boston,
2008.

[70] PNI, User Manual, CompassPoint Prime, 3-Axis Electronic Compass Module, in:
PNI (Ed.), PNI Sensor Corporation, Santa Barbara, CA, 2011.
104
[71] National Instruments (NI) Website, http://www.ni.com/, (accessed Mar. 13, 2012).

[72] P. Ioannou, Animation of complex construction simulation models, in: Proceeding of

Computing in Civil Engineering (New York), 1996.

[73] C. Microsoft, Automation programmer's reference : using ActiveX technology to

create programmable applications, Microsoft® Press, Redmond, Wash, 1997.

[74] R. Akhavian, A.H. Behzadan, Dynamic Simulation of Construction Activities Using

Real Time Field Data Collection, in: Proceeding of 18th Workshop of Intelligent
Computing in Engineering and Architecture (EG-ICE), Enschede, Netherlands, 2011, pp.
1-10.

[75] A. R. Pradhan, B. Akinci, An Approach for Fusing Data from Multiple Sources to
Support Construction Productivity Monitoring, in: Proceeding of International
Conference on Computing in Civil and Building Engineering, Nottinham, UK, 2010, pp.

[76] J. Dong, D. Zhuang, Y. Huang, J. Fu, Advances in multi-sensor data fusion:

algorithms and applications, Sensors 9 (10) (2009) 7771-7784.

105