Sie sind auf Seite 1von 12

International Journal of Transportation Science and Technology 8 (2019) 251–262

Contents lists available at ScienceDirect

International Journal of Transportation


Science and Technology
journal homepage: www.elsevier.com/locate/ijtst

Implementing a pavement management system: The Caltrans


experience
Zhongren Wang ⇑, Tom Pyle
Office of Pavement Management, Pavement Program, Division of Maintenance, California Department of Transportation, 2389 Gateway Oaks Dr. Suite
200, Sacramento, CA 95833, United States

a r t i c l e i n f o a b s t r a c t

Article history: In this paper, the fundamental characteristics of PaveM—the pavement management sys-
Received 26 November 2018 tem for the California Department of Transportation (Caltrans) is introduced first, together
Accepted 6 February 2019 with the various beta-testing efforts to validate and customize the engineering configura-
Available online 15 February 2019
tion of the software. The implementation of PaveM to support the development of the
pavement asset ten-year management plan is then described. The tools and methods
Keywords: implemented to facilitate total investment determination and individual project evaluation
Pavement
and selection are presented in detail. In the end, lessons learned and future initiatives are
Pavement Management System
Implementation
presented.
Ó 2019 Tongji University and Tongji University Press. Publishing Services by Elsevier B.V.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/
licenses/by-nc-nd/4.0/).

1. Introduction

PaveM, the pavement management system for the California Department of Transportation (Caltrans) was implemented
as a Commercial-Off-The-Shelf (COTS) system. The software package supports all the basic functions of a modern PMS, such
as an OracleÒ database and an optimization engine. The database houses all the necessary network inventory, Linear Refer-
encing System (LRS), and pavement condition and project information; while the optimization engine enables optimized
project selection with budget and other constraints. The bulk of the implementation work is to customize this package to
suit California’s pavement management needs, that is, to provide the necessary input, and then test and validate the results,
before apply the package for any intended tasks.
Implementation of PaveM enables Caltrans to be in compliance with the requirements of both the Moving Ahead for Pro-
gress in the 21st Century Act (MAP-21) and California’s Road Repair and Accountability Act of 2017 (Senate Bill 1 or SB 1). As
the data central for pavement condition and project data and an optimization tool for long range pavement financial plan-
ning, PaveM also enables Caltrans to better manage its 50,000 lane miles of the State Highway System (SHS) with unprece-
dented transparency and accountability (State of the Pavement (SOP) Report, 2016).
This paper starts by describing the customized PaveM, focusing on introduction of the engineering configuration includ-
ing network segmentation, LRS, decision trees, performance models, and major inputs, such as APCS, and As Built. Then the
testing and validation effort will be presented, followed by a presentation of the major products developed and implemented
in the ten-year plan process. A summary and lessons learned session will conclude this paper.

Peer review under responsibility of Tongji University and Tongji University Press.
⇑ Corresponding author.
E-mail address: Zhongren.wang@dot.ca.gov (Z. Wang).

https://doi.org/10.1016/j.ijtst.2019.02.002
2046-0430/Ó 2019 Tongji University and Tongji University Press. Publishing Services by Elsevier B.V.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
252 Z. Wang, T. Pyle / International Journal of Transportation Science and Technology 8 (2019) 251–262

2. Network inventory and condition

2.1. Network inventory and Linear Referencing system

Caltrans has a network inventory clearinghouse database for the entire state highway system. The database is called the
Caltrans Transportation System Network (TSN). It houses network attributes such as network geometry, traffic information
and post miles. The Caltrans dual-carriageway LRS is also developed based on TSN. PaveM uses TSN and LRS as its base net-
work descriptions.

2.2. Current pavement condition

Current pavement condition is collected through the long-standing Pavement Condition Survey (PCS) program (State of
the Pavement (SOP) Report, 2016; National Cooperative Highway Research Program (NCHRP), 1995). Before 2011, pavement
condition survey was performed by a dedicated state-owned field crew. Sampled pavement sections were manually rated for
cracking data. The International Roughness Index (IRI) and other sensor data, such as rutting and faulting was collected for all
lanes across the network using state-owned equipment. It takes about three years to complete one cycle of the entire state
highway system. Base on the survey data, Pavement Condition Report (PCR) is then developed for each one-mile segment
and shared statewide. The ‘current’ pavement condition becomes the basis for all project programming-related activities,
and field reviews are performed to validate and adjust the programmed projects.
Year 2011 saw the first application of automated pavement condition survey (APCS) in California. Both cracking and sen-
sor data, such as IRI, faulting, and rutting were automatically collected and analyzed on a lane-by-lane basis. The entire net-
work survey was completed in about eight months. By the time this paper was written, Caltrans has completed four cycles of
APCS data collection for the years of 2011, (State of the Pavement (SOP) Report, 2016), and 2018. APCS was administered as
an annual program. In 2018, semi-automated process was adopted for cracking data collection, in order to improve cracking
data quality. As PCS data, APCS data was used to develop PCR. APCS data is also uploaded into PaveM to enable the prediction
of future conditions.

2.3. History project input

Past project information is the starting point for any PaveM evaluation. Past project information, or as-built is collected
and manually inputted into PaveM, lane by lane, treatment by treatment. A detailed quality control procedure is followed to
ensure quality data. Project postmile limits, layer thickness, type of treatment, and application year are among the most
important as-built data. An army of student forces is hired, together with Caltrans engineers and researchers at University
of California, to ensure data quality and eliminate data backlog. Currently there are more than 5000 pavement-related pro-
jects in the as-built database, dated back to 1978.

3. Engineering configuration

3.1. Network segmentation

Multiple segmentation schemes are accommodated by the software. The common ones include the fine segmentation and
coarse segmentation. Fine segmentation segments the SHS into chunks with a maximum length of one mile. Each segment in
this segmentation is called a management segment. These management segment, once aged using performance models, and
then aggregated temporally and spatially, are able to model the network performance over time. Fine segmentation is effec-
tive in capturing network level performance through piece-by-piece evolution analysis; but it is challenging to propose any
practically meaningful projects. The ‘projects’ predicted using fine segmentation is too small (like a mosaic), and has to be
aggregated both laterally and longitudinally to be used as project limits. A coarse segmentation is thus developed to over-
come such shortcomings (Lea, 2015).
In PaveM, the coarse segmentation has only a total of 2800 segments, and each segment covers about 18 lane miles on
average. As this 18 lane miles include multiple lanes, so the centerline length is about 5 or 6 times longer than that of an
average segment length in the fine segmentation. Coarse segmentation, due to the length of its segments, is not intended
to accurately evaluate segment condition evolutions; instead, it is intended to propose project limits.

3.2. Priority matrix

Priority Matrix refers to the condition evaluation matrix defining the condition rating indices and associated thresholds.
Traditionally, pavement condition is rated using Red, Orange, Blue, Yellow, and Green five colors with Red representing the
worst condition while Green the best condition. Red pavement is Major Structurally Distressed pavement that may call for a
full rehabilitation treatment. Orange and Blue pavements are respectively Minor Structurally Distressed and Poor Ride Only
pavements. These two conditions are typically treated with medium overlays (thickness greater than 0.25 foot). Yellow
Z. Wang, T. Pyle / International Journal of Transportation Science and Technology 8 (2019) 251–262 253

pavement does not suffer from structural distresses but call for preventive or corrective maintenance work. Green-colored
pavements suffer from low to none distress and may not need any treatment.
In 2017, a new set of performance measures were specified by the Moving Ahead for Progress in the 21st Century (MAP-
21) Act. The MAP-21 measures were later on adopted by the California Transportation Commission (CTC), and used to
develop the Strategic Highway System Management Plan (SHSMP) (State Highway System Management Plan, 2017) accord-
ing to the Transportation Asset Management Plan (TAMP) (California Transportation Asset Management Plan, 2017). In the
MAP-21 measures, pavements are rated as Good, Fair, and Poor three categories, based on cracking, rutting/faulting, and
International Roughness Index (IRI) three indices. Rutting index is for asphalt pavements; while faulting index is used for
jointed plain concrete pavement (JPCP). Per definition, a piece of pavement is rated as Poor when two or more of the three
indices exceed the poor thresholds; and rated as Good when all three indices are within the good thresholds; and rated as
Fair, otherwise.
These two sets of performance measurement systems are very different. The distresses used as indices, individual distress
definition, and condition thresholds used are not the same. Both sets performance measures are implemented in PaveM, and
the MAP-21 system is used for all reporting purposes. The traditional system is used to assist treatment strategy selections.

3.3. Decision trees

Decision trees connect designated pavement conditions to the appropriate treatments. In PaveM, only asphalt pavement
(AC) and JPCP are considered. A deterministic set of decision trees are implemented with a total of 35 treatments. Determin-
istic means that there is only one treatment is selected for each branch of the decision trees. The decision trees for AC and
JPCP are shown respectively in Figs. 1 and 2.

3.4. Performance models

For each treatment used in the decision trees, its performance over time is predicted in two steps. First, the evolution of
individual distress, such as cracking, IRI, and rutting/faulting is predicted. Then the condition categories, such as the Good,
Fair, and Poor defined in the MAP 21 are determined by considering each individual distress following the priority matrix.
The models are all age-based single-variable regression models. There are 180 models for asphalt pavements as there are
10 treatments as shown in Fig. 2, three individual indices (IRI, Rutting, and Cracking), two climate zones (Mild and Severe),

Fig. 1. Decision trees for asphalt pavements.


254 Z. Wang, T. Pyle / International Journal of Transportation Science and Technology 8 (2019) 251–262

Fig. 2. Decision trees for jointed plain concrete pavements.

and three traffic levels (Low, Medium, and High). There are 90 models for JPCP as there are five treatments, three individual
indices (IRI, Faulting, and Cracking), two climate zones (Mild and Severe), and three traffic levels (Low, Medium, and High).
Traffic levels are specified based on Equivalent Single Axle Loads (ESAL) per day. The low, medium, and high traffic level
means an ESAL level less than 60,000 per day, between 60,000 and 300,000 per day, and more than 300, 000 per day, respec-
tively. In PaveM, there is no models for composite pavements and continuously reinforced concrete pavement (CRCP).

3.5. Benefit equations

Benefit equations are used as objective functions in PaveM software optimization engine to select and prioritize projects.
The basic concept is illustrated with two performance curves shown in Fig. 3. The left one is for the original treatment, such
as the last rehabilitation job; While the right one is for the proposed treatment. The hatched area between the two curves
represents the benefit of the proposed treatment over the analysis period, T0. The area can be adjusted using project lane
miles, costs of projects. A cut-off distress level, such as A10, can also be used to further adjust the hatched area size so that
the benefit equations will help achieve desired goals in project prioritization.

4. Standard operating procedures

Implementation of PaveM is a systematic undertaking. The general workflow for PaveM operations involves input, output,
and development three major sectors as captured in the pinwheel shown in Fig. 4. The major tasks, stakeholders and time-
lines identified for each major sector form the basis of the PaveM Standard Operating Procedure.
The input sector is the foundation of PaveM operations. As the most important dynamic input to support PaveM opera-
tions, APCS data is scheduled to start in January of each calendar year. It takes about eight months or more complete. Fed
with the latest APCS data, PaveM runs may be conducted to develop pavement condition report and recommended project
lists. Several easy-to-use spreadsheet tools have been developed and published on the PaveM Portal website for statewide
Z. Wang, T. Pyle / International Journal of Transportation Science and Technology 8 (2019) 251–262 255

Fig. 3. Illustration of formulation of a priority equation.

access. paveM portal can be accessed at http://www.ucprc.ucdavis.edu/PaveM-Portal/. The most popular spreadsheet tool is
the Pavement Condition Report, with peak usership reaches more than 50 per weekday.
PaveM entails constant technical support. Caltrans is fortunate to have the full support of the University of California
Pavement Research Center (UCPRC). Performance modeling, decision trees, and PaveM Portal upkeep are just a few items
that UCPRC supports though research contracts. On the other hand, the software vendor, AgileAssets Inc. provides hosting,
software maintenance and support, and continuous operational support. The Office of Pavement Management manages the
day to day operations and development of the entire system. OPM also coordinates with various partners internal or external
to the Department.

5. Testing and validation

5.1. APCS data quality management plan

The quality of Automated Pavement Condition Survey (APCS) data has long been a focus of the Caltrans Pavement Pro-
gram. In 2018, the first official Caltrans APCS Data Quality Management Plan (DQMP) was developed and approved by the
Federal Highway Administration (Pavement Condition Survey Data Quality Management Plan, 2018). The DQMP enables
our pavement condition data to be in full compliance with relevant federal laws and regulations.
The quality assurance (QA) protocol described in DQMP includes both vendor quality control (QC) and Caltrans QA.
Besides reviewing vendor’s QC report, the Pavement Program follows a four-step protocol to perform QA, including
boots-on-the-ground field review, office review, data upload into PaveM databases, and year-by-year consistency checks.
The full QA protocol was implemented in the 2018 APCS data collection cycle to ensure data quality.

5.2. Project location, treatment, and schedule

Before PaveM’s implementation, a project’s limit, treatment and schedule are typically recommended by some senior
pavement engineers, based on current pavement condition information and engineering judgement. When PaveM is used
to recommend projects, the reasonableness of the recommended projects is called into question. A comparison between
the engineer-recommended and the PaveM-recommended projects were conducted to see whether they ‘match’. A ‘match’
is found if the two types of projects have: (a) overlapped limits; and (b) the same or similar treatments within one level of
each other (a thin overlay treatment differs by one level from a medium overlay; while a thin overlay differs by more than
one level from a thick overlay); and (c) the same or similar schedule within two years of each other.
Two funding scenarios are used in the comparison, one with unlimited, and the other one with limited amount of funding.
For the unlimited funding scenario, PaveM was given all the money it needs to select projects according to the decision trees.
For the limited funding scenario, PaveM used a funding that is equal to the total costs of all of the engineer-recommended
256 Z. Wang, T. Pyle / International Journal of Transportation Science and Technology 8 (2019) 251–262

Fig. 4. A Pinwheel depiction of the PaveM Business Cycle.

projects. After the scenarios were run, the list of PaveM- and engineer-recommended projects were plotted side-by-side as
horizontal bars in a spreadsheet. The comparison covered four Caltrans districts and 140 projects as shown in Table 1.
Two general findings were made from this exercise. First, all engineer-recommended projects are identified by PaveM
with the unlimited budget run. The project limit, schedule and treatment match almost 100% of the time for District 3
and 9, and more than 75% for District 10 and 12. In other words, the engineer-recommended projects are a subset of
PaveM-recommended ones. This finding indicated that PaveM is able to identify the same or similar system needs as the
manual programming process. APCS is able to identify system needs as the previous manual pavement condition surveys.
When funding is limited, the matching rate ranges between 18% and 59%. This finding indicated that PaveM prioritized
projects somewhat differently. Due to the various factors engineers used to select and prioritize projects; this is not surpris-
ing, because PaveM may not have all the information engineers used in such prioritization. PaveM could be further config-
ured to better match the manual way of project prioritization. PaveM could also be configured to develop new prioritization
schemes such as minimization of Green House Gas (GHG) emission.

6. Applications

6.1. The biannual pavement asset ten-year management plan

The Ten Year Plan (TYP) is mandated by the legislature and the Streets and Highways Code (Streets and Highways Code
164, 2018). It is part of the SHSMP (State Highway System Management Plan, 2017). The purpose of a TYP is to identify future

Table 1
A comparison between PaveM- and engineer-recommended projects.

District Number of Programmed Project With Limited Budget (%) With Unlimited Budget (%)
D3, Sacramento are 30 20 100
D9, Bishop area 38 18 100
D10, Stockton area 50 41 75
D12, Irvine area 22 59 86
Z. Wang, T. Pyle / International Journal of Transportation Science and Technology 8 (2019) 251–262 257

project and funding needs to achieve performance criteria established for the next 10 years. In California, the TYP is a con-
tinuous planning process, and it consists of the following steps:

1. Document current network conditions;


2. Set the performance goals by end of the TYP;
3. Forecast future conditions and identify needs to reach goals;
4. Identify contribution of all current projects in the project delivery pipeline;
5. Identify performance gaps and determine lane miles to treat, and investment needed;
6. Document year by year project list and investment needed as the TYP;
7. Monitor the delivery progress on a quarterly basis;
8. Repeat the same process every two years to ensure continuous planning, and continuous adjustment.

As a network level pavement management system, PaveM is developed primarily to support network level pavement pro-
gramming effort, i.e. the biannual TYP effort. Specifically, PaveM is to develop the optimal total investment, and the corre-
sponding project list across the duration of a TYP.

6.2. Total investment decision-making support

What is the potential total investment necessary to reach the desired performance goals? With PaveM’s predicting capa-
bility, the pavement program developed a scheme just to answer that question. As shown in Fig. 5, engineers plotted the
10 year and the corresponding 20 year total investment in one chart. The 10-year investment refers to funds just sufficient
to reach the desired goals, say 1% Poor and 39% Fair; and the corresponding 20-year investment refers to funds, in addition to
the 10-year investment, just sufficient to maintain the goals for another ten years. ‘‘Maintaining the goals” means to keep the
total Fair and Poor lane miles steady over the next 10 years (by adjusting investments). Each point shown represents a
PaveM run reaching the desired goals by the end of the 10-year plan period, and then maintaining the goals until the end
of the 20-year plan period. After many points are plotted, it was found that too little or too much 10 year (short term) invest-
ment both lead to higher 20 year (long term) funding needs. One sweet spot exists in the middle of the ‘‘U” shaped curve,
which is deemed as the optimal total investment for a 10-year period. Such analyses offered great credibility to the optimum
total 10-year plan investment.

6.3. Project effectiveness

In project programming, an engineer’s job is focused on identifying the limit, treatment, and schedule of each project. This
is a multi-dimensional decision-making exercise, involving project development considerations, programming policies, and
pavement management philosophy. In PaveM, ‘‘project effectiveness” is introduced to facilitate such decision-making pro-
cess. For a rehabilitation project, its effectiveness is defined as the ratio of the Red lane miles eliminated over the entire pro-
ject limit as shown in Eq. (1). For a CAPM project, the effectiveness is defined as the ratio of the (Red + Orange) lane miles
eliminated over the entire project limit as shown in Eq. (2). For a maintenance project, the effectiveness can be defined as
shown in Eq. (3).
Project effectiveness is a function of project limit (both lateral and longitudinal), project schedule, and deterioration rate
of the treatment applied. Utilizing the predicting power of PaveM, one can examine the evolution of project effectiveness for
any project. Such evolution of a project effectiveness provides a wealth of information critical to program timely and effec-
tive treatment for any specified project limits.

Red Lane Miles


Rehabilitation Effectiv enss ¼  100% ð1Þ
Total Project Lane Miles

ðRed þ OrangeÞ LaneMiles


CAPM Effectiv enss ¼  100% ð2Þ
Total Project Lane Miles

Yellow Lane Miles


Maintenance Effectiv enss ¼  100% ð3Þ
Total Project Lane Miles
Fig. 6 is a project effectiveness evolution chart from 2016 to 2055 for Postmile 0.028 to 16.006 of Route 101 in District 4
(San Francisco Bay Area). Based on the predicted rehabilitation effectiveness, an engineer may choose to wait till after 2035
when the effectiveness exceeds 20% to perform a full rehabilitation job; she/he can also choose to perform a CAPM job as
early as 2024 when (Orange + Red) composition exceeds 40%. A third alternative is to choose preventive or corrective main-
tenance strategies well before 2024 when the maintenance effectiveness exceeds 20%. With predicted project effectiveness
at fingertips, pavement engineers are better informed of possible future pavement performance, and in turn program the
appropriate type of treatment at the right time.
258 Z. Wang, T. Pyle / International Journal of Transportation Science and Technology 8 (2019) 251–262

Fig. 5. Optimum 10-Year Plan Investment Determination.

6.4. H-chart

An H-chart, short for Highway Chart, is a visualization tool for pavement project review. It is a bar chart developed using
Excel spreadsheet depicting all past, current, and future projects for any specified postmile limits and number of years. The
H-chart shown in Fig. 7 is dedicated for the northbound direction of route 101 across Del Norte County in California.
In Fig. 7, the x-axis shows the county odometer and the y-axis shows Year. Projects are shown as hatched and colored
horizontal bars. The project limits (the start and end locations) are located on the x-axis by means of county odometer val-
ues. These county odometer values are computed using the state odometer associated with the start and end of a project. The
year of the project is indicated by its vertical location on the chart.
Four types of projects are displayed in Fig. 7: (1) Completed or as-built, (2) Under Construction, (3) Programmed, and (4)
PaveM Scenario-Recommended. These types of project are indicated in the chart by different cross-hatching patterns. These
types of projects are sorted in time with the earliest known as-built projects displayed at the bottom of page, followed by the
programmed and under construction projects, and finally the recommended future projects at the top. The thick dashed hor-
izontal line separates the scenario-recommended projects from the rest.
Each project shown in Fig. 7 is also associated with a budget group. There are four budget groups used in California. These
include: (1) Highway Maintenance (HM) Preventive; (2) HM Corrective; (3) Capital Program Maintenance (CAPM); and Reha-
bilitation. Each budget group is displayed in a different color. For example, the rehabilitation group is displayed in orange
color.
Each project shown has a label, which displays such project information as treatment type, lane number, Expenditure
Authorization (EA) number and Post-Mile (PM) limits. The contents of the label are configurable before the H-chart is gen-
erated. For example, the label ‘‘CinPlePrecyc-All 01-2T12LAT-II” means that this is a Cold-in-Place-Recycling for ‘‘All” lanes
project. The EA number is 01-2T12LAT. The last letter ‘‘T” in the EA number means this is a scenario-recommended project.
The ‘‘-II” following the EA number refers to the PaveM running Scenario that corresponds to the title of the chart.
Together with the depicted projects, pavement type and pavement condition information is also respectively shown at
the top and bottom of the sample H-chart. The cracking and IRI values are based on the latest available pavement condition
survey. The label for ‘crack%’, such as ‘‘0.3/2:0.7” shown at the lower left corner of Fig. 7 means that the average wheel path
cracking across all lanes from PM 0 to PM 4.4 is 0.3%, and the highest cracking percentage is 0.7%, occurring in lane number 2.
Clearly, a H chart as shown in Fig. 7 provides an effective visualized working environment for a pavement engineer to
evaluate, compare, adjust, and select projects for programming and planning purposes. With pavement condition, as-built
projects, programmed projects, and scenario-recommended projects visualized in one chart, it becomes an easy process
to identify gaps or overlaps in project schedule or limits. For example, project 01-49940 and 01-08080 overlap in schedule.
Z. Wang, T. Pyle / International Journal of Transportation Science and Technology 8 (2019) 251–262 259

Fig. 6. Project effectiveness evolution from 2016 to 2055.

Fig. 7. A Sample H-Chart.

It may not make sense to apply a medium overlay on top of a thin overlay that just applied a year ago. On the other hand, the
segment from PM 20 to 25 has not been treated ever since the year of 1994. Why is there such a long hiatus? Is it because of
lack of funding, or incomplete project history? Answers to these types of questions may well help better plan for the current
and future projects. Definitely, the H-chart shown may also be carried to the field to help adjust project limits with boots on
the ground.
260 Z. Wang, T. Pyle / International Journal of Transportation Science and Technology 8 (2019) 251–262

7. Lessons learned and future work

7.1. Is PaveM to replace human beings?

A major concern to implement PaveM is whether the system will replace human beings to make decisions for pavement
improvement projects. Before the implementation, it was made clear that PaveM is to provide information to support
decision-making, and not to make decisions. As a computer program, it reasons based on information and logics provided.
PaveM will not be able to consider situations not being provided for.
On the other hand, PaveM has way better memories, and way more consistent and reliable than human beings in reason-
ing based on established logics. At the network level, PaveM provides a screening of system needs and thus total funding
need estimate. Such information will greatly facilitate network management, although should never replace individual pro-
ject repair strategy determination. It is the corporate memory and support workforce succession planning.

7.2. Eliminate the blackbox syndrome

To many of our engineers, PaveM is like a black box. In order to dispel the myths, win confidence of our engineers, and
make sure the system is doing what it is configured to do, various efforts are attempted to eliminate the black box syndrome.
The first attempt is to clearly describe and publish the engineering configurations, including version control. A user working
group, called Yoda team is developed as a PaveM forum to bring users, developers, and administrators together to increase
communications. For example, any changes made in PaveM can be distributed on a timely basis. Various hand-calculation
was performed to ensure that PaveM produce the results that is supposed to. As a continuous process, the validation effort
is proved to be the most effective to dispel myths, and win engineer’s confidence.

7.3. LRS update

LRS update is one of the key maintenance issues with PaveM. Automated process is needed. Ramps and connectors are
being added to the system, and this will increase the level of complexity of maintaining it.

7.4. Project data maintenance

The importance of an accurate project database cannot be overemphasized. Past projects dictate the starting points and
performance models for future prediction. Current and future planned projects are ‘designated’ or ‘mandated’ ones that must
be delivered before PaveM can be used to hunt for any additional projects. Clearly, any inaccuracy in the list of mandated
projects will influence the entire delivery plan.
Project data maintenance is a routine line-function. Effort is being made to exchange of information between project
databases to eliminate repeated manual key-ins. Automation will also improve data integrity and accuracy. Ideas are
brought up to decentralize the project database maintenance task, i.e. to ask the district personnel to maintain their own
data in the system. However, as more hands at the same work may only make it more difficult to control the data quality,
Caltrans is maintaining central control of the PaveM system.

7.5. MAP-21 measures vs. Traditional Caltrans measures

MAP-21 measures presented challenges for strategy selection, especially for jointed plain concrete pavement (JPCP). In
MAP-21 measures, the cracking index is based solely on transverse cracking. With longitudinal and corner cracking ignored,
the repair needs do not align with practical knowledge developed in California for decades. Some need patterns such as that
shown in Fig. 8(a) Corner cracking, and 8(b) Longitudinal and transverse cracking are left out. In the meantime, some trans-
verse cracking only type of need may be treated more than it deserves. It is hoped that the MAP-21 measures will improve
and adjust this in the near future so that state transportation agencies can count on these measures to identify repair needs
and appropriate treatments.
Another way to improve the MAP 21 measures is to further breakdown the Fair category by specifying clearly the reason
behind the Fair rating. For example, the Fair category can be divided into ten subcategories, and the Poor two subcategories
as shown in Table 2. GGG means all three indices are rated Good; FF means two indices are rated as Fair; PP means two
indices are rated as Poor; so on and so forth. The ten subcategories can be further aggregated to help establish the relation-
ship between condition rating and the most appropriate treatment strategy.

7.6. Long-term commitment

The implementation of a COTS system is a long-term commitment. The system is procured from a sole-source provider,
meaning the current vendor is the only qualified provider not only for now, but also for any future customization, upgrade,
and system maintenance and support. Termination of the current vendor halfway is very challenging, because that will mean
Z. Wang, T. Pyle / International Journal of Transportation Science and Technology 8 (2019) 251–262 261

Fig. 8. Distress not considered in MAP 21 Measures.

Table 2
Breakdown of the MAP 21 Performance Measure Categories.

MAP-21 Condition COMBINATIONS (Cracking, Rutting/Faulting, and IRI) CATEGORY


GOOD All measures are Good GGG 1
FAIR One measure is Poor P, P, P, 10
One measure is Fair F, F, F,
Two measures are Fair FF, FF, FF,
All measures are Fair FFF
POOR Cracking-Related Poor PP, PP, PPP 2
Non-Cracking-Related Poor PP

Fig. 9. Asset extraction using APCS images.

to forgo whatever already accomplished and re-start the procurement process. Such a long-term commitment should be well
recognized at the very beginning of the procurement to avoid later-on surprises and ensure continued management support.

7.7. APCS imagery data storage and access

APCS is an annual event. Each APCS cycle, Caltrans receives more than 60 TB of data, most of which are right-of-way
images. Due to the various needs of these images, at least ten years should be stored, with one or two years stored online
for quick access. Such large amount of imagery data proves to be costly to store, and technically challenging to ensure easy
statewide access. Any potential bidder may win the APCS contract. Different winners may use different image data storage
262 Z. Wang, T. Pyle / International Journal of Transportation Science and Technology 8 (2019) 251–262

format, and different image viewer software. Compatibility of image viewer and analyzer software packages between differ-
ent vendors must be required across different vendors in the APCS contracts.
APCS data is very attractive for many programs in Caltrans and outside of Caltrans. Lawyers may use them in legal case
discoveries. Bridge deck cracking review, asset management databases, and project reviews. Fig. 9 shows how a sign post
asset can be measured and inventoried based on APCS images. Continued effort is being made to attract more usage of APCS
images to reduce unnecessary field trips and associated costs. The importance of a successful and timely APCS contract
becomes ever more critical. In order to ensure timely delivery, caution has to be exercised not to overload this contract.

8. Summary

PaveM is a network level pavement management system implemented in California in recent years. It uses APCS as cur-
rent condition data input and support a $20 billion-dollar Ten Year Plan for the pavement asset. As the tool to gauge pave-
ment performance into the future years, it is an integral component of the entire Caltrans asset management practice. With
the ever-increasing requirements for investment transparency and accountability, PaveM and the data driven philosophy is
weaving into daily activities at Caltrans. On a quarterly basis, the performance of programmed and planned projects based on
PaveM are to be reported to the California Transportation Commission. Efforts are underway to provide data governance and
interactive maps to facilitate the publication and usage of PaveM database and predicting capabilities, in post project eval-
uation, and design strategy forensic investigations.

Disclaimer and Acknowledgements

The contents of this paper reflect the views of the author who is solely responsible for the facts and accuracy of the data
presented herein. The contents do not necessarily reflect the official views or policies of the State of California. The numbers,
especially financial numbers are for illustration purposes only. This paper does not constitute a standard, specification, or
regulation. Support from Nagi Pagadala, Venkata Mandapaka, Haiping Zhou, Robert Sherrick, and Lan Nguyen are gratefully
acknowledged.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.ijtst.2019.02.002.

References

2015 State of the Pavement (SOP) Report, California Department of Transportation, Sacramento, 2016.
National Cooperative Highway Research Program (NCHRP), Synthesis of Highway Practice 222. Pavement Management Methodologies to Select Projects
and Recommend Preservation Treatments, Kathryn A. Zimmerman and ERES Consultants, Inc. National Academy Press, Washington D.C. 1995.
Lea, Jeremy D., 2015. Grouping pavement segments to form realistic projects. Transp. Res. Rec.: J. Transp. Res. Board 2523, 64–71. https://doi.org/10.3141/
2523-08.
2017 State Highway System Management Plan, prepared in accordance with Streets and Highways Code 164.6. California Department of Transportation,
Sacramento, June 2017.
California Transportation Asset Management Plan, Fiscal Years 2017/18-2016/27. California Department of Transportation, Sacramento, January 2018.
Pavement Condition Survey Data Quality Management Plan. California Department of Transportation, Sacramento, March 2018. https://maintenance.
onramp.dot.ca.gov/paveprogram/pavement-management.
Streets and Highways Code 164. State of California, Sacramento. Available online at: http://leginfo.legislature.ca.gov/faces/codes_displaySection.xhtml
accessed on July 27, 2018.

Das könnte Ihnen auch gefallen