You are on page 1of 7

Transportation Research Record 1813 ■ 253

Paper No. 02-3941

Contractor-Led Quality Control and


Quality Assurance Plus Design–Build
Who Is Watching the Quality?

Jim Ernzen and Tom Feeney

Recent innovations by the Arizona Department of Transportation (DOT) that must remain open during construction. The lure of these alter-
in the use of design–build procurement for highway construction are pre- native methods lies primarily in their ability to compress the sched-
sented. Explosive population growth in Arizona has pushed its DOT to ule by overlapping design and construction and to create team syn-
the limits of its capacity and has challenged the department to develop ergies that can often expedite the project. Sanvido surveyed 351
innovative ways to stretch its resources to meet its constituents’ needs. In building projects from 1992 to 1997 for the Construction Industry
1996 the department spearheaded the passage of a pilot design–build law Institute and reported that design–build projects were completed on
aimed at completing public-sector construction projects more rapidly average 33% faster than design–bid–build projects (1). So, although
than could be done by traditional methods. An evaluation of the mate- the increased delivery speed is appealing to state highway agencies
rial quality program used in the second design–build project in this pro- that are constantly under pressure to open lanes sooner, the concern
gram is described. The project reconstructed an extremely congested is whether they are sacrificing quality for the sake of speed. Sanvido’s
7-mi segment of Interstate 17, a primary artery carrying 180,000 vehi- study also surveyed the projects for quality; however, the data
cles per day through the city of Phoenix, widening it from 6 to 10 lanes. gathered were rather soft in nature, asking the users if they were
The design–build contract was awarded after A+B bidding, which con- “satisfied” with different aspects of the structure and so forth. The
sidered the bid price to do the work and the time required to complete results from the Construction Industry Institute survey for qual-
the project, and was the largest ever awarded at the time. It was won by ity, not surprisingly, showed very little difference in the quality
a design–builder who implemented a very aggressive schedule that of the work between projects constructed by the different delivery
required double-shift work for nearly 2 years. In another contracting methods.
first, the agency also assigned the design–builder responsibility for the Although the pressure for faster performance mounts, state high-
quality control and quality assurance functions on the project, with way agencies are also finding that the workforces needed to perform
Arizona DOT providing verification sampling and testing only. The con- these increasing numbers of tasks are shrinking. Higher wages in the
crete compressive strength and material density for the project are private sector are luring many personnel away from the state agen-
examined and are compared to statewide averages for traditional cies, and that makes their tasks doubly hard. Although national
design–bid–build projects in which Arizona DOT performed the qual- statistics are hard to determine, one midwestern state found that
ity assurance function. Analysis of the data shows that despite a highly although its workload has tripled since 1980, its personnel level
compressed schedule, the quality of the material on the project exceeded dropped by 18.5% during the same period (Dave Cox, Federal High-
the project specifications and was similar to the quality of work com- way Administrator, State of Oregon, personal interview, June 15,
pleted for the state under traditional contracting methods with an Arizona 2001). To combat this, some state agencies are contracting the
DOT–operated quality assurance program. responsibility for quality control (QC) and quality assurance (QA),
a domain once strictly held by the agency. This combination of in-
creased pressure to perform faster along with decreased numbers of
State departments of transportation face a wide variety of challenges personnel and an increased dependence on outsourcing of the quality
in serving their state constituencies. The public and the media often function has some in the transportation community concerned.
criticize the performance of agencies without always understanding
the complexity of the issues that public transportation officials face,
including budget constraints, ecological concerns, and growth or PROJECT DESCRIPTION
no-growth initiatives. State agencies are increasingly under public
pressure to complete all aspects of their work—from long-term Explosive population growth over the last decade in the Southwest
transit planning, to new road construction, to local maintenance has pushed the Arizona Department of Transportation (DOT) to the
operations—at a faster pace, usually with fewer personnel. limits of its capacity and has challenged the department to develop
An increasing number of state highway agencies are turning to innovative ways to stretch its resources to meet its constituents’
alternative project delivery systems such as design–build to help needs. In 1996 Arizona DOT sponsored a pilot design–build pro-
them deliver projects more quickly. Some agencies have found that gram that was aimed at completing public construction projects
this is especially advantageous when they are replacing or upgrad- more rapidly than could be done by traditional methods. The second
ing sections of urban freeways that are overloaded with traffic and project in the program was the reconstruction of 7 mi of Interstate
17, a primary artery carrying 180,000 vehicles per day through the
Del E. Webb School of Construction, Arizona State University, Box 870204, city of Phoenix, widening it from 6 to 10 lanes. The $80 million proj-
Tempe AZ 85287-0204. ect was the second in the design–build pilot program and at the time
254 Paper No. 02-3941 Transportation Research Record 1813

was the largest single contract awarded by the department (Doug requirements. The incentive system that Arizona DOT used for
Nintzel, Public Affairs Officer, Arizona DOT, personal interview, material strength test and smoothness results was exactly the same
June 2000). as the system that it has used for many years on its traditional
The project scope created one high-occupancy vehicle lane in design–bid–build projects. Arizona DOT paid particular attention to
each direction, constructed auxiliary lanes in both directions be- the balance between schedule incentives and quality incentives, with
tween each of the eight traffic interchanges (TIs), reconstructed the 49% of the incentives offered for schedule compression and 51%
on- and off-ramps at each TI, and added several miles of sound offered for material quality.
walls near residences. It also installed traffic-monitoring cameras This paper reports on the quality of the concrete compressive
and added variable-message signs, ramp meters, and traffic-counting strength and the material density test data for the project completed
equipment, linking I-17 to an existing freeway management sys- by the design–build team. These results are then compared with aver-
tem. The scope of the project also called for the demolition of age values gathered from other Arizona DOT projects contracted via
bridges at two major crossroads, each carrying 40,000 vehicles per the traditional design–bid–build manner in which Arizona DOT
day and the reconstruction of eight-lane single-point urban inter- provided the QA function.
changes. The freeway was widened entirely within the existing
right-of-way corridor, so the project site was extremely congested.
The work also abutted many politically influential businesses in the QUALITY FUNCTION
downtown Phoenix area, so the successful execution of the project
was deemed critical to the success of the entire statewide design– Although Arizona DOT had experimented with contracting out the
build program. Figure 1 shows a drawing of the roadway before and QA function, this was the largest and highest-profile project in
after construction. which it placed the responsibility for both QC and QA on the entity
contracted to perform the work. In the technical specifications
Arizona DOT defined the roles and responsibilities of the three
CONTRACT DESCRIPTION entities to be involved with the quality management plan: the design–
builder, Arizona DOT, and an independent assurance entity. Fig-
The contract was awarded by a two-step best-value methodology ure 2 shows an organizational chart of the roles and responsibili-
based on a combination of technical merit and price to a joint ven- ties. Although the design–builder controlled the quality manage-
ture formed by Granite and Sundt Constructors teamed with URS ment plan, the construction QC and QA, and the design QC and
Greiner as the principal design firm. Because of the location of the QA, both the design–builder and Arizona DOT shared the respon-
project and the effect that the traffic congestion would have on sibilities for the independent assurance, which was performed by
nearby business owners, the price proposal used a method known as a previously approved local testing laboratory. Arizona DOT de-
A+B bidding, which considered the bid price to do the work and the signed the roles during the proposal stage so that, although the
time required to complete the project. Although Arizona DOT esti- design–builder led the team, Arizona DOT maintained the final re-
mated the project duration to be 900 days, the winning design–build sponsibility for acceptance of the workmanship, the materials, and
team proposed to complete the project for $79.7 million in just the design and periodically audited the contractor data to ensure
609 days. The project also included approximately $3.8 million in the adequacy of the design–builder’s procedures (2). Throughout the
financial incentive clauses with a schedule component and a quality project, Arizona DOT performed random verification testing and
component. Arizona DOT offered a financial incentive for early an independent assurance sampling and testing of the final prod-
completion of the two major crossroad bridges, the auxiliary ramps, uct at a specified frequency of one test for every five tests performed
and the overhead freeway lighting. On the quality side, Arizona by the design–builder to check the reliability of the design–builder’s
DOT also paid incentives for superior performance of the public tests. However, the design–builder detailed and managed the plan
relations and quality management programs and for exceeding pave- to work with many different entities involved with both design and
ment smoothness, material strength, and construction workmanship construction. The design–builder performed tests according to a
predetermined schedule and actually performed more tests than
required by the specifications and the Materials Division within
Arizona DOT.
When a particular material was selected for verification testing,
three split (or companion) samples were taken; one was under con-
trol of the design–builder and the other two were under the control
of Arizona DOT. Two of the samples were tested immediately,
with one of the Arizona DOT samples being held in reserve for dis-
pute resolution. If the test results were similar for the two samples
tested, the test result was accepted. If the two test results were not
similar, the design–builder and Arizona DOT would work together
to find the source of the discrepancy. If no source of test discrep-
ancy could be found, an independent testing laboratory then
tested the final sample; the results of this test were binding on
both parties, and the cost for such testing would be borne by the
party found to be in error. Figure 3 shows a diagram outlining how
this process was designed, and Table 1 outlines the allowable vari-
ations between the design–builder and Arizona DOT test results
FIGURE 1 I-17 roadway before and after construction. under the verification program.
Ernzen and Feeney Paper No. 02-3941 255

Design Builder

Corporate Management

Quality Construction Design Production


Management Production (Quality Assurance)

Technical
Review

Workmanship Materials Workmanship Design


Inspection Control Control Control

Materials Design
Acceptance Oversight
(If Verified) Review

ADOT

Central/ District Office


Regional
Offices

Project Manager/Resident Engineer


(Acceptance Decisions)

Workmanship
& Materials Materials Workmanship
Independent Verification & Oversight &
Assurance Acceptance Acceptance

Correlation

FIGURE 2 QC and QA roles and responsibilities (ADOT  Arizona DOT).

DESIGN–BUILD PROJECT CONCRETE DESIGN depending on their use on the project. The different concrete classifi-
STRENGTH AND TEST DATA cations were primarily used for the following different activities:
Class S 3,000-psi concrete was used on sound walls and footings,
The project included large quantities of concrete, which was used for Class S 3,500-psi concrete was used for all drilled shaft founda-
a variety of construction activities such as footings, sound walls, tions, Class S 4,000-psi concrete was used on median barriers, and
drilled shaft foundations, and portland cement concrete pavement. Class S 4,500-psi concrete was used on all bridge decks. Furthermore,
The design–builder, being familiar with Arizona DOT and its mate- all portland cement concrete pavement was designed with Class P
rial procedures, instituted a project-level concrete mixture classifica- 4,000-psi concrete, whereas nonstructural elements such as curbs
tion system similar to that which Arizona DOT uses on its traditional and gutters and all sidewalks were designed with Class B 2,500-psi
work. The designer specified three classes of concrete: Class S for concrete. The specifications mandated that all concrete strength
structural concrete, Class B for nonstructural concrete, and Class P requirements be met at 28 days of age, although the design–builder
for concrete pavement. There were four different Class S structural did on occasion use additional mixtures that met the 28-day strength
concrete mixtures with various compressive strength requirements, requirement at an earlier date to facilitate the construction schedule.
256 Paper No. 02-3941 Transportation Research Record 1813

Sample Accepted by QC

Test by QA Team Verification by ADOT

NO
Results Compare Contractor & ADOT
Favorably Investigate Issue

YES
NO
Tests are Used Discrepancy
for Acceptance Solved

Use of Verification
Test to Accept or
Reject

FIGURE 3 Materials acceptance flowchart.

TABLE 1 Allowable Variations for Design–Builder and Arizona DOT Test Results Under the
Verification Program
SOILS
VERIFICATION TEST
TEST vs.
DESIGN-BUILDER TEST +/-
Plasticity Index 3
# 200 Sieve, % 1.5
Optimum Moisture, % 1.0
Proctor Density, pcf 2.0
Compaction, % 2.0
AGGREGATE (Base, Sub-base, Backfill, and Mineral Aggregates)
VERIFICATION TEST
TEST vs.
DESIGN-BUILDER TEST +/-
Sieve Analysis, %
# 16 and larger 4.0
# 40 3.0
# 200 1.5
Fractured Faces 9
Sand Equivalent 5
Plasticity Index 2
Moisture, % Optimum 1.0
Proctor Density, pcf 2.0
Compaction 2.0
PORTLAND CEMENT CONCRETE
VERIFICATION TEST
TEST vs.
DESIGN-BUILDER TEST +/-
Coarse Aggregate
# 8 and larger sieves, % 4.0
Fine Aggregate
# 4 mm, % 4.0
Larger than # 200 to less than # 4, % 3.0
# 200, % 1.5
Sand Equivalent 5
Air Content, % 0.5
Slump, inches 0.5
Temperature, °F 2
28-day Compressive Strength, % 15
Ernzen and Feeney Paper No. 02-3941 257

TABLE 2 I-17 Design–Build Project Concrete Compressive Strength Data


Typical Use On Total Number Average 28-day
Specified Concrete Material The Project of Tests Taken Strength (psi)
Class "S" Concrete Bridge
4500 psi Decks 44 5294
Class "S" Concrete Median
4000 psi Barriers 245 5272
Class "S" Concrete Drilled Shaft
3500 psi Foundations 101 4501
Class "S" Concrete Sound Walls
3000 psi and Footings 864 4664
Class "B" Concrete Sidewalks and
2500 psi Curb and Gutter 331 3897
Class "P" Concrete Concrete
4000 psi Pavement 315 5139
TOTAL NUMBER OF
CONCRETE TESTS 1900

Throughout the construction process the design–builder obtained by the traditional design–bid–build method, with Arizona DOT pro-
samples and recorded 1,900 compressive strength tests, of which viding the QA testing for the project. These test results were also the
25 failed to meet the strength requirements of the specifications. averages for three cylinder breaks. The classification and strength of
Although statistics on failed test results are not kept by Arizona the concrete in the historical data correlated directly with the classi-
DOT, conversations with its materials specialists revealed that the fication and strength of the design–builder’s concrete; however, be-
test failure rate on the I-17 design–build project was no higher than cause of specification changes and the introduction of new mix de-
the rate that it typically experiences on its traditional projects in signs, the exact mix designs were not compared; only the specified
which it does the QA work. In fact, the design–builder performed compressive strengths were compared. Table 3 shows the average
many more tests than required by the contract specifications and in compressive strength of concrete from Arizona DOT historical data
this light provided Arizona DOT with a much higher level of QC along with the number of tests recorded.
than is typically received (Randy Allenstein, Arizona DOT materi-
als specialist, personal interview, July 2001). In addition, 535 slump
tests were conducted on the delivered concrete materials. The strength CONCRETE DATA COMPARISON AND ANALYSIS
test results were the averages for three 4- × 8-in. cylinder breaks.
Table 2 shows the classifications of the material, the specified The I-17 project data were compared with Arizona DOT historical
strength, the typical use on the project, the number of test samples data, and conclusions can be made from the individual classifica-
taken, and the average strength obtained from the tests. tions. Table 4 shows the average compressive strength for each
classification and each data set. The most important thing to note
is that none of the test results even approaches the point of falling
ARIZONA DOT HISTORICAL CONCRETE DATA below the specification strength. The analysis becomes a question
of which data set exceeded the specified strength by a greater mar-
The Arizona DOT operates several regional testing laboratories gin. Closer inspection reveals that the strengths of the Class P con-
across the state, and these maintain databases of historical test results crete and the Class S 4,000-psi concrete are virtually equal for the
for concrete samples. The local Phoenix Office provided test results two data sets, with the strength of the concrete for each data set
for 7,663 samples obtained from 1996 through 2000 for comparison exceeding the specified strength by 32%. When the other four clas-
with the results for the I-17 project. These projects were all contracted sifications are factored in and all the scores are adjusted for their

TABLE 3 Arizona DOT Historical Baseline Concrete Compressive Strength Data


Total Number Average Compressive
Specified Concrete Material of Tests Strength (psi)
Class "S" Concrete
4500 psi 633 5646
Class "S" Concrete
4000 psi 557 5299
Class "S" Concrete
3500 psi 896 5057
Class "S" Concrete
3000 psi 2945 5152
Class "B" Concrete
2500 psi 681 4662
Class "P" Concrete
4000 psi 1951 5129
TOTAL NUMBER OF
CONCRETE TESTS 7663
258 Paper No. 02-3941 Transportation Research Record 1813

TABLE 4 Comparison of Concrete Compressive Strength Data

Specified Concrete Material Average Compressive Strength


ADOT Historical I-17 Design-Build
Baseline Data Project Data
(psi) (psi)
Class "S" Concrete
4500 psi 5646 5294
Class "S" Concrete
4000 psi 5299 5272
Class "S" Concrete
3500 psi 5057 4501
Class "S" Concrete
3000 psi 5152 4664
Class "B" Concrete
2500 psi 4662 3897
Class "P" Concrete
4000 psi 5129 5139

percentage of the total number of tests, the average concrete com- culated average density, which ranged from a low of 96.49% to a
pressive strength for the I-17 design–build project exceeded the high of 99.25%.
specified level by 45%. The average strengths for the Arizona
DOT historical database surpassed the specification requirement
by 52%. Of interest here is the fact that both the design–build proj- ARIZONA DOT HISTORICAL DENSITY TEST DATA
ect and the projects whose data are included in the Arizona DOT
historical database operated under the same pay adjustment pro- The Arizona DOT maintains a database of density test data similar
gram, which adjusts pay up or down on the basis of the increases to the database of concrete test data. The database includes data
or decreases in material quality based on a statistical standard. In from all projects across the state of Arizona. However, the informa-
other words, a new set of incentive rules was not applied to the tion in the database is organized slightly differently from the I-17
design–build project. project’s data. The data gathered from the database represent those
from projects with a wide range of characteristics, from large to
small contract value, from bridge to freeway construction, and
I-17 DESIGN–BUILD PROJECT MATERIAL from complicated to fairly simple, all of which were completed be-
DENSITY TESTS tween 1999 and 2000. Overall, the data represent a wide cross sec-
tion of 101 projects that were under construction at that time, with
Three types of density tests were performed: aggregate base ma- a total of 1,034 test results examined. Table 6 lists the density test
terial density tests, drainage density tests, and earthwork density used as well as the average density, the number of samples tested,
tests. The drainage and the aggregate base density tests consisted and the number of projects for each test category. The density re-
of proctor tests done with imported or mix materials used as sub- sults from the Arizona DOT historical database ranged from 97.64%
base materials for either pipe bedding or footings, whereas the to 101.13%.
earthwork density tests were completed with natural materials,
which required compaction before backfill was placed. Therefore,
the earthwork density tests evaluated a combination of workman- DENSITY DATA COMPARISON AND ANALYSIS
ship quality and the quality of the natural materials present on the
project site, whereas the drainage density tests and the aggregate A side-by-side comparison of the results of the density tests for the
base density tests evaluated a combination of workmanship qual- I-17 design–build project and the data in the Arizona DOT histori-
ity and the quality of the imported materials. The design–builder cal database is presented in Table 7. Again, the discussion is not one
tested samples throughout the duration of the project, taking sam- of failing materials but one of which data set exceeded the speci-
ples for a total of 1,552 tests. These test results were gathered, and fications by how much. The I-17 density specifications were 95%
an average density was calculated for each category. Table 5 shows compaction in each of the three tests, and on average, the project
the type of density test, the number of samples tested, and the cal- data clearly met that specification. The Arizona DOT historical data-

TABLE 5 I-17 Design–Build Project Density Test Data


Test Number Average
Description of Tests Density
Drainage Density 675 96.49%
Earthwork Density 591 97.70%
Aggregate Base Density 286 99.25%
Total Number of Tests 1552
Ernzen and Feeney Paper No. 02-3941 259

TABLE 6 Density Classification and Sample Size from Arizona DOT Baseline Data

Test Description Number of Projects Number of Tests Average Density


Drainage Density 20 181 98.33%
Earthwork Density 48 590 97.64%
Aggregate Base Density 33 263 101.13%
Total Density Projects and Tests 101 1034

TABLE 7 Material Density Test Data Comparison


Test ADOT Baseline Data I-17 Design-Build Project Average
Description Average Density Density
Drainage Densities 98.33% 96.49%
Earthwork Densities 97.64% 97.70%
Aggregate Base Densities 101.13% 99.25%
Total Cumulative Average 99.03% 97.81%

base includes data for many projects with various specification re- project delivery methods return quality equal to or better than that
quirements, so it was not possible to measure those data against a obtained by the traditional methods.
single standard percentage. The average for the I-17 project data was The reconstruction of Interstate 17 in Phoenix, Arizona, used two
equal to the statewide average in one of the three density categories innovative concepts: design–build and a contractor-led quality man-
(earthwork density) and was slightly below the statewide average in agement program. The I-17 project incorporated both of these con-
the other two categories. One would expect more scatter in this com- cepts into one large public reconstruction project in the center of the
parison because the data are for work done with widely different in- seventh largest city in the United States. Despite the constraints of
place materials in different areas of the state. However, the two cat- working on one of the most heavily traveled roads in the state, the
egories in which the design–builder’s averages failed to meet the design–build team of Granite and Sundt Contractors completed the
statewide averages included workmanship as well as imported ma- project in 30% less time than Arizona DOT had originally estimated.
terials, although the differences were small. Discussions with Ari- A review of the materials testing data showed that despite the com-
zona DOT materials personnel indicated that in the contracting com- pressed schedule and difficult working conditions, the results of the
munity technician knowledge of density testing is often not as strong concrete strength tests exceeded the specification requirement by
as knowledge of concrete strength testing, so this may be an area in 45%, on average. This compared very favorably with the results for
which Arizona DOT needs to remain vigilant (Randy Allenstein, traditional Arizona DOT projects whose data are included in a his-
Arizona DOT materials specialist, personal interview, July 2001). torical database, for which the average compressive strength test
However, the availability of good-quality imported subbase ma- result surpassed the specification by 52%. Likewise, the density tests
terials in the Phoenix metropolitan area has recently begun to de- showed similar results, with the average densities for the design–
cline, which could explain the differences between the two data sets build project being between 96.49% and 99.25% and those for the
(Jay Guertin, Arizona DOT project engineer, personal interview, traditional projects ranging from 97.64% to 101.13%. Therefore,
September 2000). although this comparison shows that the material quality is margin-
ally higher with traditional design–bid–build contracting and agency-
controlled QA, the differences are very small and the results are very
SUMMARY encouraging for other state highway agencies that are considering
moving in this direction and that are concerned about project quality.
It is inevitable that state departments of transportation will continue
to face more complex projects than ever before and be pressured to
do more, faster, with less. Current alternatives for these agencies are REFERENCES
to use alternative delivery systems or to create incentives for sched-
ule compression and to shed more of the traditional agency missions 1. Sanvido, V. Project Delivery Systems: CM at Risk, Design–Build,
such as QA and QC to the contracting community. As state highway Design–Bid–Build. Research Report 133-11. Construction Industry Insti-
agencies move further in this direction, it is incumbent on them to tute, Austin, Tex., April 1998.
2. Allenstein, R. Materials Sampling and Testing Report. Arizona Depart-
first plan carefully during the procurement phase to ensure that they ment of Transportation, Aug. 1999.
choose qualified teams. They must then draft contracts and specifi-
cations that put sufficient checks and balances in place so that these Publication of this paper sponsored by Committee on Construction Management.

You might also like