Professional Documents
Culture Documents
IRC IH Oct 2018-11-18
IRC IH Oct 2018-11-18
ABSTRACT
Implementation of Quality Control and Quality Assurance (QC/QA) is necessary to ensure consistent and quality
bituminous road construction in India. It should replace the present system of generally reporting “passing” test
results only during construction, which is unacceptable and leads to premature failures.
Quality control is the responsibility of the construction contractor who should prepare and execute a QC plan for the
bituminous paving project. Quality assurance is the responsibility of the owner (such as NHAI, representative of NHAI,
or a concessionaire). For quality assurance paving project is first divided into lots and sublots. Quality assurance is
done with the help of Percent Within Limits (PWL) and pay factors (price adjustments), which are determined for
each lot from test values of sublots.
This paper describes the QC/QA system in detail and the way it should be implemented in India after some pilot
projects. For QA a computer software program based on Excel has been developed at IIT Guwahati exclusively for
this paper and is available online. With this program it is quite easy to calculate the mean, standard deviation, and
PWL once the five sublot test results are entered along with the minimum and maximum specified limits for the test
parameter.
1
Associate Director Emeritus, National Center for Asphalt Technology, Auburn University, USA. Currently in Jaipur,
E-mail: pkandhal@gmail.com
2
Associate Professor, Department of Civil Engineering, IIT Guwahati, E-mail: rajandce@iitg.ernet.in
3
PhD Research Scholar, Department of Civil Engineering, IIT Guwahati, E-mail: abhinayk29@gmail.com
period is 10/12 years. About 90 percent of the total through an independent test/process (supposed to be
construction cost in EPC agreement is regulated by the included under QA), they usually act as “witness” to
government and is paid to contractor in 4 installments tests/checks carried out by the construction contractor.
of 5-6 months each, and the remaining 10 percent is Moreover, in the existing specification/concession
paid during the defect liability period. In the HA mode, agreements, there are no clear guidelines on the
40 percent cost is regulated by the government and acceptance/rejection criteria for any variability in
is paid during the construction period. Remaining 60 the production and construction process. (This paper
percent of the project cost, which was borne by the provides such reasonably clear guidelines.] Under
concessionaire during the construction period, is paid this scenario, contractors or sub-contractors have a
on bi-annual basis during the defect liability period in tendency to bring/ensure all the results (through fair or
installments. unfair means) within the specified ranges or acceptable
As per the Indian specifications and concession limits.They are afraid of some action by the owner if
agreement provisions, QC aspects are to be ensured a result outside the tolerances is reported. This has
through the concessionaire, contractor as well by an been observed by the first author while investigating
independent body. The independent body is called some premature failures of national highways in India.
Authority Engineer in case of EPC agreement and All construction records were of no use because they
showed passing results only despite the fact that NHAI
Independent Engineer in case of the HA mode.
had an Independent Engineer (IE) to oversee the
Most contractors follow the MoRTH Specifications construction quality.
and the Indian Roads Congress Standard
It must be realized that variations in test results of
IRC:111-2009 for Dense Graded Bituminous Mixes,
bituminous mixes is quite natural and common. Test
which specifies the minimum frequency of tests to be
results can be affected by the inherent variability of
performed by the contractor during production and
the materials; asphalt mix; sampling errors; and testing
laying of bituminous mixes[2]. Some examples are as
errors. For example, a segregated hot mix sample can
follows:
decrease or increase the bitumen content test value from
Bitumen content and 3 tests for each 400 tons of what the real value is[3]. Therefore, it is common that
gradation mix (minimum 2 per day) some test results may be outside the tolerance limits of
Marshall tests including 3 tests for each 400 tons of the specifications. The proposed QA process will help
mix (minimum 2 per day) to address such expected variations in bituminous road
voids analysis and & flow construction[4].
Density in the field after 1 test per 700 sq m area 3. MOVE TOWARDS STATISTICAL QA
compaction
SPECIFICATIONS
Specifications provide tolerance limits for the Job-Mix
Formula (JMF) parameters such as follows:
As mentioned earlier, the system in-place in India at
the present time for QA is based on tests performed
Bitumen content ±0.3 %
under QC program, mainly on single samples (so-called
Gradation Varies according to sieve ‘representative’ samples) on the basis of which decision
size (as given in MoRTH is made whether to accept or reject the material/process.
5th Revision)
In cases where results are not found to be within the
Compaction in field Minimum 92% of specified upper and lower limits, additional samples
maximum specific gravity called ‘check’ or ‘confirmatory’ samples are tested. It
of mix
cannot be denied that the existing system creates a lot of
QA should be carried out by the owner (such as NHAI, confusion amongst the agencies responsible for QC/QA
its representative or concessionaire). This aspect of QA regarding judging the overall quality. Therefore, there
is essentially missing currently in India. Agencies that is an urgent need to shift from the current ‘pass or fail’
are responsible for QA rely on the QC process alone. system to a more rational system based on statistical
Under current practice, the owner such as NHAI or its analysis of the results obtained. The statistically based
representative “certify” all the tests/processes/checks QA system will be described in detail, the QC System
done for QC. Instead of ensuring the quality of work is presented first.
This is where statistics is used without any bias as a tool If so desired by the contractor, the owner or his
to make informed decisions and to resolve disputes. representative can split the loose hot mix and provide
Statistical tools such as mean, standard deviation, one-half to the contractor for verification.
normal distribution curve, Percent Within Limits In the statistically based quality evaluation system, two
(PWL), and pay factors are used for quality assurance. or three important quality parameters are selected for
Therefore, QA is conducted on a statistically based testing and evaluation. This is because so many test
quality evaluation system. A detailed discussion of this parameters such as bitumen content; gradation on all
statistical approach follows. sieve sizes; Marshall test void parameters, stability
For acceptance purposes, the paving project is divided and flow; and in-situ density after compaction, make
into lots and sublots. A ‘lot’ means a quantity of the process rather complex. Moreover, many test
material produced from a single source under similar parameters are correlated with each other and there is
conditions. In the context of asphalt pavement, it no need to test all of them.
typically represents asphalt mix production in a single The following three test parameters are widely used for
day. It can also be a specified tonnage such as 400 tons. acceptance and price adjustments and are considered
The project is accepted on lot by lot basis. If one lot reasonably adequate:
is substandard, the contractor is penalized with price ● Bitumen content
adjustment or rejection of that lot only, rather than the ● Air voids in the compacted Marshall specimens
whole project. This reduces the risk for both contractor ● Percent compaction in the field based on maximum
and owner. specific gravity of mix
Sublots are divisions of a lot used for sampling purposes Proper bitumen content is necessary for the performance
and may be of an approximate length of 100-150 m of of the bituminous pavement. Excessive bitumen
bituminous paving. Generally, 5 sublots are considered content would result in bleeding and/or rutting whereas
within a lot. Samples are taken from each sublot for deficient bitumen content would significantly reduce
performing tests such as bitumen content, gradation, the durability of the bituminous pavement[5]. Proper
and Marshall Test. Sampling within the sublot is done air void content in the compacted Marshall specimens
at random by using X and Y coordinates obtained from is also important for the performance of the pavement.
a table of random numbers. If random locations are Air void content of less than 3 percent increases the
selected within the whole lot, they may not be spread potential for rutting whereas air void content above
out as shown in top of Figure 3. Rather, a stratified 5 or 6 percent increases the potential for premature
sampling plan is adopted in which a lot is divided into aging (oxidation) of the bituminous pavement, which
5 equal sublots and then one random sample is obtained may also induce raveling and stripping. Laboratory
from each sublot as shown in bottom of Fig. 3. Such a air voids also indirectly control the mix composition
sampling plan is adopted for obtaining loose bituminous (bitumen content and gradation of aggregate). Percent
mix behind the paver for determining mix composition compaction in the field is the single most important
(bitumen content and gradation) or obtaining cores for test parameter for the performance of bituminous
determining compaction level. Alternatively, loose mix pavement. The composition of bituminous mix may be
perfect but deficient compaction (high air voids in the
can also be collected at random from trucks leaving the
mat) is likely to cause premature deterioration of the
hot mix plant based on time or tonnage.
bituminous pavement such as raveling and potholes.
After the five sublot samples representing one lot are
tested for the desired parameter(s), the test values are
analyzed statistically to determine the percentage of
this lot which is within specified tolerance limits, that
is, Percent Within Limits (PWL).
Based on the mean and standard deviation of the 5
results, a normal distribution curve is fit to the data. A
normal distribution curve is a ‘bell-shaped’ symmetric
curve that describes the statistical distribution of
engineering measurements, such as asphalt binder
Fig. 3 Sampling from Lots and Sublots content, mix density, or gradation data of bituminous
mixes. Fig. 4 shows two normal distribution curves is requirement of % compaction in the field, which is
‘a’ and ‘b’. Observe that both curves have the same 92% minimum of the mix theoretical maximum specific
mean but the standard deviation for ‘b’ is larger than gravity, Gmm.
‘a’, due to which curve ‘b’ has higher spread or scatter
Referring to Fig. 5, PD is defined as the area (expressed
than curve ‘a’. In other words, curve ‘b’ represents high
as percent of the total) under the normal curve lying
variability.
outside (either towards left or right or both) of the
Using the specifications limits, the fitted normal specified upper and lower limits. Consequently, PWL
distribution curve, is further used to compute two can be defined as per Equation 1:
parameters: percent defective (PD) and percent within
limits (PWL). PD indicates percentage of a lot falling PWL = 100 – (PDU + PDL); for double-limit (1a)
outside the specification limits. PWL, on the other PWL = 100 – PD; for single-limit (1b)
hand, indicates percentage of the lot conforming to the
For a fitted normal distribution, the quantities PD and
specification limits. Fig. 5 illustrates the concepts of PD
PWL can be calculated using any convenient software
and PWL for two cases of double-limit and single-limit
specifications. An example of double-limit specification such as MS Excel. Excel includes an in-built function
is the requirement of design binder content to be in the NORMDIST that computes area under a normal
range of ± 0.3% from optimum distribution curve from negative infinity to a given
value with a given mean and a standard deviation.
A computer software program based on Excel has been
developed at IIT Guwahati exclusively for this paper. It
is very easy to use and can be accessed at the following
internet link to calculate the mean, standard deviation,
and PWL once the five sublot test results are entered
along with the minimum and maximum specified limits.
The link for the program developed at IITG is:
http://www.iitg.ac.in/rajandce/homepage/index.html#
Fig. 4 Normal Distribution Curves
[click on “Other Contributions” tab]
4. ILLUSTRATIVE EXAMPLES ON PWL
CALCULATIONS
The following two examples of binder content and %
compaction in the field have been worked out using the
software mentioned above. Review of reported sublot
test values and calculated PWL should familiarize the
reader with the expected trends. To obtain high PWL
(a) Double-Limit Specification values, a lot (with lower and upper limits) should have
the sample mean close to the target (such as JMF binder
content) and low standard deviation (less spread). If
any one condition is not met, PWL is likely to be lower.
4.1 Example 1 on Binder Content
Table 1 gives the binder content test values for 7 lots
(Lot A through Lot G) each lot consisting of 5 sublots.
Figs. 7, 8, 9, and 10 show the normal distribution
curves along with specified low and high limits for
binder content. Lot A represents a case when a good
(b) Single-Limit Specification
quality control is maintained as seen from the mean
Fig. 5 Concept of PD and PWL (same as the optimum binder content of 5.0 %) and a
binder content of 5% (say), i.e. lower limit = 4.7%; upper low standard deviation. For this case, the PWL is 98.4%
limit = 5.3%. An example of single-limit specification (Fig. 6).
(DOTs) in the US, including the state of Pennsylvania small number of test results outside the specification
[7]. Such specifications have significantly improved limits is normal and not necessarily detrimental to the
the quality of hot-mix asphalt produced [8]. It is high performance of the bituminous pavement. Pavements
time that such a system for QC/QA gets implemented with reasonable Percent Within Limits (PWL) lower
in India as well. than 100% should also be accepted.
Table 3 Example of Pay Adjustment Schedule Quality control is the responsibility of the construction
PWL (percent) Pay factor (percent)
contractor who should be required to submit a QC Plan
96-100 105
before executing the paving project. Quality assurance
90-95 PWL + 10
is the responsibility of the owner (such as NHAI, its
representative or a concessionaire) who should divide
60-89 0.5 PWL + 55
the paving project into lots and sublots; test sample from
Below 60 Reject
each sublot; and determine the Percent Within Limits
Based on experience in the US, about 5 to 10% asphalt (PWL) of each lot based on test values of at least 5
paving lots require some price adjustment. Pilot sublots and associated pay factors (price adjustments).
projects need to be considered in India to familiarize
both contractors and owners with the PWL approach A large volume of actual hot mix production test data
and assess the impact of pay factors without actually has been gathered and analyzed in many developed
imposing the price adjustments. countries such as US. Since hot mix plants are similar,
these values can be adopted in India as a starting point.
5.1 Resolution of Differences between Owner and Some QC/QA pilot projects should be undertaken
Contractor Test Results in India as soon as possible to familiarize both the
What if the contractor disputes the owner’s QA test contractors and owners with this concept and to assess
data? This can be resolved by retesting the retained the impact of PWL and pay factors.
sublot samples in presence of the contractor. Retest A computer software program based on Excel has been
does not always give the same test results as shown in developed at IIT Guwahati for this paper. It is very easy
Table 4. to use and can be accessed at the internet link provided
Table 4 Original and Retest Values of Bitumen Contents in the paper to calculate the mean, standard deviation,
and PWL, once the five sublot test results are entered
Test Bitumen contents in sublots
along with the minimum and maximum specified limits.
1 2 3 4 5
Original 6.0 5.6 6.4 5.8 6.1 7. REFERENCES
Retest 6.1 5.7 6.3 5.7 6.1 i. AASHTO. Implementation Manual for Quality Assurance,
1996.
Determine the mean and standard deviation of both lots
(original and retested) and conduct a paired t-test to ii. Indian Roads Congress. Specifications for Dense Graded
Bituminous Mixes. IRC:111-2009.
determine if the two sets of results came from the same
iii. Kandhal, P.S. and S.A. Cross. Effect of Aggregate Gradation
population or not. If yes, the original test results prevail
on Measured Asphalt Content. Transportation Research
and the contractor is charged the cost of retests. If no, Board, Transportation Research Record 1417, 1993.
the retest results prevail and are used for recomputing iv. Roberts, F.L., P.S. Kandhal, and E.R. Brown. Hot Mix Asphalt
the PWL and associated pay factor since testing was Materials, Mix Design and Construction. US Textbook.
done in contractor’s presence. It must be realized that NAPA Education Foundation, Maryland 1996.
decision cannot be made by considering means of v. Kandhal, P.S. Bituminous Road Construction in India.
original and retest values only. Textbook. Prentice Hall of India, New Delhi, July 2016.
vi. Akkinepally, R. and N. Attoh-Okine. Quality Control and
6. SUMMARY Quality Assurance of Hot Mix Asphalt Construction in
Implementation of quality control and quality assurance Delaware. Delaware Center for Transportation Report DCT
(QC/QA) is necessary to ensure consistent and quality 173, July 2006.
bituminous road construction in India. It should replace vii. Pennsylvania Department of Transportation. Specifications
for Highways, Publication 408, 1990.
the present system of generally reporting “passing” test
results only during construction, which is unacceptable viii. Kandhal, P.S., Cominsky, R.J., Maurer, D., and Motter, J.B.
Development and Implementation of Statistically-based End
and highly unfavorable for the long-term performance Result Specifications for Hot Mix Asphalt in Pennsylvania,
of a bituminous paving project. A change in the current Transportation Research Board, Transportation Research
mindset of highway engineers is required to realize that Record1389, 1993.