Académique Documents
Professionnel Documents
Culture Documents
by
Bhupinder Singh, Ph.D.
Associate Professor
Department of Civil Engineering
Indian Institute of Technology Roorkee
Roorkee 247 667
bhupifce@iitr.ernet.in
• What is quality?
Integrity
Work should be executed according to drawings and
specifications
Solid Fluid
However, fluids such as blood, foods, cosmetics and concrete shown Non-Newtonian behaviour i.e.,
flow only initiates above some level of stress called as the ‘yield stress’. The best example of yield
stress is the force required to remove tooth-paste from a tube!
Examples of viscosities and shear rates
Non-ideal Bingham plastic (cement-based materials)
Yield stress
Newtonian (Water)
Rheological models
Plastic viscosity
Shear stress
Yield stress
Shear rate
Yield stress
Plastic viscosity
100
25
13
2mmThick
300
2mm Thick
200 Dia
-5.09
σ = 5.09 mm. Let the allowable tolerance in the measured initial slump value be σ.
This implies that the acceptable range of the measured slump values is 75 – 5.09 =
69.91 or say 70 mm and 75 + 5.09 = 80.09 or say 80 mm.
Hence, measured slump values less than 70 mm or more than 80 mm should invite
corrective action!
5.
09
75
5.
09
Slump values lower than the Slump values higher than the
minimum acceptable value maximum acceptable value
4σ
About 95% of the measured
slumps will lie within the
acceptable limits.
2σ
Properties of the bell-shaped curve: Approximate distribution of area under normal frequency
distribution curve
Monitoring concrete quality in the hardened state
For first 50 m3 For second 50 m3
If early age strength testing has to be done, then an additional 18 cubes have to be
collected for 7-day strength testing.
Sublots of length 0.3 km each Casting done in one day for example
0.3 km
1 2 3 4 5 6 7 8 9 10
• From which sublots will the four randomly selected samples collected?
• Consider any starting point. Pick up any four numbers whose last two digits are between
1 and 10. Repetitions are not allowed in the selected numbers!
Generate a table of random numbers
X
X
X : Last three digits are greater than 300 √
√
X
√
√
• Consider any starting point. Pick up any four numbers whose last two digits
are between 1 and 10. Repetitions are not allowed in the selected numbers!
Starting point
Sample no. 2
Sample no. 3
Sample no. 4
No repetitions
Sample no. 1
Sublot time interval = (9 h per lot) x (60 min per hour) / (5 sublots per lot) = 108 min per sublot
What is the exact time at which the sample should be collected within each sublot?
Sublot 1 - 7:30 a.m. + 51 min = 8:21 a.m. (Collect the first sample at this time)
Sublot 2 – 9:18 a.m. + 75 min = 10:33 a.m. (Collect the second sample at this time)
Sublot 3 – 11:06 a.m. + 66 min = 12:12 p.m. (Collect the third sample at this time)
Sublot 4 – 12:54 p.m. + 86 min = 2:20 p.m. (Collect the fourth sample at this time)
Sublot 5 – 2:42 p.m. + 7 min = 2:49 p.m. (Collect the fifth sample at this time)
Example: Random sampling by material weight
The total tonnage of cement required for a construction project is 15000 bags. The cement will be
delivered at site in lots of 3000 bags each. According to the contract documents, each lot shall be
divided into 5 equal sublots and one cement bag has to be collected for testing from each of the
sublots. Determine the sampling sequence.
600 bags 600 bags 600 bags 600 bags 600 bags
A random sampling procedure has to be followed. It is required to find which bag number has to be
collected from each of the sublots.
For randomised sample collection, choose any five random numbers from the random number table.
Choose five random numbers from Column Y of the Table starting with S. No. 17. These five random
numbers are: 0.1818, 0.8908, 0.9181, 0.8362 and 0.2364.
To identify the bag number to be sampled, multiply the number of bags in each sublot with the above
random numbers
In sublot 1, the bag number to be sampled is 0.1818 x 600 = 109 th
If a cumulative log of the 3000 bags to be delivered at site in one lot is available, then the bag
P
a
1000 m
ve
m
Sublot 1 Sublot 2 Sublot 3 Sublot 4 Sublot 5 e
nt
w
id
th
=
One lot of total length = 5000 m 4
m
1000 m
P
a
Randomly selected sampling point for
ve x depth measurement
m
e
nt
w y
id Coordinate system for the random sampling point
th
=
Coordinates of the sampling points
Sublot No. Longitudinal random number Sublot length Sampling point location Sampling point location
(from the X column of the (m) from start of sublot from start of lot
random number table) (m) (m)
Local X coordinate Global X coordinate
1 0.5651 1000 565.1 565.1
Sublot No. Transverse random number Sublot width Sampling point location Sampling point location
(from the Y column of the (m) from start of sublot from start of lot
random number table) (m) (m)
Local Y coordinate Global Y coordinate
1 0.1545 4 0.618 0.618
µP
P
Cracks extend throughout the specimen height
De
hc
sce
an
br
nd
g
ing
in
nd
br
ce
an
As
ch
Decreasing platen restraint
increasing brittleness
Voi
Bond crack d
Aggregate particle
Longitudinal strain
Failure criteria
for concrete in
uniaxial comp.
IS 456:2000
To exercise
control on
‘within-test’
variation’
Individual test result shall be > this strength
Mean of a group of 4 test results shall be > this strength
Mix design is carried out for this strength
Concrete grade
SAMPLE NUMBER
Overall probability becomes 75% pass
and 25% fail
1 2 3
4 5 6
How to apply the acceptability criteria if the number of samples is less than 4?
Amendment No. 4 to IS 456:2000
No. of samples required
to be collected = 4+1 = 5
Batch-to-batch variation
Additional cubes
fck - 4 = 46 MPa
for 7-day strength Mean = 56.58 MPa
test (0.66 x 28-day ➢50 + 4 = 54 MPa
strength = 33 MPa) ➢Hence, OK.
Results of 15
cubes
• Although 56.58 MPa is greater than 54 MPa, this result
gives us no idea about the degree of quality control at site.
• For example, at one site, the mean strength is 54.2 MPa (>
54 MPa) and at the other site the mean strength is 57.2
MPa (> 54 MPa). As per IS 456:2000, both the results are
satisfactory but these results tell NOTHING about the
degree of quality control at the two sites
The IS 456:2000 recommendations do not make any
attempt to analyse the variability of measured strengths.
Except for control on within-test variations by way of the
+/- 15% criteria no evaluation of the batch-to-batch
variation of concrete strengths is specified.
Variability of measured strengths
The test results are said to follow a normal distribution if they are more or less
equally spaced about the mean value !
Average strength
ơ ơ
15.9% of 15.9% of
results are results are
less than more than
f-ơ f-ơ
2.3% of 2.3% of
results are results are
less than more than
f - 2ơ f - 2ơ
σ
σ
Areas under various segments of the normal distribution curve for deviations of
1.64 and 2 standard deviations from the mean
Areas under various segments of the normal distribution curve as a function of
standard deviation
Normal distribution curves for three different cases with the same mean but
different variability. For which case is the quality control the best?
Quantitative Assessment of Variability
Determination of standard deviation
Example: At a construction site, concreting has been completed in three stages and the results of
standard 28 days cube tests are given in Table below. Determine the standard deviation for the grade
of concrete:
STAGE 1 STAGE 2 STAGE 3
Sample No. Cube strength, Sample No Cube strength Sample No Cube strength
(MPa) (MPa) (MPa)
Sample no. Cube strength (MPa) Sample No Cube strength(MPa)
1 2
3 4
5 6
1 2 3 4 5 6
͌
<
Analysis of trends in measured concrete
strengths – How to detect intervention points?
Individual test results and their average (of two test results)
High degree of noise and scatter associated with the average results
Average of 1 and 2
A plot of the ‘Moving Average Strength’ is a useful tool for
identifying trends in concrete strength and particularly for
identifying when intervention is required!
Moving average of three results
The first data point in this plot is obtained only after 3 test results
are available
Control Zone
3
1
2
5
1 3
The first data point in this plot is obtained only after 5 test results
are available
Moving average of ten results
The first data point in this plot is obtained only after 10 test results
are available
7
8
2
6
1 3 5
9
4 10
Detection of intervention points using CUSUM
CUMulative SUM control chart or CUSUM chart can be used to help detect
relatively small changes or shifts in average concrete strength for example,
quickly.
(37+34.7+32.8)/3 = 34.8
Average of 19 test
results
(34.7+32.8+37.8)/3 = 35.1
No obvious indication of a Trend detection is improved by using
change in data is seen in this the moving average chart because
plot scatter and noise in the data is reduced