Transcript Workshop on Specifications and Producer QC

Workshop on Specifications
and Producer QC – Ken Day
I shall start by spending a few minutes describing
some provisions of the Australian Code before
inviting your participation.
I will then make provocative statements about
the effects of prescription and performance
specifications, plant and project control,
achievable variability, quality of testing, the
analysis of data and financial benefits
I enjoy arguments, PLEASE OBLIGE!
Workshop on Specifications
and Producer QC
Australian Code Requirements
AS1379 / DR05253
www.standards.com.au
Specification of Concrete
• Concrete shall be specified as either: Normal
class or Special class and by strength grade ‘or
other readily verifiable parameter’
• Standard grades
20, 25, 32, 40, 50,/ 65, 80, 100 MPa
(20=2,900, 50=7,250 / 100 =14,500psi)
• 32MPa (4,640psi) is the minimum strength
allowed to be specified for external reinf. concrete
Also max agg size, slump, method of placement, air
entrainment are to be specified.
Other Standard Requirements
(wide basic requirements for all concrete)
•
•
•
•
Density 2100-2800kg/m (130-175lb/cu ft)
Acid soluble chloride and sulphate limits
Shrinkage (max 1000)
7 day strength 50% of grade strength (up
to N50)
• Cement complying with AS3972 alone or
plus ‘one or more supplementary
cementitious materials’
Special Class Concrete
• Still preferably uses strength grade can be
compressive, flexural or indirect tensile
• One of 3 exposure classifications - which
imposes limits on aggregate durability and
class and type of cementitious material
• >50MPa (7,250psi) has to be special class
• Can specify other requirements
‘in consultation with the supplier’
Special Class Concrete
• It is certainly allowed to specify tighter limits on
shrinkage, low permeability, the use of fly ash,
silica fume, or ggbfs etc.
• This will usually restrict the range of producers
prepared to supply to those having
arrangements with particular material suppliers
and suitable storage arrangements (extra bin?)
• Other special class items could include selfcompacting concrete, lightweight, very high early
strengths, colour control, wear resistance, heat
generation, shotcrete, underwater etc.
Assessment for Compliance
• Tests at least 1 per 100cu m (133cu yd).
• Assess at selected ‘production interval’
from 2 weeks to 3months (1 month usual)
• One grade to be selected as ‘control
grade’ and to have at least 10 results per
production interval (most have many
more)
• Variability of production to be assessed on
basis of this grade
Compliance Requirements
• The ‘Characteristic Strength’ is that
exceeded by 95% of results so:
= Mean –1.65SD (90% so -1.28SD in USA)
However it is recognised that there is an error in
assessment so k in mean-kSD is varied
according to number of samples in control grade
from 3.2 for 4 samples to 1.5 for 10samples and
1.25 for 15 or more samples (1.25 gives all error
margin to producer, I do not agree with this)
Compliance Requirements
• For ‘associated grades’ the SD is scaled from
that of the control grade using relative factors:
<20(2,900psi)-0.9,
20-1.0,
25-1.1,
32-1.2,
40-1.3,
50 (7,250psi) -1.4
Note that these are far from assuming the
same coefficient of variation applies,
eg 40 @ 1.3 not 2x20 @1.0
Dissemination of Information
• A long list of information is to be prepared
at the end of each ‘production interval’ and
‘kept readily available for inspection’
• Customers can require relevant data to be
submitted to them within 15 days
• The supplier shall notify the relevant
customers, within 2 working days, if a
particular quantity of concrete is likely to
be below the specified strength
Project Assessment
• At least 1 sample per 50cu m
• Moving average of 3 to exceed f’c
Not much different to US, less likely to find
any problem than the plant control
The Purpose of Testing Concrete
• In 1958 I wrote:
‘The only rational objective for any but
100% testing is not to discover and
reject faulty products but to ascertain
the minimum quality level of the
production’
-Some ideas take a while to sink in!
The Purpose of Testing Concrete
• More recently I have added a second
objective of quality control:
‘To detect at the earliest possible
moment any change in the quality
of concrete being supplied’
The factors involved in such a detection are:
the frequency of testing,
the basic variability of the concrete,
the analysis system in use.
Plant v Project Control
1. It is far more efficient to ensure that no defective
concrete is produced at a plant than to ensure
that no defective concrete is delivered to a project
because:
2. More data is available at less cost per cubic yard
3. Problems can be detected and identified earlier
4. Producers can react to a limited amount of early
age test data but project control usually requires a
significant number of 28 day results to demand
action on marginally defective concrete.
5.
A limited number of standard mixes can be accurately
maintained with data also on shrinkage, durability, fresh
properties (slump, pumpability, bleeding) etc
DISCUSSION
1.
2.
3.
4.
Prescription specifications provide no incentive for
producers to know or care anything about
designing or controlling concrete.
As a consequence such specifications have to provide
a very large safety margin to cover high variability and
inefficient design and material selection.
Since there is no incentive for the producer to employ
competent staff, purchase good materials, and have
good production facilities, it is necessary for all such
matters to be specified in detail and to employ
supervision to ensure compliance.
Therefore prescription specification concrete is
inevitably more expensive than producer-controlled
performance concrete.
Variability of Concrete
(COMMENTS?)
1. Under good quality control, the SD of concrete
strength should not exceed 450psi and can be
as low as 300psi (achieved on concrete of
14,000psi mean strength on Petronas Towers)
2. Under the UK QSRMC system a figure of
600psi is regarded as normal
3. Without formal QC, 800 to 1000psi would not
be surprising (what is yours?)
4. Even under the relaxed US criterion of 10%
defective, each extra 100psi of SD requires a
mean strength increase of 128psi and so at
least 20lb/cu yd additional cement
Testing of Concrete
1. In a paper ‘Bad Concrete or Bad Testing’ to
ACI San Diego in 1989 I showed that
individual tests can be inaccurate to the
extent of over 1,000psi but that an analysed
pattern of results was very reliable.
2. A single test result is not an invariably
accurate assessment of the quality of
concrete in a single truck
3. ALSO we should show as much concern for
those trucks we did not test as for those we did
Testing of Concrete
1. Best criterion of testing quality is average pair
difference of 28day results.
2. 75psi is best attainable, 150psi just OK,
200psi is POOR
3. Average strength of a pair is depressed by at
least half the pair difference (more likely twice
this)
4. So sub-standard testing costs you money even
without considering rejections and penalties
(Prof Juran’s ‘Gold in the Mine’)
Testing of Concrete
1. Who does the best job of testing?
2. In Australia a lab is required to be NATA
registered. An independent lab must achieve
registration but derived no benefit from
additional quality since the lab was chosen by
the main contractor in the 1970s
3. Since higher testing quality meant a higher
mean and lower variability, supplier’s labs had
more incentive to achieve a higher standard
4. So, as an independent analyst and a NATA
assessor, I found that supplier’s labs generally
achieved at least a slightly higher standard
Testing of Concrete: Duplicate
Testing
1.
2.
3.
4.
5.
When concrete producers began doing the main
testing, specifiers were initially keen to have a
proportion of check tests by independent labs
It proved very inconclusive to test different trucks
Duplicating tests by two lab teams on the same
truck was more enlightening, but expensive
Requiring producer’s sampling personnel to take a
proportion of double samples and deliver half to an
independent lab worked well
Now, after years of experience of such checks,
very few purchasers require independent checks
Conclusion on Producer Testing
1. After 20+ years of experience it is clear that it
is worthwhile for any substantial producer to
establish a high standard, officially certified,
laboratory
2. The laboratory and its test data and analysis
records must be open to inspection by
customer’s representatives at all times
3. In these circumstances, plant control with
producer testing is substantially more
economical and reliable than project
control
Physical and Financial Benefits
1. It is clear that producer operated plant
control results in better controlled,
lower variability concrete at a lower
real cost of production. So there is
scope for everyone to benefit
2. The question is how to get started –
specifiers probably happy to use plant
control once it is seen to exist but
producers have to provide it first?
Physical and Financial Benefits
3) I suggest specifiers include an alternative
clause waiving requirements such as
minimum cement content and aggregate
gradings ‘where an approved system of
plant control is in operation’. They
could initially be generous as to what is
required for approval.
4) Some US producers are already using
plant control and will immediately benefit.
Competition will ensure others follow.
Physical and Financial Benefits
1. The initial benefit will be to those producers
who first achieve low variability and so are
more competitive
2. As other producers catch up, the benefit is to
the customers as competition reduces price
3. There may be some tendency for takeovers as
some small producers are unable to compete
4. However small producers headed by
technically competent persons can compete
and may be able to react quicker than larger
producers and gain an initial advantage
(eg Don Bain at Maricopa)
Analysis of Data
1. The efficient analysis of data is of as
much significance as quality of testing
2. It is to be judged by the rapidity of
corrective action in terms of number of
results needed to reach a decision
3. Multigrade, Multivariable, Cusum
analysis is 3 to 10 times as fast as
normal Shewhart graphing
4. A semi-automatic mix correction system
is also required for rapid response to
give low variability
Analysis of Data
1. Such analysis can be undertaken using the
free programs on my website:
www.kenday.id.au
2. There are far more elaborate systems (such as
ConAd) giving greater insight into the causes
of problems and more precise mix correction
3. Such programs may also be of substantial
value in production engineering matters,
locating batching plant and constituent material
problems and also monitoring the performance
of individual testing personnel and truck drivers