Documente Academic
Documente Profesional
Documente Cultură
David Goodyear, PE
Consulting Engineer
Olympia, Washington
(formerly Senior Engineer with
Arvid Grant and Associates)
108
does not have a high or low bias — we
simply want to have the greatest chance
of meeting the predicted values. Re- Synopsis
garding stress redistribution due to free Computations for creep and shrink
cantilever construction, we want to have age affect a variety of details in con-
confidence that we provide section ca- crete bridge structures. From the size
pacity (for stress design) in excess of the of expansion joints and bearings to
future demands on the section — a more the amount of prestress loss and long
traditional design objective. term deflection, creep and shrinkage
Clearly, no matter how precise the in concrete can either govern or
calculation method, the probability of greatly influence final design details
achieving both objectives with one "av- and construction of our modern
erage" creep coefficient is very low in- bridges.
deed. Even in the case of camber com- In current practice, the treatment of
putations where mean values are appro- creep and shrinkage is handled differ-
priate, there are generally well-defined ently than other loads. White we strive
consequences for "missing" the target to design our bridges for the maximum
camber. The practical options for demands of live loads, dead loads,
geometry correction can best be evalu- temperature and other parameters,
ated by inspecting a range of creep co- we have consistently chosen to look at
efficients, rather than by a refinement of average demand due to creep and
analytical procedures. shrinkage.
This paper presents the viewpoint
that the large degree of variability in
Material Behavior both concrete properties and method
Branson ° lists the following param- of design justifies a change from our
eters as affecting creep or shrinkage deterministic approach to design for
strains in concrete: creep and shrinkage to one that ac-
—Member size commodates the variability in con-
— Water-cement ratio crete properties. Because it is a more
— Mix proportions complex time dependent strain than
— Aggregate type shrinkage, the focus of this paper is an
— Length of curing creep.
— Curing temperature The current state of the art in analy-
— Curing humidity sis of creep effects is reviewed, with
— Environmental temperature ACI and CEB-FIP type analyses re-
— Time of initial loading lated to the real world difficulties in
— Duration of load modern bridge design.
— Number of load cycles
— Unloading period
— Stress distribution
— Stress magnitude
— Stress rate — the (pseudo) elastic modulus, the
Some of these factors are more vari- creep coefficient, and the loading his-
able than others, but certainly few of tory. Each plays a major role in both the
these factors can be taken as determi- ACI and the CEB-FIP methods for de-
nistic - especially in the design stage of termining creep effects.
a project. Material research presented by Nil-
To look at the variability and uncer- son' and Branson et al. 8 is typical of the
tainty in time dependent deformations data available on concrete modulus and
we will look at three major parameters creep coefficient. These particular data
Test ff E x 10''
Number psi psi E/ ^
Mean =48067.68
Standard deviation = 2676.23
yield coefficients of variation in the state of the art material testing program
range of 0.15 for modulus (expressed as during the design phase indicated that
E/) and 0.3 for creep coefficient. In the modulus of the laboratory mix was
comparing the data with both AC! and approximately 6,500,000 psi (44850
CEB-FIP procedures, Bazant g• " finds MPa), The construction contract called
that both ACI and CEB-FIP yield coef- for field testing of concrete and the re-
ficients of variation for creep between sults of that testing are indicated in
test data and predicted values of about Table 1. The most noteworthy point is
0.3, which is in agreement with the pre- that, while the variation across the sam-
viously mentioned data. ple was fairly low, the mean value for
While project specific testing can modulus was approximately 20 percent
certainly help to narrow the range of below that determined for design using
these variables, such testing at the de- project specific testing.
sign stage will not eliminate the char- Creep coefficients were also deter-
acteristic variability in concrete prop- mined for the East Huntington project.
erties. The East Huntington Bridge Of necessity, any such testing must be of
project is a good example of this point. A short duration, and prediction methods
MIS]
must he used to extrapolate the creep Code Procedures
data to obtain total creep coefficients.
Figs. 1 through 4 show results from this The ACI and GEB-FIP methods are
gob specific testing. Creep results were both empirical. Rather than being based
obtained for loading ages of 14, 28, 90 on development of a theoretical material
and 365 days. Tests were run for a load- model, they are expressions of numeri-
ing duration of one year. Plotted along cal correlations with test data. While the
with the results are both the ACI 209 format and standard correction factors
and CEB-FIP Code projections for these for the two methods differ, the primary
cases. difference is that the CEB-k 1P provi-
For designers considering creep be- sions address creep recovery (rebound
havior of concrete, these figures illus- of the concrete after unloading) sepa-
trate several important characteristics: rately from creep development. This
First, the scatter among the tests is difference leads to a two component
very large. The authors are aware of creep model for the CEB-FIP method,
other major construction projects in with one component representing the
which the high and low creep coeffi- recoverable portion of creep and an-
cients determined during construction other representing the irrecoverable
varied by as much as 100 percent. component.
Second, the CEB-FIP method gener- In the development of creep tinder
ally indicates higher creep than does the load, the similarities between the two
ACI method. However, all ACI projec- methods are more significant than the
tions shown are based on the "average" differences. Both methods subscribe to
ultimate creep coefficient of 2.35, the principle of superposition, i.e., the
pointing to a major shortcoming (or mis- thesis that the total creep effect from
understanding) in the AC! guidelines - multiple loads is a linear superposition
there is no such thing as an "average" of the individual effects from each Ioad.
creep coefficient. A more definitive Therefore, creep development for both
guideline is needed in ACI 209 for esti- methods depends not on total load, but
mating creep coefficients for design. on load history,
Fig. 5 shows the difference between For a single component creep model
the ACI 209 and CEB-FIP methods for like the ACI method, this means that
the 14-day loading age case with the ul- (with correction for loading age, etc.,)
timate creep coefficients normalized to the total creep is the sum of the creep for
the same value. The graph clearly shows each load, based on each load duration.
what the authors have found to be a For a two component creep model like
significant difference between the two the CEB-FIP method, this means that
codes, and that is the curvature (or rate) the total creep is the sum of the recover-
of the creep curve — ACI creep burns able and the irrecoverable creep for
out faster than CEB-FIP. This differ- each load based on each load duration,
ence can have a significant effect on the and the total creep recovery is the sum
analytical results for certain construc- of the recoveries for each load removed,
tion sequences that involve changes to according to each load history.
the statical system. Fig. 6 shows a comparison of creep
A third observation is that the trend of deformation computed by both ACI and
the loading age correction factor in the CEB-FIP methods for a free cantilever
ACI method for late loading ages (one under a concentrated load. Prior to un-
year in this case) is to cause an apparent loading, the two methods agree quite
increase in creep, and a reversal of posi- well. After unloading, divergence de-
tion with CEB-FIP provisions in terms velops due to the limited recoverable
of magnitude of creep. creep developed in the CEB-FIP
--'-----------
__-i --
20
U 1/ - -
0-..r
0 100 200 300 400
TIME AFTER LOADING (DAYS)
CEB-FIP PREDICTION
2.0 / ~ ^_
r ,^
U //
a
1.0
UW / /
rJ^^
,'' ACI PREDIC110N
0 1 r I 1
0 100 200 300 400
TI%IE AFTER LOADMNG (DAYS)
112
Z? /
aa ^^ C£B-FIP PREDICTION
aF
W
1
r ^
_
_r-
KQ f ^'^~ ^` r
ACI PREDICTIDN
3.0
v CEB-FIP PREDICTION
U /
model. The same initial elastic modulus nition of load as loading or unloading is
and total creep coefficient was chosen equally obscure.
for each method in this example. Most modern analysis codes that deal
Clearly, a variation in creep coefficient with creep utilize the so-called "target
of 25 to 30 percent would result in wide creep" method," - 15 in which creep strain
overlap of the results using either is described with a Dirichlet series, and
method. the creep history is carried in a state
variahle,A E,J , as described below:
114
P
f
df
V)
w
2
Z CEB-FIP
v2
z rr ---
U ^.
ACI
100 200 300 400 500 600 700 800 900 1000
11ME (DAYS)
Fig. 6. Comparison of cantilever deflections by ACI and CEB-FIP methods.
The key to this method is that the the state variables that are incremented
loading history does not have to be car- with each load application retain the
ried along in the analysis because it is effects of each load for the remainder of
already stored in the state variable. The the analysis. While in both the ACI and
state variable reflects the accumulation CEB-FIP methods creep development
of strain increments throughout the ceases with the removal of a load, in the
loading history and decays according to target creep method the creep continues
a numerical prescription to the total due to the inclusion of the initial load in
creep due to all stress increments ap- the decaying state variable.
plied on an element. In addition, for the CEB-FIP method
The curve fitting coefficients can be the amount of creep recovery for loads of
derived from the ACT method for a short duration depends on the length of
single component creep method, from time the Ioad has been acting. Since
the CEB-FIP provisions for a two com- load history is not carried in the target
ponent (two state variables) creep creep analysis, this information is not
method, or from test results. The latter is available. Artificial corrections based on
preferable once mix designs have been the sign of incremental creep strain in
established. the target creep method to accommodate
Because load history is not carried in creep recovery make intuitive sense, but
the analysis, unloading is treated as they are not an application of the empir-
loading with opposite sign. This creates ical base of the CEB-FIP method, for
complications for both target creep they imply that creep recovery is inde-
methods (single component and two pendent of die initial load.
component), The curve fitting coeffi- The practical result of the target creep
cients provide for a fairly representative method is that for an ACI type target
model of the ACI and CEB-FIP creep creep model subject to unloading, the
curves for applied loading. However, target is overestimated. But since the
z
0
F NEW LOADING+UNLOADING CURVE
(BY TARGET CREEP CALCULATION)
0
w
0
a
w
w
U
0 LOADING+UNLOADING CREEP CURVE (BY CODES)
TIME OF UNLOADING
TIME
116
BRG. PIER PIER BRG.
15.29
PLAN
ELEVATION
6 d 17 11 a R 6 5 2
East Huntington data that resembled as twice the increase in creep coeffi-
the CEB-FIP type curve (Fig. 14). While cient when using the steeper ACI creep
the increase in bottom flange tension for curve.
the ACI type curve in going from C, = The actual numbers presented here
2.35 to 3.35 is only 78 psi (0.54 MPa), the serve only for illustration. The impor-
increase when using the flatter creep tant point is that, despite all the elabo-
curve from the East Huntington data rate computation methods available
was 172 psi (1.19 MPa). Thus, this flat- today, the final stress state in a concrete
ter curve, which leaves more creep bridge structure depends on the time
strain until after main span closure, re- dependent response of highly variable
sults in almost as much stress change concrete material.
20,000 FOOT-KIPS
1000 PSI
118
4.36 ACI CREEP COEFFICIENT
`l:^rN4lf
1000 PSI
1000 PSI
120
REFERENCES
1. ACI Committee 209, "Prediction of tember-October 1985, pp. 665-675.
Creep, Shrinkage and Temperature Ef- 10. Bazant, Z. P., Chern, JC., "Bayesian
fects in Concrete Structures," Special Statistical Prediction of Concrete Creep
Publication SP-76, Designing for Creep and Shrinkage," ACJ Journal, Proceed-
and Shrinkage in Concrete Structures, ings V. 81, No. 4, July-August 1984, pp.
American Concrete Institute, Detroit, 319-330.
Michigan, pp. 193-300. 11. Khalil, M. S., "Time Dependent Non-
2. CEB-F1P Model Code for Concrete linear Analysis of Prestressed Concrete
Structures, Comit6 Euro-International Cable Stayed Girders and Other Con-
du B6ton (CEB), 1978. Available from crete Structures," University of Calgary,
Federation Internationale de la Pr6con- Alberta, Canada, 1979, 246 pp.
trainte, 11 Upper Belgrave Street, Lon- 12. van Zyl, S. F., "Analysis of Segmentally
don SW 1X SBH, England. Erected Prestressed Concrete Box
3. Tadros, M_ K., Ghali, A., and Dilger, Girder Bridges," Report No. SESM
W. H., "Time Dependent Analysis of 78-02, University of California at Berke-
Composite Frames," Journal of the ley, January 1978, 265 pp.
Structural Division, ASCE, V. 103, No. 13. Kahir, A. F., "Nonlinear Analysis of Re-
ST4, April 1977, pp. 871-884• inforced Concrete Panels, Slabs and
4. Dilger, W. H., "Creep Analysis of Pre- Shells for Time Dependent Effects," Re-
stressed Concrete Structures Using port No. SESM 76-06, University of Cal-
Creep-Transformed Section Properties," ifornia at Berkeley, December 1976, 219
PCI JOURNAL, V. 27, No. 1, January- pp.
February 1982, pp. 98-118. 14. Bazant, Z. P., and Wu, S. T., "Dirichlet
5. Tadros, M. K., Chali, A., and Dilger, Series Creep Function for Aging Con-
W. H., "Long-Term Stresses and Defor- crete," journal of the Engineering Me-
mation of Segmental Bridges," PCI chanics Division, ASCE, V. 99, No, EM2,
JOURNAL, V. 24, No. 4, July-August April 1973, pp. 367-387.
1979, pp. 66-87. 15. Ketchum, M. A., "Redistribution o'
6. Branson, D. E., Deformations of Con- Stresses in Segmentally Erected Pre-
crete Structures, McGraw Hill, New stressed Concrete Bridges," Report No.
York, N.Y., 1977, 546 pp. SESM 86-07, University of California at
7. Ngab, A_ S., Nilson, A. H., and Slate, Berkeley, May 1986, 185 pp.
F. 0., "Shrinkage and Creep of High 16. Elbadry, M.; Chali, A.; "Thermal
Strength Concrete," AC! Journal, Pro- Stresses and Cracking of Concrete
ceedings V. 78, No. 4, July-August 1981, Bridges," ACI journal, Proceedings V.
pp. 255-261. 83, No. 6, November-December 1986,
8. Meyers, B. L., Branson, D. E., and pp. 1001-1009.
Schumann, C. G., "Prediction of Creep 17_ Precast Segmental Box Girder Bridge
and Shrinkage Behavior for Design from Manual, Post-Tensioning Institute,
Short Term Tests," PCI JOURNAL, V. Phoenix, Arizona, 1978, 116 pp.
17, No. 3, May-June 1972, pp. 29-45. 18. Podolny, W., and Muller, J. M., Con-
9. Bazant, Z. P., Chern, JC., "Log Double struction and Design of Prestressed
Power Law for Concrete Creep," ACI Concrete Segmental Bridges, John Wiley
journal, Proceedings V. 82, No. 5, Sep- & Sons, New York, N.Y., 1982, 561 pp.