IMPROVED PREDICTION OF PRE-CAMBER OF

POST-TENSIONED PRESTRESSED I-BEAM

(Date received: 22.11.2007)

Wahid Omar1, Tan Pui Lai2, Lee Poh Huat3 and Roslina Omar4
Faculty of Civil Engineering, Universiti Teknologi Malaysia, 81310, UTM Skudai, Johor
Email: [email protected], [email protected], [email protected], [email protected]

abstract
An improved method of calculating pre-camber prediction in prestressed concrete beam is presented. It has been found that
the current method of estimation that is widely practiced by local engineers which is based on BS 8110 is over simplified and
inaccurate. Accurate prediction of the upward deflection or pre-camber due to prestressing of prestressed concrete beams is
always difficult to attain. Inaccurate estimation of pre-camber usually leads to complications in construction and may incur
additional cost. This paper presents an improved pre-camber prediction of prestressed I-beam that adopts the equivalent weight
method with the inclusion of concrete creep factor into the deflection equation. While creep is often ignored or underestimated, it
is being considered in the calculation as it was identified to significantly increase pre-camber, even at an early age. Laboratory
tests were conducted to obtain compressive strength, elastic modulus, creep and shrinkage of concrete. These properties were
used in the evaluation of pre-camber. The accuracy of the prediction was compared against the actual pre-camber measured on
four post-tensioned ‘I’ beam with overall height of 1.98m and 36m in length for a duration of 15 days after the application of
prestressing force. Based on the comparison, the improved method seems to provide better results compared to the commonly
adopted simplified method. The suitability of the concrete properties, especially creep coefficient recommended by various
foreign standard codes - BS 8110, EC 2, ACI and AS 3600 for application on local concrete is also examined. The method
presented is readily available for local engineers to adopt.

Keywords: Creep Coefficient, Pre-Camber, Post-Tensioned Beam, Standard Codes

1. INTRODUCTION also occur when the vertical movement of each segment is not
In a prestressed concrete beam, pre-camber or upwards properly compensated. Considering the various damaging effects
deflection occurs due to the sustained eccentric compression force caused by pre-cambering, an accurate prediction value should not
that induces negative moment in the section. A proper design of be taken lightly.
prestressing force can make this initial pre-camber to act as an The common method of design estimation of pre-camber
offset to the downward deflection due to self weight and live currently practiced by the local engineers is the equivalent weight
load, reducing or eliminating the deflection and cracking of the method and the required concrete properties are taken directly
prestressed member considerably. Ideally, this feature offers a from design codes such as BS 8110. A site study carried out
means of controlling deflection and cracking, making it possible indicated that this method was inaccurate. The results showed that
to design longer span members. However, one common problem the actual pre-camber measured on site was 48.6% higher than the
usually faced by most local designers is the difficulty to accurately predicted values and various reasons were identified as the cause
predicting the pre-camber development. The actual upwards of the vast differences [2]. The two main causes identified were
deflection often deviates significantly from the estimated value, the calculation method and the concrete properties such as elastic
causing difficulty in the construction process, primarily in the modulus, used in the equation which was normally based on the
construction of prestressed bridges [1]. values recommended in standard codes and did not represent the
Under the circumstances when the pre-camber is under or over actual properties of local concrete.
predicted, various detrimental effects may take place. A typical Having recognised the significant difference between the
example of the effect of an inaccurate prediction of pre-camber prediction and actual deformation, an improved calculation of pre-
of prestressed bridge beam is uneven road profile and producing camber prediction of prestressed concrete beam is discussed. The
uncomfortable or even dangerous driving condition. An accurate suitability of adopting concrete properties from standard code’s
estimation is also crucial to achieve a proper finished level of the recommendation for local concrete is also examined. The accuracy
bridge. Inaccuracy may results in the application of additional of the currently practiced design prediction and the proposed
thickness of bituminous wearing course on the bridge deck in improved calculation is verified by comparing with the actual pre-
order to achieve the designed finish level. This inevitably results camber of post-tensioned ‘I’ beams measured on site. Beside the
in additional dead load, compromising the safety of structure, site measurement of pre-camber, laboratory tests were conducted
incurs additional construction cost. In most cases, the contractors to obtain the actual mechanical properties of concrete such as
have to bear the additional cost. Besides that, misalignment at creep, shrinkage, elastic modulus and compressive strength which
the jointing of segmentally constructed prestressed bridges may are required in order to predict the pre-camber more accurately.

32 Journal - The Institution of Engineers, Malaysia (Vol. 69, No.1, March 2008)
 improved prediction of pre-camber of post-tensioned prestressed i-beam

2. THE DESIGN CALCULATION OF PRE-

CAMBER IN PRESTRESSED CONCRETE BEAM
Pre-camber resulting from prestressing force can be calculated
either on the basis of curvatures, moment-area method, or based
on the equivalent weight method [3]. Deflections due to the dead
and live loads are calculated as for any other flexural members,
and to obtain the total deflection, the deflection due to prestressing
force and all the deflections that are calculated separately are
superimposed together. The common prediction method adopted
by the local design engineers is the equivalent weight method in
which the elastic modulus are usually taken directly from BS 8110
based on concrete strength. Only short term prestress losses are
taken into account in this calculation.
According to the equivalent weight method, the upward Figure 1(a): Schematic view of the experimental setup
deflection, δpi at the mid-span of a uniformly loaded simply
supported beam is obtained using Equation (1), when the tendon
profile is parabolic. The sagging of beam, δo is calculated based
on Equation (1) with Pe being substituted by moment due to the
dead load as given in Equation (2). The resultant short-term or
instantaneous deflection is obtained by using Equation (3) in
which the hogging produced by the prestressing force is deducted
by the sagging due to the self-weight of the beam. Usually only
the pre-camber at the mid span is obtained. Assuming that the
deflection profile is a perfect parabolic curvature represented by
equation y = Ax2 + Bx + C, the deflection at other locations can
be easily obtained.
5L2
δpi = (Pe) (1)
48 Ec Ic

5wL4
δo = (2)
384 Ec Ic

δshort-term = − δpi + δo
(3)

In Equations (1) and (2), the term L represents the beam span,
Ec and Ic refers to elastic modulus of concrete and moment of
inertia of section, respectively. P is the prestressing force after Figure 1(b): Cross section of I-Beam
initial losses, e is the eccentricity of tendon at the mid-span and w
is the self weight of the beam. applied at the age of 28 days. Five points on the top of each beam
Generally in practice, this short term deflection is an with equal distance of 8.95 m were chosen as the leveling reference
important information for contractors to ensure a smooth points, as shown in Figure 1(a). Figure 1(b) shows the cross-section
construction process. However this crude calculation method of the beam. The maximum pre-camber was determined from the
adopted only provides an approximate estimation and only difference between point No. 3 to the average of point No. 1 and
considers instantaneous deflections. Accuracy of this prediction No. 5. The leveling points were measured at an interval of every
is compromised due to the over simplified calculation method and two days using a theodolite and a reference datum was established
based on Equations (1) and (2), it is observed that the reference to counter check for vertical movement or settlement of the beams
to the concrete property is only on the elastic modulus. The usual during the period of monitoring. Details of site monitoring work
practice adopted is that the elastic modulus is taken directly from are described in [2].
the BS 8110 based on the concrete strength. The deviation of the
recommended elastic modulus obtained from foreign codes based 3.1 Comparison of Actual Pre-camber
on concrete strength, from the actual local concrete properties, to Prediction
further intensify the digression. Based on the measurement, it shows that the upward deflection
of the four beams continue to increase after day one until day 15 as
3. SITE MEASUREMENT OF PRE-CAMBER OF shown in Table 1. Immediately after prestressing, the mid-span pre-
PRESTRESSED CONCRETE camber of the four beams ranges between 37mm to 57mm. 15 days
The monitoring of pre-camber development was conducted on after prestressing, pre-camber of the beams increased to an average
four post-tensioned prestressed concrete I-beams with an overall of 63mm with a maximum of 75mm. It is observed from the results
height of 1.98m and 35.7m span over a duration of 15 days. The that substantial increase of the upward deflection occurred during
beams were cast with Grade 50 concrete and prestressing force was the first three days after prestressing, and subsequently, it increases

Journal - The Institution of Engineers, Malaysia (Vol. 69, No.1, March 2008) 33
 wahid omar, et al

Table 1: Comparison of Pre-camber Measured on Site and

Predicted Value [2]
Beam Pre-camber Measured Predicted
Day on Site (mm) Value Difference*
Beam Beam Beam Beam Average (mm) (%)
1 2 3 4
1** 45 57 45 37 46.0 42.4 8.50
3 55 71 56 47 57.3 42.4 35.00
6 59 73 57 49 59.3 42.4 40.33
9 58 74 57 51 60.0 42.4 41.50
12 60 75 59 55 62.3 42.4 46.80
15 62 75 59 56 63.0 42.4 48.58
* Between predicted value and average pre-camber of 4 beams measured on site Figure 3: Pre-camber of beams measured immediately after
** Immediately after prestressing prestressing [2]

Figure 2: Pre-camber of beams at different days after prestressing [2] Figure 4: Pre-camber of beams measured 15 days after prestressing [2]

steadily within a small margin. The pre-camber of the four beams is when the pre-camber has almost stabilised. A vast difference
at different time after prestressing is plotted in Figure 2. The steep is observed when the calculated 42.4mm pre-camber is compared
increment at the initial stage is due to the effect of elastic strain to the actual deflection at 15 days, when the pre-camber is more
of concrete when prestressing force is first applied. Thereafter, the constant as plotted in Figure 4. On average, the actual pre-camber
rate of increment is lower under the influence of time-dependent is higher than the estimated pre-camber in the range between
deformation of creep and shrinkage. The standard deviation of all 8.50% to 48.58% and the values are given in Table 1. Through this
the pre-camber measurement obtained from the four beams is within site measurement, the approximate calculation method currently
the range of 8.3mm to 10.0mm, with the coefficient of variation is being practiced is proven to be inaccurate and under- predicts the
between 13.3% and 17.9%. Statistically the data collected on site actual pre-camber.
can be considered as relatively consistent and acceptable as far as Referring to Figures 3 and 4, it is observed that the pre-camber
concrete testing is concern. for the four beams are different although they are designed to be the
The ultimate pre-camber of 42.4mm, obtained from the simplified same. The difference in deformation may be due to the influence
calculation method given by Equations (1) to (3) discussed earlier of various intrinsic and extrinsic factors. Therefore properties and
is also shown in Table 1. The calculation is based on the site data performance of concrete play an important role in prestressed
as given in Appendix A. This ultimate calculation refers to the final concrete members. Accurate estimation of pre-camber cannot
pre-camber at transfer before service load is applied. However the be obtained without a rational account of prestress losses and an
time frame for the structure to stabilise to that stage is ambiguous. accurate prediction of the concrete properties [4].
Therefore for comparison purpose, the predicted value of 42.4mm is
set constant from day 1 to day 15 without any demarcation, as shown 4. IMPROVED PRE-CAMBER ESTIMATION
in Table 1. A better assessment of the accuracy of the prediction is According to Nawy [1], the best estimate of pre-camber should
to compare it to actual pre-camber at day 15, when the actual pre- be based on accumulated experience, and a correct choice of elastic
camber has almost stabilised. The difference between the predicted modulus of concrete. The accuracy of pre-camber is very much
pre-camber and the initial pre-camber of the four beams at different dependent on the prediction of concrete properties such as elastic
times after prestressing is also plotted in Figure 2. modulus and prestress losses, especially due to creep and shrinkage.
The pre-camber profile of the four beams immediately The difficulty in estimating the total prestress losses affects the task
after prestressing and the calculated design value is shown in to precisely estimate the magnitude of expected pre-camber.
Figure 3. Even though the predicted profile is within the range Therefore better accuracy can be achieved when creep
of the measured pre-camber, a better assessment of the accuracy factor is considered in the initial deflection prediction and creep

34 Journal - The Institution of Engineers, Malaysia (Vol. 69, No.1, March 2008)
 improved prediction of pre-camber of post-tensioned prestressed i-beam

is considered in prestress losses [5]. Creep strain affects the strength, elastic modulus, creep and shrinkage. The compressive
deflection in two opposite ways. While it produces loss of prestress strength was obtained from tests carried out on 150mm cubes at
force, hence reducing the pre-camber, creep strains in the concrete 7 days and 28 days. Test on elastic modulus, creep and shrinkage
usually increase negative curvatures and hence, increase the pre- were conducted on 100mm diameter x 300mm cylinders at the
camber. Generally the second effect predominates and pre-camber age of 28 days. The creep and shrinkage testing were carried out
increases with time, in spite of the reduction of prestress loss [3]. according to ASTM C512-87, under the ambient of 27 ± 4oC and
When the creep coefficient and prestress losses are considered, RH of 70 ± 10% to simulate the tropical climate condition. The
the upward deflection is given as Equation (4), as proposed by elastic modulus test was conducted in accordance to ASTM C469-
Nilson [3], where δpe is the midspan deflection due to prestressing 87a. Full details of the laboratory testing are reported in [6].
after considering prestress losses and φ is the creep coefficient. The results of concrete properties obtained from laboratory
testing and standard codes are tabulated in Table 2 for comparison.
δpi + δpe
δupwards = − δpi + (δpi − δpe) − .φ (4) The recommended values of modulus of elasticity and creep
2 coefficient obtained from each code and their respective clauses
Both δpi and δpe in Equation (4) above are calculated based on are also indicated in Table 2. The creep strain, ε in Table 2 is
Equation (1) for prestress beam with parabolic tendon profile. The calculated based on Equation (7) below. σ is the applied stress due
first term in Equation (4) is the initial negative curvature and the to prestressing force applied on the beam, as given in Appendix A.
second term is the reduction in that initial curvature because of The comparison reveals that the magnitude obtained from
the loss of prestress. The third term is the increase in negative laboratory tests are within the range provided by various design
curve because of concrete creep. In Equation (4), the important codes. It should be noted that elastic modulus and creep coefficient
approximation is made that creep occurs under a constant prestress provided by BS 8110 shows a significant difference from the tested
force, equal to the average of the initial and final values, as values. This is one of the reasons identified for the significant
expressed in the third term. difference between estimated pre-camber based on BS 8110 and
The deflection due to self weight is also modified by creep and the site measurement. The modulus of elasticity given by BS
may be obtained by applying the creep coefficient to the instantaneous 8110 and ACI 209 are both comparatively low compared to the
value. Thus the downward deflection is obtained through Equation actual values. However it is worth noting that the Ec of 30 kN/mm2
(5) with δo being calculated based on Equation (2). represents the mean recommended value and actually BS 8110
allows Ec to be taken in the range of 28 kN/mm2 to 36 kN/mm2
δdownwards = δo (1 + φ ) (5) for concrete Grade 50. Therefore in the prediction of pre-camber,
the value of 34 kN/mm2 is adopted by the contractor as indicated
The total member deflection is obtained based on principle of in Appendix A.
superposition after consideration of losses and creep, effective φ
prestress and self weight is given by equation (6). ε= x σ (7)
Ec
δpi + δpe
δtotal = − δpe − φ + δo (1 + φ) (6) As appear in the Table 2, BS 8110 does not recommend
2 equivalent cylinder strength for the corresponding cube strength.
Equation (6) is formulated based on principle of superposition As for ACI 209 and AS 3600, both codes are based on concrete
from Equations (4) and (5). cylinder strength, therefore no equivalent cube strengths are given.
These are reflected in Table 2. It should be noted that the correlation
4.1 Concrete Properties of G50 cube strength to G40 cylinder strength is recommended by
Besides a better calculation formula, accurate concrete EC2. The accuracy of cube-cylinder strength conversion especially
properties prediction is important to contribute towards a better for Malaysian concrete has not been fully established and further
pre-camber prediction of prestress beams. The formulae described investigation should be initiated.
above require the properties of concrete such as strength, elastic It is interesting to observe the closeness between the creep
modulus and creep coefficient to be taken into account. Therefore, coefficient values provided by EC2 and the test values, which
laboratory testing were conducted on cube and cylinder specimens indicates that EC2 may be a suitable replacement to BS 8110 in the
prepared using the same batch of concrete that has been used near future for Malaysia.
to cast the prestressed beams to obtain the actual compressive
Table 2: Concrete properties results from experiment and standard codes 5. EVALUATION OF IMPROVED
PRE-CAMBER ESTIMATION
Concrete BS 8110 EC 2 ACI 209 AS 3600
Using the improved pre-camber
Properties Experiment [7] [8] [9] [10]
calculation method as given by Equation
fcu cube (N/mm2) 59 50 50 − − (6), and with concrete properties taken from
fcu cylinder (N/mm2) 42 − 40 40 40 Appendix A, the magnitude of predicted pre-
camber 15 days after prestressing is 69.31
Elastic Modulus, Ec 35.1 30 35 29.9 32.7 mm. This result is 38.8% higher than the value
(kN/mm2) (ref. Clauses) (Pt. 2, Cl. 7.2) (Cl. 3.1.3) (Cl. 2.2.2) (Cl. 6.1.8) obtained from Equation (3). Comparison of
Creep Coefficient at 0.725 0.375 0.723 0.567 0.954 the new predicted value and the pre-camber
15 days, φ (ref. Clauses) (Pt. 2, Cl. 7.3) (Cl. 3.1.4) (Cl. 2.4) (Cl. 6.1.8) measured on site is presented in Figure 5. With
Creep Strain, ε, at 342.6 250.0 332.7 302.8 482.4 the inclusion of the effect of creep, the new
15 days (µ strain) predicted values are closer to the average site

Journal - The Institution of Engineers, Malaysia (Vol. 69, No.1, March 2008) 35
 wahid omar, et al

Table 3 Experimental creep coefficient used for prediction

as given in Figure 6 reveals that the prediction is more accurate
Days Experimental Creep than the initial predicted values that ignoring the effect of creep.
Coefficient
By taking creep coefficient and concrete properties from EC2,
0 0 the pre-camber is within 1.73% lower and 11.1% higher than the
1 0.261 actual deformation. ACI 209 on the other hand overestimates the
3 0.406 pre-camber up to 14% higher than the actual magnitude. Therefore
6 0.455 it can be concluded that the concrete properties recommended by
9 0.533 EC 2 are better to be applied to our local concrete than the ACI
12 0.599 recommendation.
15 0.725
6. CONCLUSIONS
Based on this study, it can be concluded that:-
1. The pre-camber of the 36m I-beam measured on site is
between 37 mm to 57 mm immediately after prestressing and
it increases to between 56mm to 75mm after 15 days due to
time-dependent deformation.
2. The results of this study reveals that the actual beam pre-cambers
measured are 8.5% to 48.6% higher than the design estimation
based on BS 8110. This shows that the current simplified pre-
camber prediction based on BS 8110 is inaccurate and may
cause complications during construction.
3. Creep factor has to be taken into account in the pre-camber
estimation for a more accurate prediction. The proposed design
calculation with consideration of creep factor is found to be
simple for application and provides a closer prediction to the
Figure 5: Comparison of average site measurement to the prediction actual site measurement.
values 4. With the limited creep data for local concrete, the creep
coefficient catered for RH of 80% derived from foreign
standard codes is sufficient to achieve a good pre-camber
prediction. The EC 2 has proven to provide the best result for
local concrete properties under the condition when laboratory
testing is not feasible.
5. More laboratory test data is still required in order to develop
a comprehensive and reliable set of creep coefficient for local
tropical concrete. n

references

[1] Nawy, E. G. (2003). ‘Prestressed Concrete: A Fundamental

Approach, Forth Edition’. New Jersey: Prentice Hall.

Figure 6: Estimated precamber using different values of creep [2] Lee, P. H. (2005). ‘Case study to determine the pre-camber of
coefficient
post-tensioned I beam.’. Universiti Teknologi Malaysia: Master
measurement. This drastic improvement compared to the initial of Engineering Thesis.
prediction that stays at 42.4mm reveals the important effect of this
time-dependent deformation. This new prediction is within the [3] Nilson, A. H. (1978). ‘Design of Prestressed Concrete’.
range of 5.76% lower and 6.56% higher than the actual pre-camber New York: John Wiley & Sons.
of the beams.
The creep coefficient value, φ which is used in equation (6) is [4] Tadros, M. K., Ghali, A. and Dilger, W. H. (1975). ‘Time-
obtained from laboratory creep testing and the values for 15 days dependent prestress loss and deflection in prestressed concrete
are given in Table 3. members’. PCI Journal, 20 (3), 86 – 98.
In the occasion that the laboratory testing for creep is not
feasible to be conducted, the improved calculation method can still [5] Saiidi, M., Hutchens. E. and Gardella, D. ‘Bridge prestress
be applied to obtain an acceptable prediction. This is made possible losses in dry climate’. Journal of Bridge Engineering, ASCE,.
with the reference to other standard codes for creep coefficient 3 (3), 111 -116.
values. The suitability of foreign codes prediction for creep
coefficient is presented in Figure 6. The creep coefficient from EC [6] Omar, W., Tan, P. L. and Omar, R. (2005). ‘Pre-camber in Post-
2 and ACI 209 that caters for surrounding relative humidity of 80% Tensioned I-beam: An Improvised Prediction for Local Concrete’.
is used to replace the experimental creep coefficient. The result, Seminar Kebangsaan Kejuruteraan Awam. Johor Bahru.

36 Journal - The Institution of Engineers, Malaysia (Vol. 69, No.1, March 2008)
 improved prediction of pre-camber of post-tensioned prestressed i-beam

[7] British Standard Institution (1985). ‘British Standard 8110: [9] ACI Committee 209. ‘Prediction of creep, shrinkage and
Structural Use of Concrete.’ BSI, London. temperature effects in concrete structures’. SP-277, ACI, Detroit,
1971.
[8] BS EN 1992-1-1:2004 –Design of Concrete Structures, British
Standard Institutions, UK, December 2004. [10] Australian Standard (2001). ‘Concrete Structures’. Sydney:
AS 3600 – 2001.

APPENDIX A
PROFILEs
PRE-CAMBER CALCULATION (Site data) engr. assoc. prof. Dr wahid bin omar
Engr. Assoc. Prof. Dr Wahid bin Omar is an Associate
Section Properties Professor at the Faculty of Civil Engineering, UTM.
He graduated with B.Sc (Civil) from University of
Concrete cross section area Ac = 628000 mm2
Strathclyde, Scotland (1986), obtained his MSc in Bridge
Centroid of beam from bottom yb = 862 mm Engineering from University of Surrey (1989) and PhD
Centroid of beam from top yt = 1120 mm from University of Birmingham (1998). Currently he is
Area of tendon Aps = 98.77 mm2/strand working on time-dependent deformation of Malaysian
Tendon elastic modulus E s = 190 kN/mm2 concrete with funding from CIDB. His other research at
present is to study the possibility of using low cost steel
strap to improve the ductility of high-strength concrete.
Materials Properties
Concrete elastic modulus, Ec = 3.4E+07 kN/m2
Length of beam, L = 35.7 m tan pui lai
Beam self weight, W = 15.072 kN/m Tan Pui Lai conducted her doctorate research on
time-dependent deformation of concrete in the tropical
Moment of Innertia, I = 0.275 m4 climate in Universiti Teknologi Malaysia. She is
currently a consulting engineer in Parsons Brinckerhoff
Prestressed Forces Pte. Ltd. while she completes her doctorate part-time.
Total Initial Prestress, Pi = 7590 kN
Total Prestress after initial losses, P = 7255.4 kN
Eccentricity (mid-span) e = 0.742 m lee poh huat
Mr. Lee Poh Huat is a managing director of a construction
company Zibina Jaya Sdn Bhd. He obtained his M.Eng.
Pre-camber calculation (Civil-Structure) from Universiti Teknologi Malaysia.
The article published is part of his work in completing
5L2
Hogging due to prestress di = (Pe) the Master’s project during his study in UTM.
48EcIc
= 76.87 mm

roslina binti omar
She was born in Terengganu, Malaysia in 1981. She
5wL4
Sagging due to S/W do = received her secondary education from SMKA Sheikh

384EcIc Abdul Malek, Terengganu. She obtained her B.Eng.
= 34.48 mm (Civil) from Universiti Teknologi Malaysia in 2003.
Currently, she is a tutor at Universiti Malaysia Pahang and
working on her M.Eng. (Civil-Structure) dissertation at
Beam deflection at transfer dshort-term = di - do Universiti Teknologi Malaysia. Her research is related to
the prediction of time-dependent deformation of normal
= 42.4 mm strength concrete in tropical climate.

Journal - The Institution of Engineers, Malaysia (Vol. 69, No.1, March 2008) 37