See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/323342953

SHAPE CORRECTION FACTORS FOR DRYING SHRINKAGE OF SEVERAL

TYPICAL CONCRETE CROSS-SECTIONS

Cover Page · November 2016

CITATIONS READS

0 962

4 authors:

Nicky Reybrouck Roman Wan-Wendner

Ghent University Ghent University
11 PUBLICATIONS 23 CITATIONS 215 PUBLICATIONS 1,906 CITATIONS

SEE PROFILE SEE PROFILE

Robby Caspeele L. Taerwe

Ghent University Ghent University
176 PUBLICATIONS 1,115 CITATIONS 330 PUBLICATIONS 5,524 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Fastenings View project

Influence of concrete age and aggregate type on anchor load capacity View project

All content following this page was uploaded by Roman Wan-Wendner on 22 February 2018.

The user has requested enhancement of the downloaded file.

SHAPE CORRECTION FACTORS FOR DRYING SHRINKAGE OF
SEVERAL TYPICAL CONCRETE CROSS-SECTIONS

Nicky Reybrouck1 Roman Wendner2 Robby Caspeele1 and Luc Taerwe1

1
Magnel Laboratory for Concrete Research, Department of Structural Engineering, Ghent University,
Ghent, Belgium
2
Christian Doppler Laboratory, Institute of Structural Engineering (IKI), Department of Civil Engineering
and Natural Hazards, University of Natural Resources and Life Sciences, Vienna, Austria

ABSTRACT
When a concrete member is loaded for a long period of time, creep and shrinkage of concrete gradually
develop. The prediction of this time-dependent behaviour is important as it may cause serious
serviceability problems in concrete structures. The shape and size of a concrete cross-section and
environmental conditions have an important effect on the moisture flow and moisture distribution of a
concrete cross-section and affect the drying shrinkage and drying creep behaviour and thus the long-term
behaviour of the concrete member. The impact of the shape and size of the cross-section on the net
macroscopic temporal strain and curvature due to shrinkage is investigated in this contribution using
numerical simulations. A time-dependent analysis is commonly based on equations given by design codes.
The effect of moisture diffusion in these equations is included in an empirical manner and is commonly,
among others, defined based on the notional size of the element. As a result of the investigations
performed in this contribution, simple shape correction factors are proposed which can be applied to the
notional size of elements with different shapes in order to improve the accuracy of the time dependent
prediction of drying creep and shrinkage of existing code formulations. The proposed correction factor is
based on numerical simulations of different geometries (rectangular, circular, T-shaped beams, ribbed
plates, box girders, ribbed plates, hollow core slabs, TT-elements) and a wide range of element sizes.

Keywords: concrete, moisture diffusion, drying shrinkage, shape effect.

1. Introduction
Creep and shrinkage of concrete result in long-term deformations that may cause serviceability problems
and it is therefore important to predict this effect properly. The mechanisms causing drying creep and
drying shrinkage of concrete is caused by the difference in moisture content within the concrete cross-
section as function of time. The drying shrinkage of concrete members depends highly on the possible
moisture transport routes affected by for example a different shape geometry (T-beams, ribbed plate, box
girders, etc.), hollow areas (hollow core elements), obstructions (imbedded steel) and boundary conditions
(isolations and exposed surfaces). The effect of differential moisture distribution should therefore be taken
into account in order to enhance the accuracy of the prediction of strain development. International
accepted time functions for drying shrinkage (such as in Eurocode 2 EN 1992-1-1, Model B3, Model B4,
the fib Model Code 90/99 and 2010, etc.) use the notional size parameter ℎ0 = 2𝐴⁄𝑆 (A: area, S:
perimeter in contact with the atmosphere). This study is focussing on the determination of the influence of
different 2D geometrical shapes of concrete members on the drying shrinkage response and indirectly
attempts to provide suggestive corrections for the parameter ℎ0 to take into account the influence of the
shape on the drying shrinkage.
Many physical phenomena strongly affect the drying shrinkage behaviour such as the local aging degree,
temperature, pore humidity, hydration energy and creep due to internal induced stresses. Ideally, an
advanced coupled analysis is required such as a Hydro-Thermo-Chemical (HTC) model which simulates
the evolution of temperature and moisture both spatially and temporally in combination with damage,
creep and aging (Di Luzio and Cusatis, 2009). These methods are required in order to accurately simulate
the mechanical behaviour and transport mechanisms of the concrete as function of time. In such an
analysis, the thermal and relative pore humidity behaviour is temporally and spatially simulated and
combined with a coupled mechanical analysis of concrete composition using an experimentally verified
Lattice Discrete Particle Model (LDPM) approach (Cusatis et al., 2011). The global deformation can be
obtained based on the total strain rate composed out of inelastic flow, temperature rate and shrinkage rate.
Performing a full detailed analysis by taking the mentioned effects into account requires a complex model
and increases the amount of input parameters required for analysis. However as a first approach presented
here, a simplified numerical study was performed in order to have a basic understanding of the effect of
the shape on the notional size. The notional size was obtained based on the moisture transport as function
of time and its effect of drying shrinkage for different concrete shapes assuming a homogeneous concrete
material. As a simplification the effects due to thermal effects, cracking, aging and creep due to internal
stress redistributions were neglected. Also only the drying effect based on constant environmental
conditions is considered. This requires less computational time while still providing good insight in the
general behaviour. By using this simplification and methodology the time-development of drying
shrinkage 𝛽𝑑𝑠 (𝑡) is obtained directly with maximum value 1.0 at 𝑡∞ .
The time-development of drying shrinkage is proportional to the square of the notional size of the element
given by the volume-surface ratio. However there is also an additional effect of the cross-sectional shape
which can be taken into account by using a shape correction factor. In model B3 and B4 shape correction
factors 𝑘𝑠 have been specified for typical shapes, i.e. the slab, cylinder, prism, sphere and cube. The
values were based on the half-time of average pore relative humidity, based on solving the nonlinear
diffusion equation of drying of concrete with fixed environmental humidity, ℎ𝑒𝑛𝑣 = 65%. The values
proposed were 𝑘𝑠 = 1.00 for an infinite slab, 1.15 for an infinite cylinder, 1.25 for an infinite prism, 1.30
for a sphere and 1.55 for a cube (Bazant and Murphy, 1995). These values were recently updated for the
same typical shapes which were calculated with larger accuracy and for several environmental humidities
between 30% and 80%. These recently corrected values are for example for ℎ𝑒𝑛𝑣 = 80%: 1.00 for a slab,
1.19 for a cylinder, 1.24 for a square, 1.30 for a sphere and 1.41 for a cube (Dönmez and Bazant, 2016).
In this contribution the effect of different practical geometries were simulated using only the nonlinear
diffusion equation in which the effects of selfdessication, aging and creep are neglected as a first
approximation and in the viewpoint of developing a simplified calculation tool. In this contribution a first
step towards a shape correction factor applicable for more complex and frequently used construction
geometries as shown in Fig. 1 is provided.

Square Circle T-beam Ribbed plate Hollow core

Rectangle TT-element Box girder

Fig. 1. Overview of several types of cross-sections used for the numerical analyses.
2. Numerical analysis of drying shrinkage

2.1 Moisture transport

The diffusion equation governing the moisture transport using Darcy’s law is given by

𝜕ℎ
= ∇[𝐷(ℎ)∇h] (1)
𝜕𝑡

The moisture transport governing the pore relative humidity can be formulated in terms of the pore
relative humidity itself. The progress of moisture at each point of the concrete section in time is calculated
using a non-linear diffusion coefficient expressed as a function of the relative pore humidity. The
diffusivity depends on ℎ and is given by CEB-FIP (2013) as (in m²/sec)

1−𝛼 10−8
𝐷(ℎ) = D1 [𝛼 + ] 𝑤𝑖𝑡ℎ D1 = (2)
1−ℎ 𝑛 𝑓𝑐𝑚 − 8
1+[ ]
1 − ℎ𝑐

In which 𝐷(ℎ) is the diffusion coefficient in m²/s, 𝑓𝑐𝑚 is the mean compressive strenghth in MPa, 𝛼 is the
ratio of minimum and maximum diffusion coefficient, ℎ𝑐 is the relative pore humidity at 𝐷(ℎ) = 0.5𝐷1 , ℎ
is the relative pore humidity and 𝑛 is a parameter. Typical values were assumed as given by CEB-FIP
(2013): 𝛼 = 0.05, ℎ𝑐 = 0.80 and 𝑛 = 15. The graph 𝐷(ℎ)/𝐷1 as function of relative pore humidity ℎ is
given in Fig. 2. The equation can be corrected for non-isothermal conditions (Bažant and Najjar, 1972).
The correction for thermal effects was not taken into account. The decrease of diffusivity of concrete with
decreasing pore relative humidity makes the diffusion equation highly nonlinear.

1.00
0.90
0.80
0.70
0.60
D(h)/D1

0.50
0.40
0.30
0.20
0.10
0.00
0.00 0.20 0.40 0.60 0.80 1.00
Relative pore humidity h

Fig. 2. Diffusion coefficient as function of relative pore humidity.

The maximum time-step for numerical stability during analysis using the Forward-Time Central-Space
method in 2-dimensions is given by

Δ𝑥 2 Δ𝑦 2
Δ𝑡 = (3)
2𝐷𝑚𝑎𝑥 ⋅ (Δ𝑥 2 + Δ𝑦 2 )
An example of spatial moisture content in case of a T-shaped section is given in Fig. 3.

Fig. 3. Example of spatial moisture content evolution

The global drying shrinkage as function of time is determined based on the local shrinkage increments in
each finite element of the considered geometry by assuming that cross-sections remain plane after
deformation. The relation between the local shrinkage strain increments and the loss of relative pore
humidity obtained by the diffusion analysis is given as (Bažant and Yunping, 1994)

𝑑𝑓𝑠 (ℎ)
𝑑𝜀𝑠ℎ = (𝑒𝑠0 𝑔𝑠 (𝑡) ) ⋅ 𝑑ℎ (4)
𝑑ℎ

in which 𝑓𝑠 (ℎ) = 1 − ℎ is assumed. This assumption is based on the hypotheses that the drying shrinkage
strain is approximately proportional to the water loss and the isotherm of water loss is approximately
linear to the decrease in pore relative humidity (> 30% RH) (Hubler et al., 2015).
The relative humidity in the capillary pores of concrete at various points and times is computed using the
nonlinear diffusion equation for drying concrete. The equation is solved numerically with the initial
relative pore humidity at 𝑡0 assumed equal to 100% and the boundary condition equal to the ambient
1
external conditions ℎ = ℎ𝑒𝑛𝑣 . When assuming 𝑒𝑠0 = and 𝑔𝑠 (𝑡) = 1 (no aging) the time-
1−ℎ𝑒𝑛𝑣
development of shrinkage 𝛽𝑑𝑠 (𝑡 − 𝑡𝑠 ) can be obtained. The ultimate strain in each cell is proportional to
the ultimate drying shrinkage and is equal to 0 at beginning and is set equal to 1 at 𝑡∞ . Hence, values are
obtained relative to the ultimate drying shrinkage.

2.2 Numerical study

The average drying shrinkage strain is related to as the square of the effective specimen size. This is
commonly characterized by the notional size and is calculated as the volume-surface ratio as:

2𝐴
ℎ0 = (5)
𝑆

in which ℎ0 is the notional size or effective thickness or size of the member and 𝐴 and 𝑆 are the area and
the exposed surface of the concrete member respectively. A factor of 2 is used to make ℎ0 equal to the
actual thickness when an infinite flat slab is considered (Dönmez and Bazant, 2016).
Numerous simulations are performed in a wide range of dimensions for several types of shapes in order to
define a “generally applicable” shape correction factor. An overview of several types of cross-sections
used for the numerical analysis is given in Fig. 4. The dimensions were simulated with values ranging
from 200 mm to 1000 mm. The numerical simulations were performed for different mesh sizes equal to 5
mm and 10 mm and using ℎ𝑒𝑛𝑣 = 80% (outside environmental condition) or ℎ𝑒𝑛𝑣 = 60% (lab condition)
as boundary conditions. The moisture content history in each cell is stored at 100 time-points
logarithmically distributed in the period 0-10 years. The moisture content data is then used to estimate the
notional size parameter based on Model Code 2010 and Eurocode 2 formulations.
 SQUARE CIRCULAR RECTANGULAR INFINITE PLATE

∞ ∞
H
H

Z D W

Type K1
Type K2 Type K3
tc = (H-tk)/2

H tk tk tk tk tk tk tk

Fig. 4. An overview of several types of cross-sections used for numerical analysis

(areas in orange indicate ‘exposed to boundary conditions’)

3. Numerical results and shape correction factor

3.1 General observations

Comparing the simulations performed with a mesh with element size equal to 5 mm and 10 mm show only
a small difference during the first 10 days and quickly converge to the curve as calculated with a mesh
with element size equal to the 5 mm (see Fig. 5a). When comparing using 𝑓𝑠 (ℎ) = 1 − ℎ or 𝑓𝑠 (ℎ) = 1 −
ℎ3 (both equation mentioned by Bažant and Yunping (1994) during simulation it yields similar results (see
Fig. 5b).

(a) Mesh size equal to 5 mm (blue) and 10 mm (red) (b) using 𝑓𝑠 (ℎ) = 1 − ℎ (red) or 𝑓𝑠 (ℎ) = 1 − ℎ3 (blue)

Fig. 5. Effect of mesh size and 𝑓𝑠 (ℎ) function on drying shrinkage (Square cross-section, RH = 80%).
The effect of external relative humidity at the boundaries is shown in Fig. 6a where the global drying
shrinkage for square cross-sections (dimension between 200 mm and 1000 mm) simulated with ℎ𝑒𝑛𝑣 =
60% and ℎ𝑒𝑛𝑣 = 80% are shown. Further, it was already mentioned that the mean compressive strength
𝑓𝑐𝑚 affects the diffusion coefficient (see Eq. 2). Higher compressive strength causes a delayed drying
shrinkage effect as shown in Fig. 6b.
(a) Global drying shrinkage for square cross-sections (b) Effect of mean compressive strength on the global
with different sizes, ℎ𝑒𝑛𝑣 = 80% (blue) and ℎ𝑒𝑛𝑣 = drying shrinkage in case of an infinite slab with
60% (red) thickness of 300 mm and ℎ𝑒𝑛𝑣 = 80% (fcm = 28MPa
and fcm = 48MPa)
Fig. 6. Effect of relative humidity and concrete compressive strength on drying shrinkage.

3.2 Evolution of drying shrinkage according to international standards

The time development of drying shrinkage in a concrete cross section is given by widely used shrinkage
models as:
𝜀𝑐𝑑𝑠 (𝑡, 𝑡𝑠 ) (𝑡 − 𝑡𝑠 )
𝛽𝑑𝑠 (𝑡 − 𝑡𝑠 ) = =√ (MC90 & MC2010) (6)
𝜀𝑐𝑑𝑠0 (𝑓𝑐𝑚 )𝛽𝑅𝐻 (𝑅𝐻) 0.035 ⋅ (ℎ0 )2 + (𝑡 − 𝑡𝑠 )

𝜀𝑐𝑑 (𝑡) (𝑡 − 𝑡𝑠 )
𝛽𝑑𝑠 (𝑡, 𝑡𝑠 ) = = (EN 1992-1-1) (7)
𝑘ℎ 𝜀𝑐𝑑0 (𝑡 − 𝑡𝑠 ) + 0.04 ℎ01.5

𝜀𝑐𝑑 (𝑡) 𝑡 − 𝑡𝑠
= (EN 1992-2) (8)
𝐾(𝑓𝑐𝑘 )[72 exp(−0.046𝑓𝑐𝑘 ) + 75 − 𝑅𝐻]10−6 (𝑡 − 𝑡𝑠 ) + 𝛽𝑐𝑑 ℎ0 2

𝜀𝑠ℎ (𝑡) (𝑡 − 𝑡𝑠 )
= tanh (√ ) ; 𝜏𝑠ℎ = 𝑘1 ⋅ (𝑘𝑠 ⋅ ℎ0 )2 (B3 and B4) (9)
𝑘ℎ 𝑘𝑎 𝜀𝑠∞ 𝜏𝑠ℎ

where 𝑡, 𝑡0 is the current time and concrete age at exposure to drying (all times are in days), 𝜏𝑠ℎ is the
shrinkage halftime, 𝑘1 is the empirical factor depending on concrete strength, 𝑘𝑠 is a shape factor derived
by fitting numerical simulations for the nonlinear diffusion equation, 𝑓𝑐𝑚 is the mean compressive strength
and 𝑅𝐻 is the ambient relative humidity.
There are large differences between the widely used models for the time-development of drying shrinkage
𝛽𝑑𝑠 (𝑡, 𝑡𝑠 ). Model B3 and Model B4 are based on diffusion theory and have very similar time-functions
compared to Model Code 2010 but differs significantly from Eurocode 2. The unknown parameter ℎ0 is
found for every shape based on the results from numerical analysis and by means of Least-Squares curve
fitting.
3.3 Numerical results for square and circular cross-sections and infinite slabs
The calculated notional size based on the numerical simulations ℎ0,𝑠𝑖𝑚𝑢𝑙𝑎𝑡𝑖𝑜𝑛 is determined based on a
least squares approximation of the drying shrinkage curve obtained by numerical simulation using the
ℎ
described time development given by Model Code 2010 or Eurocode 2. The ratio 𝛼 = 0,𝑠𝑖𝑚𝑢𝑙𝑎𝑡𝑖𝑜𝑛
2𝐴⁄𝑆
as
function of size for circular and square geometries and infinite slabs is shown in Fig. 7 and Fig. 8. The
ratio 𝛼 is expected to be equal to one, which would indicate that the model predicts the same drying
shrinkage development as the current model suggests. The obtained ratio 𝛼 is in the range of 0.8 to 1.3 and
increases as function of size in case of Eurocode 2 while the ratio 𝛼 is in the range of 0.4 to 0.5 and
remains fairly constant as function of size in case of Model Code 2010. Model Code 2010 has a similar
drying shrinkage function compared to model B3 which is based on diffusion theory. This could explain
the fairly constant ratio 𝛼, however it appears that the ℎ0,𝑠𝑖𝑚𝑢𝑙𝑎𝑡𝑖𝑜𝑛 is always underestimated by a factor 2.
This difference is further explained in section 4. The large differences between the obtained variable
ℎ0,𝑠𝑖𝑚𝑢𝑙𝑎𝑡𝑖𝑜𝑛 for the different models can be explained by the difference in shape in the time function 𝛽𝑑𝑠
(see Eqs. 6 and 7).

Fig. 7. Shape correction factor: h0 obtained by simulation divided by (2A/S) as function of size for
circular and square geometries and infinite slabs in case of Eurocode 2.

Fig. 8. Shape correction factor: h0 obtained by simulation divided by (2A/S) as function of size for
circular and square geometries and infinite slabs in case of Model Code 2010.

The values for the notional size obtained from least square fitting to the drying shrinkage according to the
drying shrinkage time development of Model Code 2010 is shown in Eqs. 10-15 for different cross-
sectional shapes (square, circle, infinite slab, hollow tube and hollow core slabs) and in Fig. 9 in case of
infinite slabs. In case of hollow elements “K1”, “K2” and “K3” (see Fig. 4) the correlation coefficient 𝑅 2
decreases related to the amount of hollow cores and therefore more local effects such as corners or T-
corners (see Fig. 11) are of importance. The parameter t c (wall thickness) of the hollow core elements
describe the ℎ0 factor well for this type of geometries.
1000
Plate thickness

y = 2.2x
500 R² = 0.9946
[mm]

0
0 100 200 300 400 500
h0 [mm]

Fig. 9. Notional size parameter ℎ0 obtained from simulations for an infinite plate
 1
ℎ0 𝑆𝑄𝑈𝐴𝑅𝐸 = ⋅ 𝑧 (𝑅2 = 0.9983) (10)
4.0
1
ℎ0 𝐶𝐼𝑅𝐶𝐿𝐸 = ⋅ 𝐷 (𝑅2 = 0.9993) (11)
4.1
1
ℎ0 𝐼𝑁𝐹𝐼𝑁𝐼𝑇𝐸 𝑆𝐿𝐴𝐵 = ⋅ ℎ (𝑅2 = 0.9954) (12)
2.2
1
ℎ0 𝐾1 = ⋅ 𝑡𝑐 (𝑅2 = 0.9933) (13)
2.2
1
ℎ0 𝐾2 = ⋅ 𝑡𝑐 (𝑅2 = 0.9753) (14)
2.2
1
ℎ0 𝐾3 = ⋅ 𝑡𝑐 (𝑅2 = 0.9222) (15)
2.2
3.4 Numerical results for rectangular cross-sections and hollow core slabs
The values for the notional size (in mm) obtained by least square fitting of the numerical results to the
drying shrinkage equation as function of notional size according to Model Code 2010 in case of hollow
core elements and rectangular sections with dimension as shown in Fig. 4 is shown in Fig. 10.

(a) contour lines of notional size ℎ0 for rectangular (b) contour lines of notional size ℎ0 for hollow
elements as function of height and width tube “K1” elements

(c) contour lines of notional size ℎ0 for hollow (d) contour lines of notional size ℎ0 for hollow
tube “K2” elements tube “K3” elements
Fig. 10. notional size ℎ0 obtained from the numerical simulations for rectangular and hollow tube
elements.
4. Discussion
4.1 Nonlinear diffusion equation
The nonlinear diffusion equation depends on the local pore humidity. When exposed at low relative
humidities the edges dry out first and can form barriers which block moisture transport affecting the
global shrinkage behavior.
4.2 Element decomposition
The hollow tube “K1” can be considered as a big square subtracted by a smaller square. However this
geometry can also be described by its characteristic wall thickness 𝑡𝑐 . This hollow tube might be
considered as an infinite slab with corrections for the local effects near the four corners of the infinite slab
defining its shape effect. Similarly, long rectangular sections are composed out of parts which behave
similar to infinite slab and two parts at the end as shown in Fig. 11. The global drying shrinkage as
function of time 𝛽𝑑𝑠 (𝑡, 𝑡0 ) could be related to a linear combination of the drying shrinkage as function of
time of the composing sub-elements multiplied by the ratio of the area of the sub-elements to the total
concrete area.

SUB - ELEMENTS

Infinite slab Corner Infinite slab T-Corner

Edge effect Infinite slab Corner

Exposed edge

Fig. 11. Decomposition of basic shapes into infinite slabs and local effects
2𝐴
The notional size ℎ0 = is widely used in Model B3, Model B4, Eurocode 2 and Model Code 2010 to
𝑆
take into account the effect of shape and size. The notional size in this contribution was determined using
Least Squares fitting on approximate numerical simulations and taking into account the models for the
drying shrinkage development according to Model Code 2010. The obtained results are of importance for
relatively thin members such as thin plates, hollow core slabs and slender elements. In case of classical
structures composed of massive elements more complex simulations are required (including thermal
effects and creep). However an analytical formulation of ℎ0 related to a combination of the composing
sub-elements is complex and not fit for practical use. Therefore practical graphs were proposed which can
be employed in order to include the shape effect.
The diffusion time is defined by (𝑡 − 𝑡𝑠 ) in which 𝑡 is the concrete age in days and 𝑡𝑠 is the concrete age
at the beginning of drying in days. The diffusion coefficient (𝐷1 ) is related to the diffusion time (𝑡 − 𝑡𝑠 )
and is proportional to 1⁄(𝑓𝑐𝑚 − 8) (see Eq. 2). Higher compressive strength of concrete causes a delayed
drying shrinkage effect resulting in higher values for the notional size ℎ0 . Therefore the notional size
based on the simulated drying shrinkage using mean compressive strength 𝑓𝑐𝑚,0 can be transformed to the
notional size as if the simulation was performed using mean compressive strength 𝑓𝑐𝑚 . Using Eq. 6 we can
write:

𝑓𝑐𝑚,0 − 8
ℎ0,𝑓𝑐𝑚 = √ ⋅ ℎ0,𝑓𝑐𝑚;0 (10)
𝑓𝑐𝑚 − 8
 in which √(𝑓𝑐𝑚,0 − 8)⁄(𝑓𝑐𝑚 − 8) is a proposed term in this contribution in order to include the effect of
concrete strength. The simulations were performed assuming 𝑓𝑐𝑚 = 48 MPa, however the results can
easily be transformed using the proposed correction term of Eq. 10. Dönmez en Bazant (2016) assumes
𝐷1 = 30 𝑚𝑚²⁄𝑑𝑎𝑦 which corresponds to 𝑓𝑐𝑚 = 36.8 𝑀𝑃𝑎 when the non-linear diffusion coefficient of
Model Code 2010 is used.
The notional size ℎ0 obtained from numerical simulations according to Model Code 2 differs from the
expected values according to 2A/S for basic shapes (see Fig. 8: correction factor 𝛼 = 0.5). For example:
𝑍
the notional size based on numerical simulations on square elements is ℎ0 𝑆𝑄𝑈𝐴𝑅𝐸 = 4.0 where ℎ0 𝑆𝑄𝑈𝐴𝑅𝐸 =
𝑍
2𝐴⁄𝑆 = 2 is expected and similarly the notional size based on simulations on infinite slabs is
1
ℎ0 𝐼𝑁𝐹𝐼𝑁𝐼𝑇𝐸 𝑆𝐿𝐴𝐵 = 2.2 ⋅ 𝐻 where ℎ0 𝐼𝑁𝐹𝐼𝑁𝐼𝑇𝐸 𝑆𝐿𝐴𝐵 = 2𝐴⁄𝑢 = 𝐻 is expected. The difference between the
notional size obtained from numerical results compared to the notional size according to 2𝐴⁄𝑆 is a factor
0.5 for all the performed simulations where the notional size was obtained by fitting to the drying
shrinkage formulation as provided by Model Code 2010. However a good match is obtained when fitting
the notional size to the drying shrinkage formulation as provided by Eurocode 2. The unexpected
𝑓𝑐𝑚,0 −8
prediction for ℎ0 in case of Model Code 2010 cannot be explained by the ratio √ which is only 1.18
𝑓𝑐𝑚 −8
but is due to the difference in ultimate drying shrinkage where Model Code 2010 is 30% larger compared
to Eurocode 2. The notional size obtained from simulations in case of Model Code 2010 is however
predicted correctly when this effect due to higher ultimate drying shrinkage is taken into account.

5. Conclusions
The shape of the simulated drying shrinkage as function of time agrees well with the function as given by
Model Code 2010. The notional size parameter ℎ0 obtained by fitting the numerical results to the drying
shrinkage development function according to Model Code 2010 is linearly proportional to the size of the
element. This behaviour is similar to the notional size as calculated by the simplified equation ℎ0 = 2𝐴⁄𝑆.

References
BAZANT, Z. P. & MURPHY, W. P. 1995. Creep and shrinkage prediction model for analysis and design of
concrete structures - model B3. Materiaux et constructions, 28, 357-365.
BAŽANT, Z. P. & NAJJAR, L. J. 1972. Nonlinear water diffusion in nonsaturated concrete. Matériaux et
Constructions, 5, 3-20.
BAŽANT, Z. P. & YUNPING, X. I. 1994. Drying creep of concrete: constitutive model and new experiments
separating its mechanisms. Materials and Structures, 27, 3-14.
CEB-FIP 2013. fib Model Code for Concrete Structures 2010, Germany, Ernst & Sohn.
CUSATIS, G., PELESSONE, D. & MENCARELLI, A. 2011. Lattice Discrete Particle Model (LDPM) for failure
behavior of concrete. I: Theory. Cement and Concrete Composites, 33, 881-890.
DI LUZIO, G. & CUSATIS, G. 2009. Hygro-thermo-chemical modeling of high performance concrete. I:
Theory. Cement and Concrete Composites, 31, 301-308.
DÖNMEZ, A. & BAZANT, Z. P. 2016. Shape factors for concrete shrinkage and drying creep in model B4
refined by nonlinear diffusion analysis. Materials and Structures.
HUBLER, M. H., WENDNER, R. & BAŽANT, Z. P. 2015. Statistical justification of Model B4 for drying and
autogenous shrinkage of concrete and comparisons to other models. Materials and
Structures/Materiaux et Constructions, 48, 797-814.

View publication stats