Modeling of Stiffness Degradation and Expansion in Cement Based Materials Subjected to External Sulfate Attack Barzin Mobasher Department of Civil and Environmental Engineering, Arizona State University
Transport Properties & Concrete Quality Workshop Arizona State University October 10-12, 2005, Tempe, AZ
You created this PDF from an application that is not licensed to print to novaPDF printer (http://www.novapdf.com)
Outline of Presentation Overview
of the model Governing mechanisms Mathematical model System Parameters Parametric studies Comparison with experimental results Summary and Conclusions
You created this PDF from an application that is not licensed to print to novaPDF printer (http://www.novapdf.com)
Overview of the Model Initial Material Parameters Cement composition w/c ratio Degree of hydration Mineral admixtures Initial sulfate content
Size & Shape of Members Specimen geometry Reinforcement type Boundary conditions
Computed Material Parameters Mechanical properties Capillary porosity Diffusivity, D Reactive calcium aluminate content
Exposure & Loading Sulfate exposure conditions Time & humidity
Service Life Expansion-Time History Degradation levels
Tixier, Mobasher, ASCE J. of Materials, 2003 You created this PDF from an application that is not licensed to print to novaPDF printer (http://www.novapdf.com)
Schematics of 1-D Diffusion of Sulfates
Penetration of Sulfates, Chemical reactions of ettringite formation or gypsum crystallization, expansion induced damage causing both decalcification or microcracking of the CSH paste.
tetracalcium aluminate hydrate, C4 AH 13 calcium aluminate monosulfate hydrate C 4 A S H 12 residual tricalcium aluminate
C3 A
ettringite formation
SO4-Aggressive solution
unreacted calcium aluminate sites
SO4--
Sulfates ingress front
You created this PDF from an application that is not licensed to print to novaPDF printer (http://www.novapdf.com)
expansion
Sulfate Attack - Simplified approach
1st step: formation of gypsum Ca(OH)2 + Na2SO4.10H2O CaSO4.2H2O
2nd step: formation of ettringite. C4 AH 13 3 C S H 2 14H C6 A S 3 H 32 CH C4 A S H 12 2 C S H 2 16 H C6 A S 3 H 32 C3 A 3C S H 2 26 H C6 A S 3 H 32
You created this PDF from an application that is not licensed to print to novaPDF printer (http://www.novapdf.com)
Sulfate Attack - Simplified approach All reactions lumped in one:
CA q S C6 A S 3 H 32 with CA equivalent of C4AH13, monosulfate and residual C3A combined, and q equivalent stoichiometric coefficient obtained from the proportions and molar volume. Phase
Ettringite C3A C4AF Gypsum
Molar Volume 3 (cm /mole) 735.0 89.1 128.0 74.2
Molecular Weight (g/mole) 1249.5 269.9 477.4 172.1
You created this PDF from an application that is not licensed to print to novaPDF printer (http://www.novapdf.com)
Specific Gravity 3 (g/cm ) 1.70 3.03 3.73 2.32
Solution of diffusion Equations: 1-D diffusion with first order reaction U=U0 @ T
U 2U D kU 2 T X
U=U0 @ T
U=0 at T=0 reaction, k
Solution:
U 4 1 n X 1 sin ( k exp T (k ) U0 m 0 n(k ) L n 2m 1
n , D L
2
You created this PDF from an application that is not licensed to print to novaPDF printer (http://www.novapdf.com)
Diffusion & Reaction in Time & Space Standard Fick’s law Diffusion
f(p)= f(porosity, gel space ratio,..)
calcium aluminates
f(p)*[SO4--] Sulfate ingress ettringite front
SO4-D, [CA], p
[SO4--]
Present Model reacted calcium aluminates
calcium aluminates
Bound SO4--
f(p)*[SO4--] You created this PDF from an application that is not licensed to print to novaPDF printer (http://www.novapdf.com)
2nd Order Reaction Equation CA q S C6 A S 3 H 32
Rates of reaction :
d U SO
4
dt
- k U SO U CA 4
k U SO U CA d U CA dt q 4
with k rate constant, U molar concentration
You created this PDF from an application that is not licensed to print to novaPDF printer (http://www.novapdf.com)
Diffusion Reaction equation for 2nd Order Reaction Equation
Fick’s second Law- Diffusion as a function of time is driven by the concentration gradient. Assume no advection takes place, and no diffusion term for Calcium aluminates (immobile). Moving Boundary condition problem since sulfates are consumed, and the diffusion coefficient may be a function of location depending on the crack percolation. Let:
CA q S C6 A S 3 H 32 T time,
U U SO and C U CA
X distance,
4
U 2U D - kUC 2 T X
C T
kUC q
D diffusion coefficient
You created this PDF from an application that is not licensed to print to novaPDF printer (http://www.novapdf.com)
Concentration profiles for k=10-8 m3 /(mol.s), CA0=8.15 mol/m3 concentration/U0
1 0.9
SO4
0.8
CA
0.7 0.6 0.5
time
0.4 0.3 0.2
time
0.1 0
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
x You created this PDF from an application that is not licensed to print to novaPDF printer (http://www.novapdf.com)
0.4
0.45
0.5
Modeling of expansion: stress-strain response stress
damage 1
B
f’t
A
0
w log(sec / 2) 1 w0 f 't log(sec 0 / 2)
0
O e th
• OA: linear elastic E= E0, = 0 • AB: crack initiation E=E0(1-) • BC: crack propagation, E derived from expression:
ep
C strain Karihaloo and Fu, 1990
You created this PDF from an application that is not licensed to print to novaPDF printer (http://www.novapdf.com)
Damage Evolution Law 0 i ( ) 1 um 0
f (1 ) 1
, 0 i 1 f ( 1 ) 1 1.00
Karihaloo and Fu, 1990 a= 0.16, b= 2.3 eum= ultimate strain for uniaxial tension
1
E( ) / E0
Em
Em
b = 0.8
0.80 b = 0.6 b = 0.4
0.60
16 i (1 - m2 ) 3
Nemat Nasser and Hori, Micromechanics: overall properties of Heterogeneous Materials, 1993
0.40 0.000
You created this PDF from an application that is not licensed to print to novaPDF printer (http://www.novapdf.com)
0.002 0.004 Axial Strain, mm/mm
0.006
Simplified Approach- Fick’s second law of diffusion with a first order reaction
U is the concentration of sulfates, k is a rate constant, D is the diffusivity m2/s, X is depth and T is time, n=2m+1.
U 2U D kU 2 T X n X U 4 1 1 sin k exp T ( k ) L U0 m0 n ( k )
D=1.14xx10-11
D=2.24x10-13
You created this PDF from an application that is not licensed to print to novaPDF printer (http://www.novapdf.com)
Simplified solution algorithm
Solve for sulfate distribution at a given time with a given diffusivity and rate of reaction. (D, and k= finite amount) Solve the PDE with k=0 for the amount of sulfates that would have penetrated if there were no reactions with C3A. Difference between the two levels is the amount of reacted sulfates. The average values were found by integrating the U/U0 values over the entire depth of the specimen The average value of U/U0 must be multiplied by the initial sulfate concentration at the surface (U0= 0.35moles/Lit, input) to find the concentration of reacted sulfates. Stoichiometric and molar volume relations are used to convert reacted sulfates to reacted aluminates, and ettringite formed. Volumetric changes are related to linear expansion values.
You created this PDF from an application that is not licensed to print to novaPDF printer (http://www.novapdf.com)
Modeling of expansion: volumetric changes Reaction
P aS ettringite
P: expansive CA
Volume change in paste due to unit change in vol. of Hyd. Products 1
mv
mv ettringite VP 1 1 a VP mv P mv gypsum
d M
Volumetric strain due to Volumetric change of Paste in concrete VP VP V CAreacted V P VC
CA reacted = Ca – Cremaining
You created this PDF from an application that is not licensed to print to novaPDF printer (http://www.novapdf.com)
Modeling of expansion: Effect of porosityvolumetric change V corrected V f f
fraction of capillary porosity being filled capillary porosity max f c
w 0.39 c ,0 w 0.32 c
fc = cement content w/c = water/ cement ratio = degree of hydration
You created this PDF from an application that is not licensed to print to novaPDF printer (http://www.novapdf.com)
Cracking – Diffusivity relationship finite difference approach plane of symmetry
D
D
cracked zone: D=D1
uncracked zone: D=D2
D1 D2
D2
XS (0)=0
At T=0
L/2
X
0
XS (T)
At T>0
You created this PDF from an application that is not licensed to print to novaPDF printer (http://www.novapdf.com)
L/2
X
Adaptation of the 1D solution to a 2D problem- finite difference approach concentric layers perimeters uncracked zone Cracked zone
Average Stiffness= Weight-averaged at nodes
: point
where U is computed
X
X
You created this PDF from an application that is not licensed to print to novaPDF printer (http://www.novapdf.com)
Strain generation across a 1/4 portion of the sample exposed on both sides 0.0165
Strain, mm/mm
0.0165 0.016
0.016 0.0155
0.0155
0.015 0.015
0.0145 0.014
0.0145
0.0135 0.013
0.014
5
25
10
20 15
15
X coordinate, mm
10
20 25
5
Y coordinate, mm
You created this PDF from an application that is not licensed to print to novaPDF printer (http://www.novapdf.com)
D D1 D=D1
Effect of Concrete Diffusivity
uncracked zone: D=D2
D2 0 X
HPC
Control Concrete 3
3
D1/D2=1000 D2=1x10-13 m2/s
D1/D2=10
2
Expansion, %
Expansion, %
D1/D2=1000
D1/D2=100
1
D2=1x10-12 m2/s
D1/D2=10
2 D1/D2= 100
D1/D2=1
1
D1/D2=5
0
0
500
1000 1500 Time, days
2000
2500
0
0
200 400 Time, Days
600
As time progresses, an increased diffusivity due to cracking results in faster rates of expansion. You created this PDF from an application that is not licensed to print to novaPDF printer (http://www.novapdf.com)
Effect of Initial Sulfate Content on Expansion and Degradation 0.4
20 C3A = 8% Gypsum= 0-10% Degree of hyd. of C3A =0.5 -2
-2
SO4 = 4-10%
Modulus, GPa
Expansion, %
0.3
C3A = 8% Gypsum= 0-10% Degree of hyd. of C3A =0.5
16 SO4 = 2%
0.2
12 -2
SO4 = 0%
8
-2
0.1
SO4 = 2%
-2
4 SO4 = 4-10% -2
SO4 = 0% 0
0
100 200 Time, Days
300
0
0
You created this PDF from an application that is not licensed to print to novaPDF printer (http://www.novapdf.com)
100 200 Time, Days
300
Effect of Internal porosity on the Dissipation of Reaction Products 0.4
Linear Expansion, %
f=0.15 0.3
f=0.35 0.2
f=0.45
0.1
f=0.55 0
0
200
400
Time, Days
You created this PDF from an application that is not licensed to print to novaPDF printer (http://www.novapdf.com)
600
Validation of model Accuracy of prediction of experimental results depends on availability of values for parameters such as: • sulfate diffusivity, • rate constant, • capillary porosity, • hydrated calcium aluminates content and form. • Experimental conditions may not match model hypotheses, as for pH monitoring.
You created this PDF from an application that is not licensed to print to novaPDF printer (http://www.novapdf.com)
Experimental Program
Sulfate attack on blend of mortars under controlled environment. Length change of mortar bars, dynamic modulus, SEM. C3A content ranging from 2 % to 11.6 % The w/c ratio used was 0.485 Sand/cement ratio of 2.75. 5% solution of Sodium Sulfate Specimen size: 25 mm x 25 mm x 275 mm and 10 mm x 10 mm x 40 mm. Constant pH and variable pH environments Dimensions recorded on a weekly basis for periods of up to three years
You created this PDF from an application that is not licensed to print to novaPDF printer (http://www.novapdf.com)
Effect of C3A 1
0.25 Experiment, Initial C3A= 6% Simulation
f=0.3; -C3A=0.9 E0=28000 MPa f 't= r = 3 MPa D1/D2=12 D1 = 1x10 -13
0.15
0.1
6% C3A
0.05
0
0.8 Expansion, %
Expansion, %
0.2
0
200
400 600 Time, Days
Experiment, C3A=11.6% Simulation
800
=0.9; f=0.25 -C3A=0.9 E0=28000 MPa f 't=3 MPa r = 10 MPa D1/D2=20 D2 = 1x10 -13
0.6
0.4
0.2
1000
0
0
40
finite difference approach You created this PDF from an application that is not licensed to print to novaPDF printer (http://www.novapdf.com)
80 120 Time, Days
11.6% C3A
160
200
Simulation of Effect of C3A content from 4% to 11.6% 0.6
Expansion, %
11.6% 0.4 8% 6% 0.2
4% 0
finite difference approach
0
200
400 600 Time, Days
You created this PDF from an application that is not licensed to print to novaPDF printer (http://www.novapdf.com)
800
Effect of Specimen Size 1
Same variables for same materials, Only input Variable: specimen size
Experiment Simulation
0.6 0.4 0.2
25.4x25.4x279.4 mm
0 -0.2
finite difference approach
10x10x40mm
0.8
Expansion, %
w/c=0.485; C3Ai=0.06%; Gypsum=0.06%; -C3A=0.9; E0=28000 MPa; ft'=3 MPa; =0.9; f=0.45;
0
40
80 120 Time, Days
You created this PDF from an application that is not licensed to print to novaPDF printer (http://www.novapdf.com)
160
200
Comparison of Expansion and stiffness degradation 40
0.6
30
0.4
20
PCA-174Asodium
0.2
0
0
100 200 Time, Days
Modulus, GPa
Expansion, %
0.8
Expansion (Exp) Modulus (Exp) Expansion (simulation) Modulus (simulation)
10
0 300
You created this PDF from an application that is not licensed to print to novaPDF printer (http://www.novapdf.com)
finite difference approach
Effect of Curing Duration on Expansion 2.5 D2 = 1x10-13 D1/D2 = 38 f = 0.15
Modeled By three variables:
Expansion, %
1) Degree of Hydration, j 2) Internal porosity, f 3) Diffusivity changes due to cracking D1/D2
2
1.5
D2 = 1x10-13 D1/D2 = 25 f = 0.25
LTS-2 1 year curing
TXI-18 1 day curing
1
0.5
0
finite difference approach
0
40
80 Time, Days
You created this PDF from an application that is not licensed to print to novaPDF printer (http://www.novapdf.com)
120
160
Reduction of diffusivity due to the use of class F flyash 0.16
Expansion, %
0.12
w/c=0.44; f=0.55; C3A=8%
C100FA0 C45FA55 Simul. Control Simul. , FA%55
0.08 D = 1.14x10-11 D = 2.0 x10 -13
0.04
SO4 -- = 6% -C3A=0.9 E0=28000 MPa f 't= 3 MPa r = 10 MPa
Class F Flyash effect
0 0
100
200
300
Time, Days
finite difference approach You created this PDF from an application that is not licensed to print to novaPDF printer (http://www.novapdf.com)
Effect of W/C ratio and C3A content 0.25
0.8
C3A=12% C3A=7% Simul., C3A=12% Simul., C3A=7%
W/C Effect, C3A=8% w/c = 0.45 w/c = 0.6 Simul., w/c=0.45 Simul., w/c=0.6
0.15
D= 1 x 10-12 D1/D2 = 10 f= 0.35
0.1
D= 5 x 10-13 D1 /D2 = 5 f= 0.5
0.05 0
0
40
0.6
Expansion, %
Expansion, %
0.2
80
Time, Days
120
160
0.4
C3AEffect, w/c=0.6
0.2
0 0
100
200
300
400
Time, Days -C3A=0.9 E0=28000 MPa f 't= 3 MPa finite difference approach r = 10 MPa
You created this PDF from an application that is not licensed to print to novaPDF printer (http://www.novapdf.com)
Simulation of Effect of C3A using the simplified approach 1
Expansion, %
0.8 0.6 0.4
w/c=0.6; Exp.C3A=4.3% f=0.45; Exp.C3A=7% -C3A=0.9 Exp.C3A=8.8% E0 =28000 MPa Exp.C3A=12% f 't= 3 MPa model, k=2.2e-8 r = 10 MPa k=3.5e-8 D = 1.14x10-11 k=6e-8 k=1e-7
0.2 0 0
100
200
300
400
Time, Days You created this PDF from an application that is not licensed to print to novaPDF printer (http://www.novapdf.com)
Series solution
Simulation of Tertiary mixes with Flyash and Silica fume 1 Control 3%SF+40%FA Finite Difference, Control Finite Difference, 3%SF+40%FA Series Solution, Control Series Solution, 3%SF+40%FA
Expansion, %
0.8 0.6
Pozzolan Effect
0.4 M.D.A. Thomas, et. al. CCR, 29 (1999)
0.2 0
0
100
200
300
Time, Days You created this PDF from an application that is not licensed to print to novaPDF printer (http://www.novapdf.com)
400
Series solution
Conclusions
A Predictive model is presented based on cement chemistry, concrete physics and mechanics. important properties: C3A content, internal porosity, use of flyash, and diffusivity. input parameters compatible with experimental data. Effects of degradation, specimen size, curing duration, and material ingredients can be verified using the model. Simplified solution approach requiring a spreadsheet approach.
You created this PDF from an application that is not licensed to print to novaPDF printer (http://www.novapdf.com)
Acknowledgement Financial
Support of Salt River Project, SRP in the Program on Blended cements is greatly appreciated
You created this PDF from an application that is not licensed to print to novaPDF printer (http://www.novapdf.com)
www.PDFCool.com