Particle Packing Models and Their Use
in the Design of Ultra High
Performance Concretes
C. Kennan Crane
Ph.D. Candidate
Overview
z
z
z
z
z
z
z
Definition and Motivation
History of Particle Packing in Concrete
Research
Packing Models
Current Usages in Concrete Design
Use of Particle Packing in UHPC Design
Experimental Comparisons
Conclusion
Theory Behind Particle Packing
(after Stovall, 1986)
Motivation for Particle Packing in
UHPC
z
z
z
z
Decrease permeability by decreasing pore
size (Roy, 1995)
Further reduce W/C
Densify pre-hydrated mixture (Fu, 2003)
Densify hydration product (Bonneau, 2000)
Does it work?
4
History of Particle Packing
z
First used in concrete research in 1892 by R.
Feret to optimize aggregate gradation
(Wang, 1997)
Initial objective was to achieve most dense
packing possible in order to reduce cement
usage (Mehta, 2006)
Governing Assumption of All Packing
Models
MATH = FUN
(Crane, 2007)
Traditional Model Types
Particle Packing
Discrete models:
-use idealized sets of
specifically sized particles
in creating packing models
Continuous models:
-use continuous gradations
of particle sizes
(Kumar, 2003)
Discrete
Models
Binary
Furnas
Powers
Aim and Goff
Ternary
Multimodal
Toufar, Klose,
and Born
de Larrard
(Kumar, 2003)
Continuous Models
Multimodal
Fuller Thomson
Andreassen
Rosin-Rammler
(Kumar, 2003)
Furnas Model (Kumar, 2003)
z
A course and a fine phase are each
characterized by
diameter, d
volume fraction, y
packing density,
Two cases can arise
Fine Grain Dominant
Course Grain Dominant
10
Furnas Model (cont.)
z
z
Model is only valid for fine particles much
smaller than course particles
As the size of the fines approaches the size
of the course, two interaction effects occur
Wall effect: increased voids near course particles
or other boundaries
Loosening effect: increased voids when fine
particles disrupt optimum course particle packing
11
Loosening effect
Wall effect
(de Larrard, 2002)
12
de Larrard Models
z
Linear Packing Density Model (LPDM) (Stovall,
1986)
Solid Suspension Model (SSM) (de Larrard, 1994)
basic multimodal model
updated version of LPDM
includes a virtual packing factor that accounts for the
difference between ideal and random packing of particles
Compressible Packing Model (CPM)
includes compression effort
13
Fuller Thomson packing model
(Kumar, 2003)
z
Originally proposed in 1907
n
d
CPFT = 100
D
z
z
z
z
CPFT = the cumulative (volume) percent finer
than
d = the particle size
D = the maximum particle size
n = .5
14
Fuller Thomson packing model
(Young, 1998)
15
3-D Computer Models
z
z
z
z
Model begins with random placement of particles
from largest to smallest.
This is followed by a compaction that consists of a
constriction of the container (Stroeven, 1999).
The spheres interact with one another to settle into a
more preferable configuration.
This type of model was later adapted to prevent
edge effects (Fu, 2003).
16
3-D Computer Models
17
Optimizing with Particle Packing
Models
z
Self Consolidating Concrete (Brouwers,
2005)
Optimize the aggregate for maximum flowability
Prevent segregation by optimizing paste
Ultra High Performance Concrete (de
Larrard, 1994)
Optimize the aggregate and paste for maximum
density
Close particle packing helps permeability and
strength
18
SCC Pour
19
(Kim, 2007)
UHPC Flow Table Test
(Garas, 2005)
20
10
Particle Packing Effect on Strength in
UHPC
z
De Larrard showed
increased strength (Kg)
in UHPC with
decreasing Maximum
Paste Thickness (MPT)
Very little data available
currently
(de Larrard, 1995)
21
Particle Packing Effect on Durability in
UHPC
z
Roy, et. al. studied
particle packing affect
on durability of HPC.
Showed reduction in
average poor size
decreases the
permeability of the
concrete, making it
more durable.
Not directly relatable
to UHPC
(Roy, 1995)
22
11
Experimental Verification (Gallias,
2000)
z
z
z
z
Binary mixtures were made with cement and varying
percentages of 11 fine admixtures.
Water was added to achieve a standard consistency.
Water demand was measured in each of the
mixtures.
According to the particle packing models, the water
demand should decrease as small particles are
added to fill the voids between the larger particles.
23
Binary Mixtures of Admixtures
50m quartz + 3m chalk
50m quartz + 5m calcite
5m calcite + 3m chalk
24
12
Binary Mixtures of Cement and
Admixtures
25
Possible Explanation of Discrepancies
z
Interaction of cement and admixture particles (Gallias,
2000)
Clumping that changes effective particle size prevents
optimum packing
Repulsion between particles that prevents optimum packing
Shear rate dependencies could also explain lack of full
diffusion (Hodne, 2006)
Shear thinning observed in concrete could be attributed to
these clumps requiring more energy to disperse
26
13
Conclusion
z
Particle Packing Models can begin to give an
idea of how concrete constituents will interact
on multiple scales.
More research is needed, particularly in
particle interaction in order to improve these
models.
More experimental data needs to be
obtained to verify model results.
27
Questions?
14
