Scribd
Upload
47 views11 pages

ANN Soil Fly Ash

The document describes a study that developed an artificial neural network model to predict the permeability of soil mixtures blended with fly ash. Tests were conducted to determine permeability and other properties of soil-fly ash mixtures. The neural network used 5 input variables and a 5-9-2 structure. It produced acceptable R values between 0.968-0.984 and mean square error of 0.006. The study concluded fly ash increases permeability by enlarging inter-particle voids and the neural network is effective at predicting permeability of soil-fly ash mixtures.

Uploaded by

Gerard Velicaria
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
47 views11 pages

ANN Soil Fly Ash

The document describes a study that developed an artificial neural network model to predict the permeability of soil mixtures blended with fly ash. Tests were conducted to determine permeability and other properties of soil-fly ash mixtures. The neural network used 5 input variables and a 5-9-2 structure. It produced acceptable R values between 0.968-0.984 and mean square error of 0.006. The study concluded fly ash increases permeability by enlarging inter-particle voids and the neural network is effective at predicting permeability of soil-fly ash mixtures.

Uploaded by

Gerard Velicaria
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 11

ARTIFICIAL NEURAL NETWORK

PERMEABILITY MODELING OF
SOIL BLENDED WITH FLY ASH
Authors: Dungca, J., Galupino, J. (2016)
Presented by: Gerard Jude Velicaria
INTRODUCTION
■ It is very important for engineers to understand the soil underneath because it will
affect the way the structures are designed. Without its knowledge, lives would be at
stake and it can reveal a lot of information by just analyzing the soil profile of a
certain area.
■ Geotechnical properties of the soil such as grain size distribution, Atterberg limits,
specific gravity, maximum and minimum index densities, soil classification,
permeability, shear strength and compressibility should be considered by engineers
when designing foundation, retaining walls, etc.
■ The utilization of fly ash may be a viable alternative for porous backfill material
because fly ashes generally consist of silt-sized particles and consequently possess
high permeability but tests must be done to determine the permeability of soil-fly
ash mixture since there was a lack of information on the horizontal permeability of
the said mixes.
METHODOLOGY
■ Soil mixture’ properties such as
density, Atterberg’s limit, index density,
and particle size were determined
using tests given by the ASTM
■ The permeability of the different soil
mixes were determined using the
constant head test and falling head
test method.
■ The direction of the flow was also
considered, thus horizontal and
vertical orientations of permeameter
were used.
METHODOLOGY
RESULTS AND DISCUSSION
INDEX PROPERTIES
■ The specific gravity of the soil mixtures
was reduced by the addition of fly ash
since the usual specific gravity of fly
ash is much lower compared with the
soil.

■ Fly ash is considered as silt material, it


is expected to have a plasticity index
less than 1, thus, the addition of fly
ash in the mixture has reduced the
plasticity of a soil mixture
INDEX PROPERTIES
■ The Maximum Void Ratio (emax) ranges
from 1.78 to 1.99 because the fine
contents of the fly ash contributed to
the percentage of voids. 100S has the
lowest value while 100FA has the
highest

■ Their fines content and microfabric


may have contributed to the minimum
and maximum void ratio. These
minimum and maximum void ratios
together with the target relative
density of 90% were used to
determine the void ratio to be utilized
for the permeability specimens.
PERMEABILITY
■ Silt particles have almost similar size,
forming larger inter-particle void,
contributing to a much larger inter-
particle voids, thus the permeability of
pure fly-ash ranges: vertical oriented
4.53x10-05 cm/s to 5.52x10-05 cm/s
and horizontal oriented 6.02x10-05
cm/s to 7.28x10-05 cm/s.
■ Due to the contribution of fly ash to
the inter-particle voids of the soil
mixtures, the permeability of mixture
of soil and fly-ash ranges: vertical
oriented 2.05x10-05 cm/s to
3.80x10-05 cm/s and (2) horizontal
oriented 3.04x10-05 cm/s to
5.02x10-05 cm/s.
ARTIFICIAL NEURAL NETWORK
■ Matlab was utilized for the creation of the ANN model
■ Five variables were considered: (1) The percentage of added fly ash in the soil, (2)
Specific gravity (3) Liquid Limit (4) Maximum Void Ratio, (5) Minimum Void Ratio
■ Data was divided into three groups: 70% training, 15% validating, 15% testing.
■ Tan-sigmoid transformation function utilized for hidden layer, linear transformation
function for output layer
■ Feed-forward backpropagation technique was employed to generate the best model
ARTIFICIAL NEURAL NETWORK

■ After numerous trials, it is found out


that ANN structure 5-9-2 (5 input
layers, 9 hidden layers, 2 output
layers) is the best model to estimate
the permeability of the soil mixtures
■ The model was able to give acceptable
R values of 0.98414, 0.97570, and
0.96843 for validation, training and
testing,
■ Mean square error of 0.0061175 at
epoch 25
CONCLUSIONS
■ The addition of fly ash to soils changes the inter-particle void ratio, it increases the
permeability, thus, the microscopic characteristics of the soil mixtures may
contribute to the increase in permeability. Based on the tests, fly ash is a
combination of larger silt grains and smaller silt grains to form the micro fabric. Silt
particles have almost similar size, forming larger inter-particle void, contributing to a
much larger inter-particle voids.
■ ANN Structure 5-9-2 is the best architecture since it has the highest correlation
coefficient, R2, and the said model has a good predicting ability compared with
previous studies. To improve the results, extensive manipulation of parameters
should also be considered. Sensitivity analyses may be conducted to further validate
the model and the results.

You might also like

pFad - Phonifier reborn

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


Alternative Proxies:

Alternative Proxy

pFad Proxy

pFad v3 Proxy

pFad v4 Proxy