Вы находитесь на странице: 1из 23

Determination of Consolidation Properties

and Selection of Computational Methods to


Calculate Settlements of Mine Tailings

Prepared for:
Tailings and Mine Waste 2011
Vancouver, Canada
9 November 2011

D. Geier, Golder Associates Inc.


G. Gjerapic, Golder Associates Inc.
K. Morrison, AMEC Earth & Environmental
Introduction

Lab Testing Requirements


Small Strain vs. Large Strain
1D vs. 3D
Sources of Error
Why/When Important?
Why:
Influences storage capacity
Influences pore pressures (and therefore strength)
Settlements during closure
High vs. low consolidation rates and selection of deposition method (i.e.,
conventional slurry, thickened tailings or paste)
Factors influencing rate of consolidation:
Rate of rise
Geometry (e.g., TSF layouts with narrow vs. wide base)
Tailings gradation (e.g., high clay content)
Solids content at deposition (segregation)
Climate (low evaporation, freezing temperatures, desiccation potential)
Liner systems / foundation drainage
Surface Strength Soft Tailings
Surface Strength Soft Tailings
Methods of Analysis
Small-strain consolidation (i.e., traditional Terzaghi)
Compressibility and rate of consolidation modeled using constants
d v 1 de 1 Cc
k = const. mv = = =
d v ' (1 + e) d v ' ln(10) (1 + e) v '
1 k u 2u
e = Cc log cv = => = cv 2
0 mv w t z
Large-strain (Gibson,1967):
Compressibility and rate of consolidation function of void ratio

k = C eD d k (e) e k (e) d ' e e


(Gs 1) + + =0
de 1 + e z z w (1 + e) de z t
e = A ' B + M
Lab Data
Representative samples are critical
Preferred: samples from existing TSF (e.g., beach and pool)
Next best: samples from pilot plant / TSF discharge
Not recommended: use of typical values

Traditional: Oedometer test


void ratio vs. effective stress (log), coefficient of consolidation (cv)

Better: SICTA or Slurry Consolidation


compressibility and permeability over a target stress range
Example Problem
Problem: TSF filling faster than expected. Why?
Tailings properties (design vs. operating):

Project description:
Geomembrane lined TSF (valley construction)
Tropical environment
Variation of Tailings Density
and Beach Development
Lower density, more segregation
Design density (i.e. 46%) ~ non-segregating
Operation density (i.e. 35%) highly segregating

45% slurry
density
35.5% slurry
density
Variation of Tailings Density
and Beach Development
Lower density, more segregation
Design density (i.e. 46%) ~ non-segregating
Operation density (i.e. 35%) highly segregating
Segregated Tailings (sampled from
beginning and end of flume)

Feed
Implications of Current Management
Reduced tailings
slurry density: 35.5% solids (end of beach)
Higher void ratios
46% solids
Higher
compressibility 35.5% solids (upper beach)
Implications of Current Management
Reduced tailings
slurry density: 35.5% solids (end of beach)
Higher void ratios
Higher
compressibility 46% solids
Lower permeability
35.5% solids
(upper beach)
Bottom line:
--Fine grained slimes make up
large percentage of tailings
--Lower tailings density
than planned, slower
consolidating
Reduced storage capacity
1D Large Strain Model
Use 1 column
Discretize TSF
Calculate capacity by multiplying tailings production rate by filling
time
Calculate average density by dividing tailings production by TSF
volume
Compressible Boundaries Error
1D method provides upper bound on tailings density/capacity
1D assumes TSF foundation side-slopes deform at the same rate as tailings

(Gjerapic et al, 2008)


Compressible Boundaries Error
1D method provides upper bound on tailings density/capacity
1D assumes TSF foundation side-slopes deform at the same rate as tailings

Can lead to large errors


95% in this case!
3D Large Strain Model
Use multiple columns (4 in this example)
Enforces incompressible boundary at base of each column

With multiple columns, must adjust filling times to account for


settlement in adjacent columns

(Gjerapic et al, 2008)


3D Large Strain - Correction

(Gjerapic et al, 2008)


Conservation of Mass Error
Due to convergence error in modeling software
Magnitude can be estimated by calculating average density two
ways:
Method 1: (Filling Time)x(Filling Rate)/(TSF Volume)
Method 2: Integrate void ratio profile

Average Dry Average Dry


Average
Density Density
TSF Shape Rate of Rise % Error
Method 1 Method 2
(m/yr)
(tonnes/m3) (tonnes/m3)

Flat 0.3 1.44 1.34 7.6


Cone 2.7 1.63 1.23 27.7

Method 2 generally more accurate and conservative


Results of Tailings Beach Surveys
Surveys performed 3, 7, and 12 months after TSF startup
Tailings production provided by mill operators
Calculated densities from 3D large strain model using 2 types
of tailings
Calculated
Cumulative TSF Average In-Situ
Average Dry
Filling Time Dry Density % Difference
Density
(months) (tonnes/m3)
(tonnes/m3)
5 1.22 1.08 -11.9
7 1.21 1.06 -12.5
12 1.17 1.03 -12.1

Calculated values consistently ~12% conservative


Model could be calibrated by adjusting assumed division
between un-segregated and segregated tailings
Examples of Modeling Output
Beach Tailings

Tailings Slimes
Examples of Modeling Output
8 Years After Deposition 95% Consol.
Conclusions
Use representative materials for lab testing
Assign tests to mimic predicted operational conditions
1D large strain methods appropriate for trade-off studies or
preliminary engineering
3D methods for feasibility or advanced stages
Estimate conservation of mass errors
If possible, calibrate model with measured in-situ data

Вам также может понравиться