Dynamic Analysis of Fundao Tailings Dam Failure
Presented by: TIMOTHY D. STARK & JIALE LIN
University of Illinois at [email protected]
Presented to:
51st Geotechnical Engineering ConferenceUniversity of Kansas
Lawrence, KansasNovember 14, 2019
T.D.Stark-Slides-©
1/58
Fundao Failure Outline
• 5 November 2015 Failure• Chronology of events• Field observations• Subsurface investigation• Effect of blasting and earthquakes• Slope stability analyses• Summary
T.D.Stark-Slides-©
2/58
2015 Fundao Tailings Dam Failure
Left photo: https://www.theguardian.com/world/2018/feb/28/brazil-dam-collapse-samarco-fundao-mining Right photo: lindsaynewlandbowker - WordPress.com
T.D.Stark-Slides-©
19 deaths
3/58
2015 Flow Failure
Photo from: bbc.com
Photos from: abc.net.auT.D.Stark-Slides-©
4/58
T.D.Stark-Slides-©
https://www.youtube.com/watch?v=zy41SxJsSbM
2015 Flow Failure Videos
https://www.youtube.com/watch?v=k0PWRiEWMTY
https://www.youtube.com/watch?v=WMHqSp7lse8
•1:00 – 1:45
•3:36 – 4:17
•5:14 – 6:10
5/58
Flow of Tailings
Fundao
Image from: https://www.samarco.com/en/rompimento-da-barragem-de-fundao/
5.5 M m3 of tailings traveled 537 km to mouth of Doce River
T.D.Stark-Slides-©
6/58
Photo from: miningreview.com
Flow of Tailings
17+ deaths and 16+ non fatal injuries
Tailings entering Atlantic Ocean
T.D.Stark-Slides-©
7/58
T.D.Stark-Slides-©
2D Cross-Sections - 02
(Morgenstern et al., 2016)
116 Piezometers & Water Levels
8/58
Piezometric Level
T.D.Stark-Slides-© (Morgenstern et al., 2016)
A-A’ 02
9/58
Piezometric Surface
T.D.Stark-Slides-© (Morgenstern et al., 2016)
10/58
T.D.Stark-Slides-©
Kα
Blanket Drains
funneling seepage flow(Morgenstern et al., 2016)
11/58
T.D.Stark-Slides-©
Kα
El. 826 m blanket drains
El. 860 m blanket drains
Blanket Drains
(Morgenstern et al., 2016)
12/58
Witness Location and Description
#1: worker standing on plateau, felt it begin to move beneath him and crack around him, detaching from the setback slope, and moving downstream.
#2, #3, and 5: slope movement having propagated “from the bottom up” on the lower benches, not from the crest down, placing the seat of movement at lower elevations.
T.D.Stark-Slides-©
02
(Morgenstern et al., 2016)
13/58
Witness Location and Description
#9: Avalanche of mud-like tailings cascaded down. The starter dam had no movement
#5: Crack open up along crest of left abutment setback and propagating in both direction
T.D.Stark-Slides-©
#5
(Morgenstern et al., 2016)
14/58
Left Abutment Overview
White dashed line represents crack locationa small bulldozer on bench at El. 875 m moving or being pushed outward
T.D.Stark-Slides-© (Morgenstern et al., 2016)
15/58
Outline
• 5 November 2015 Failure• Chronology of events• Field observations• Subsurface investigation• Effect of blasting and earthquakes• Slope stability analyses• Summary
T.D.Stark-Slides-©
16/58
Pre-Failure CPT and SPT locations
T.D.Stark-Slides-©
02
(Morgenstern et al., 2016)
17/58
Cross-Section 02– Left Abutment
CrackScarp
FUND15-SPTFUND06-CPT FUND16-CPT
T.D.Stark-Slides-©
FUND 07-CPT
SandSlope SandPlateau SandToe
Cross-Section 02– Left Abutment18/58
Sand tailings and slime particle distribution
T.D.Stark-Slides-©
Boundaries for most liquefiable soil (Ishihara et al., 1989)
19/58
Sand tailings and slime particle distribution
T.D.Stark-Slides-©
Boundaries for most liquefiable soil (Ishihara et al., 1989)
20/58
Composition of Sand tailings and slime
T.D.Stark-Slides-©
Sand tailings mainly comprise loose sand and silt, high potential to liquefy
% S
and,
Silt
, and
Cla
y21/58
Pre-Failure Blasts & Earthquakes
Table 2.1: Pre-failure mine blasts and earthquakes on 5 November 2015 (Atkinson, 2016)
T.D.Stark-Slides-©
Local TimeMoment
Magnitude Mw
Distance from
FundaoIndentification
PHGA DurationReference(g) (second)
1:01:49 PM 2.1 2.6 km mine blast 0.0001g <5 Dowding (1996)
1:06:06 PM 2.3 2.6 km mine blast 0.00004 <5 Dowding (1996)
2:12:15 PM 2.2 <2 km earthquake (foreshock) 0.06 70 Atkinson
(2016)
2:13:51 PM 2.6 <2 km earthquake (main shock) 0.08 80 Atkinson
(2016)
2:16:03 PM 1.8 <2 km earthquake (aftershock) 0.06 60 Atkinson
(2016)
22/58
Blast induced ground motion
Magnitude 2.3 (2.1) is equivalent to 620 kg (330 kg) of TNT.Distance of 2.6 km
R = distance,c = particle propagate velocityW = weight of TNT per delayρ = density of rock/soil
M = 2.45 + 0.73*Log(Y)
Blasts had no significant impactT.D.Stark-Slides-©
23/58
Pre-Failure Blasts & Earthquakes
Local TimeMoment
Magnitude Mw
Distance from
FundaoEvent
2:12:15 PM 2.2 <2 km earthquake (foreshock)
2:13:51 PM 2.6 <2 km earthquake (main shock)
2:16:03 PM 1.8 <2 km earthquake (aftershock)
3:45 PM Dam Failure
Pre-failure earthquakes on November 5, 2015 (Atkinson, 2016)
Eyewitness Accounts:- Shaking strong enough to cause computer to fall from tabletop and minor structural
cracking (Morgenstern et al., 2016)- Viviane Rezende – 2nd eqk shook truck on dam- Daviely Silva – Desk shaking & broken glass- MMI ~ 5 to 6
T.D.Stark-Slides-©
24/58
Pre-Failure Blasts & EarthquakesEyewitness Accounts of 11/5/15 Earthquakes:- Shaking strong enough to cause computer to fall from tabletop and minor structural
cracking (Morgenstern et al., 2016)- Viviane Rezende – 2nd eqk shook truck on dam- Daviely Silva – Desk shaking & broken glass- MMI ~ 5 to 6- M ~ 4 to 5
T.D.Stark-Slides-©
https://earthquakenepal2015.weebly.com/the-richter-scale-and-modified-mercalli-intensity-scale.html
25/58
T.D.Stark-Slides-©
26/58
Cumulative Moment Magnitude from ru
Moment magnitude/Richter magnitude SandSlope SandPlateau SandToe
Hypo central
distance and magnitude
2 km 3.4/ 3.5/ 3.45/5 km 4.2/ 4.3/ 4.2/
10 km 4.8/ 5.0/ 4.9/
Reference: Atkinson, G. (2015). Ground Motion Prediction Equation for Small-to-Moderate Events at Short Hypocentral Distances, with Application to Induced-Seismicity Hazards. Bull. Seism. Soc. Am., 105, doi: 10.1785/0120140142
Cross-Section 02– Left Abutment
CrackScarp
FUND15-SPTFUND06-CPT FUND16-CPT
T.D.Stark-Slides-©
FUND 07-CPT
SandSlope SandPlateau SandToe
Cross-Section 02– Left Abutment27/58
Pre-Failure CPT Data-FUND 06-CPT
T.D.Stark-Slides-©
Tailings SandSlope:Depth = 4.25 m
(Morgenstern et al., 2016)
28/58
Pre-Failure CPT Data-FUND 06-CPT
T.D.Stark-Slides-©
29/58
Pre-Failure Site Response Analysis
T.D.Stark-Slides-©
SandSlope
SandSlope
Tailings SandSlope:Depth = 4.25 m
30/58
Magnitude Scaling Factor
T.D.Stark-Slides-©
Idriss and Boulanger (2014)
31/58
Kσ
T.D.Stark-Slides-©
Idriss and Boulanger (2014)
• CRRL(σ’0=?) = CRRL(σ’ =1 tsf) *(Kσ)σ’0 = insitu effective vertical stress
32/58
Cross-Section 02– Left Abutment
CrackScarp
FUND15-SPTFUND06-CPT FUND16-CPT
T.D.Stark-Slides-©
FUND 07-CPT
Kα
SandSlope SandPlateau SandToe
33/58
Kα
T.D.Stark-Slides-©
Idriss and Boulanger (2010) K0 = 0.45 and Q = 10
34/58
Kα
T.D.Stark-Slides-©
Idriss and Boulanger (2010)
35/58
Pre-Failure CPT Data-FUND 06-CPT
T.D.Stark-Slides-©
FS
FS
36/58
Pre-Failure CPT Data-FUND 06-CPT
T.D.Stark-Slides-©
Tailings SandSlope :- Apparent fines content ~ 10%- Clay size fraction < 2% (Morgenstern et al., 2016)- Silt deposit has no cohesion
37/58
Excess Pore Pressure Ratio
T.D.Stark-Slides-©
Marcuson, W. F., Hynes, M. E., and Franklin, A. G., (1990). “Evaluation and use of residual strength in seismic safety analysis of embankments.” Earthquake Spectra, 6(3), 529–72.
38/58
Pre-Failure CPT Data-FUND 06-CPT
T.D.Stark-Slides-©
Tailings SandSlope :- Apparent fines content ~ 10%- Clay size fraction < 2% (Morgenstern et al., 2016)- Silt deposit has no cohesion
39/58
Witness Location and Description
T.D.Stark-Slides-©
#5
#4: waves developed in the central portion of the reservoir, accompanied by cracks forming on the left side and blocks of sand moving up and down on the left abutment setback
waves
Sand boils
(Morgenstern et al., 2016)
40/58
Pre-Failure CPT Data-FUND 16-CPT
T.D.Stark-Slides-©
Tailings SandToe:- Apparent fines content ~ 10% & Depth = 3.76 m
41/58
Pre-Failure CPT Data-FUND 16-CPT
T.D.Stark-Slides-©
FS
FS
FS
Depth = 3.8 m
42/58
Pre-Failure SPT Data-FUND 15-SPT
T.D.Stark-Slides-©
Tailings SandPlateau:- Apparent fines content ~ 10%- Clay size fraction < 2% (Morgenstern et al., 2016)- Silt deposit has no cohesion
(Morgenstern et al., 2016)
43/58
Pre-Failure SPT Data-FUND 15-SPT
T.D.Stark-Slides-©
FS
FS
Depth = 5.0 m
44/58
Cross-Section 02 – Main Shock
CrackScarp
T.D.Stark-Slides-©
ru = 0.11ru ~ 0.5ru = 0.3
Static ru =0.17 SandSlope SandPlateau SandToe
45/58
Cross-Section 02 – Fore and After Shocks
CrackScarp
T.D.Stark-Slides-©
ru = 0.07ru ~ 0.2ru = 0.13
Static ru =0.17 SandSlope SandPlateau SandToe
46/58
Cross-Section 02 – Total ru Values
CrackScarp
T.D.Stark-Slides-©
ru = 0.42ru ~ 1.00ru = 0.73
Static ru =0.17
SandSlope and SandToe => liquefied strength
SandSlope SandToe SandPlateau
47/58
Pre-Failure Blasts & EarthquakesTable 2.1: Pre-failure blasts and earthquakes on 11/5/ 2015 (Atkinson, 2016)
Failure 1.5 hours after shaking – NO seismic coefficient (Marcuson et al., 2003)
Local Time
Moment Magnitude Mw
Dsitance from Fundao
Identification ru for SandToe
ru for SandPlateau
ru for SandSlope
1:01:49 PM 2.1 2.6 km mine blast 0 0 0
1:06:06 PM 2.3 2.6 km mine blast 0 0 0
Static ru 0.17 0.17 0.17
2:12:15 PM 2.2 <2 km earthquake
(foreshock) 0.13 0.07 0.2
2:13:51 PM 2.6 <2 km earthquake
(main shock) 0.3 0.11 0.5
2:16:03 PM 1.8 <2 km earthquake
(aftershock) 0.13 0.07 0.2
3:45:00 PM
Dam Failure
ru increase 0.56 0.25 0.9
Total ru=static+dynamic 0.73 0.42 1
T.D.Stark-Slides-©
48/58
Figure . excess pore water pressure generation during cyclic loading with CSR =0.05 (reprinted from Morgenstern et al., 2016)The first solid points represent the pore pressure generation after 30 cycles loading with CSR =0.01. Some amount of pore water pressure is built up on three test within a few cycles with CSR = 0.05. Comment: The value of first red solid point is inconsistent with that presented on the table on page 2 of attachment D5
Excess pore water pressure
T.D.Stark-Slides-©
49/58
Critical state line – SandSlope
Liquefied
T.D.Stark-Slides-©
50/58
Critical state line – SandPlateau
Not Liquefied
T.D.Stark-Slides-©
51/58
Critical state line – SandSlope
Liquefied
T.D.Stark-Slides-©
52/58
Outline
• 5 November 2015 Failure• Chronology of events• Field observations• Subsurface investigation• Effect of blasting and earthquakes• Slope stability analyses• Summary
T.D.Stark-Slides-©
53/58
Cross-Section 02 – Total ru Values
CrackScarp
T.D.Stark-Slides-©
ru = 0.42ru ~ 1.00ru = 0.73
Static ru =0.17 SandSlope SandPlateau SandToe
SandSlope and SandToe => liquefied strength
54/58
Light Blue, Pink and Magenta Layer ru
Each dynamic event generates excess pore water pressure that reduces shear resistance along left abutment
ru = 0.13 /0.07/0.2
Static ru = 0.17
T.D.Stark-Slides-©
Initial Condition: ru=0.17
Final State:ru = 0.73/0.42/1.0
ru = 0.3 /0.11/0.5
ru = 0.13 /0.07/0.2
55/58
Limit Equilibrium Input Parameters
Material Type Color Strength
Typecohesion
(kPa)φ (°)
Liquefied su/σv' Ratio
ru
Sand Tailing Mohr-Coul
omb 33
Compacted Sand Mohr-Coul
omb 5 35
Foundation Mohr-Coul
omb 40
SandToe
Mohr-Coulomb 22 0.03
SandPlateau
Mohr-Coulomb 22 0.42
SandSlope
Mohr-Coulomb 22 0.03
T.D.Stark-Slides-©
56/58
Global Stability
T.D.Stark-Slides-©
Initial Global FS ~ 2.0
After Shallow Movement FS <1.0
SandSlope SandPlateau
SandToe
57/58
Summary
• Tailings dam stability- consider dynamic loads
• Fundao Tailings Dam Failure- accumulation of excess pore-water pressures
• Tailings Dam Failure usually results in an EnvironmentalProblem
T.D.Stark-Slides-©
58/58
Dynamic Analysis of Fundao Tailings Dam Failure
Presented by: TIMOTHY D. STARK & JIALE LIN
University of Illinois at [email protected]
Presented to:
51st Geotechnical Engineering ConferenceUniversity of Kansas
Lawrence, KansasNovember 14, 2019
T.D.Stark-Slides-©
59/58
T.D.Stark-Slides-©
60/58
Panel’s failure mechanism
T.D.Stark-Slides-©
Lateral extrusion of slime leading to collapse loose sand above it
61/58
Mobilized Instability Ratio
T.D.Stark-Slides-©
62/58
- Static ru = ?- γ*h = 52 m (18.0 kN/m3) = 936.0 kPa - u = 40 m (9.81 kN/m3) = 392.4 kPa- Static ru = 392.4 kPa/936.0 kPa = 0.42 - Static ru = 0.42
- MIR = Δσ/σ’m- Δσ = σ’1 − σ’3- σ’m = σ’1 + σ’2 + σ’3
Mobilized Instability Ratio
T.D.Stark-Slides-©
63/58
- FSLiq = ?- FSLiq = 3 to 4- Seismic ru = 0.0 to 0.1 - Static ru = 0.42 - Total ru < 0.7 so NO liquefied strength
- MIR = Δσ/σ’m- Δσ = σ’1 − σ’3- σ’m = σ’1 + σ’2 + σ’3
Horizontal displacement contours
T.D.Stark-Slides-©
Low shear strength was assign to slime layer. Global failure surface will not match that from field observation
64/58
T.D.Stark-Slides-©
Stability Analysis & Mobilized Instability Ratio65/58
Slime configuration in Section 01
Predominantly slimes – 100% slimes properties;Mixed sand and slimes – 50:50 sand and slimes properties; Interbedded slimes – 20% slimes properties and 80% sand properties; and Isolated slimes – 100% sand properties
T.D.Stark-Slides-©
Little slimes near slope – field observations
66/58
Mobilized Instability Ratio
T.D.Stark-Slides-©
Figure 6-6 Mobilized Instability Ratio
Global Stability67/58