1. Experimental Investigation on Limitation of the
Progression of Internal Erosion in Zoned Dams
by
Ricardo N.
Correia dos Santos
ricardos@lnec.pt
Under scientific supervision of
Dra. Laura Caldeira
Dr. Emanuel Maranha das Neves
PhD thesis prepared at
PhD thesis defence | 29th October, 2014
2. Question | Objective
Research question: ‘What is the influence of the upstream zone
limiting the progression of internal erosion through a crack in the core ? ‘
Sandy gravel Sandy gravel
Upstream
Upsztorenaem
zone
Core Filter
> Flow-limiting action
> Crack-filling action
Objective: Experimentally investigate what are the upstream
materials that may provide these two actions
3. Previous laboratory testing on erosion in soils by others
Sample
Ø ~ 10 cm
h ~ 12 cm
Core
Point
gauge
Cylindrical cell
Ø ~ 0.5 m
Sample Sample
Pea gravel
Ø ~ 2.5 cm
h ~ 3.8 cm
Core
Ø ~ 10 cm
h ~ 12 cm
Core
> Erosion along a
concentrated leak
Hole Erosion Test (HET)
Wan and Fell (2004)
> Erosion of an
immersed soil
JET Erosion test
Hanson and Cook (2004)
> Identification of
dispersive clays
Pinhole test
Sherard,
Dunningan and Decker (1976)
4. Previous laboratory testing on erosion in soils by others
> Susceptibility to suffusion of a soil
Upward Flow (UF) seepage test
Wan and Fell 2008
Sample
Ø = 30 cm
h = 25 cm
Downward Flow (DF) seepage test
Wan and Fell 2008
Sample
Ø = 30 cm
h = 30 cm
Drainage
layer
5. Previous Previous llaabboorraattoorryy tteessttiinngg oonn eerroossiioonn iinn ssooiillss bbyy ootthheerrss
Core
Filter
Sample
Ø=20.5 cm
Core: h = 10 cm
Filter: h = 20 cm
Section A-A'
A
A'
Filter
Crack
Pea gravel
Core
Perspex
Crack
Sample from
Proctor test
Core
> Filtering in a hole
Continuing Erosion Filter test
Foster and Fell 2000
> Filtering in a crack
Crack Erosion Test
Maranha das Neves (1989, 1991)
Crack filling by upstream uniform sand
6. New test cell developed at LNEC | Two laboratory tests
63 cm 40 cm
> Flow Limitation Erosion Test > Crack Filling Erosion Test
Upstream
material Core
Core
Upstream
material
Filter
Ø 30 cm
1 cm thk
12 17 25
> Test Cell
F
L
E
T
C
F
E
T
8. Experimental study in the FLET/CFET (Soils tested)
> Upstream material
A
5 broadly-graded soils
N1, N2 and N3 – Rb Grande Dam
P1 and P2 – Odelouca Dam
2 uniform granular soils
Sand A0 and Gravel A
6 gap-graded granular soils
GA1, GA2, GA3 and GA4
GN and GP (5% fines)
13
Upstream
soils
9. Experimental study in the FLET/CFET (Soils tested)
> Core and Filter
2
Filters
2
Cores
10. Experimental study in the FLET/CFET (Soils tested)
> Characterisation of soils used in the FLET/CFET
Standard
laboratory
testing
• Standard compaction tests
• Maximum/minimum density tests
• Permeability tests
Theoretical
analysis
• Susceptibility of soils to internal instability
• Ability of the soils to support an open pipe
Internal
erosion tests
• 9 Upward Flow (UF) tests on gap-graded soils
• 25 Hole Erosion Tests (HET) on core soils
11. Experimental study in the FLET/CFET (Soils tested)
> Erosion behaviour of gap-graded soils in the simpler UF test
UF test on soil GN
25% sand | 5% fines (NP)
4 24
3.5 21
Hydraulic gradient, i
Flow rate, Q
3 18
2.5 15
2 12
1.5 9
1 6
iboil
0.5 3
0 0
Time (min)
uf test_sufusion_time_gn.grf
1E-3 1E-2
5E-4 5E-3
icr
2E-4 2E-3
1E-4 1E-3
i i start boil
5E-5 5E-4
Velocity, V
Permeability, k
2E-5 2E-4
1E-5 1E-4
5E-6 5E-5
2E-6 2E-5
1E-6 1E-5
Hydraulic gradient, i
Average velocity, V (m/s)
Permeability, k (m/s)
0 0.5 1 1.5 2 2.5 3 3.5 4
Hydraulic gradient, i
Flow rate, Q (cm3/s)
0 40 80 120 160 200
12. > Erosion behaviour of the core soils in the simpler HET
Water content, w (%)
Dry unit weight, d (kN/m3)
4.61
12 13 14 15 16 17 18 19 2020.5
20
19.5
19
18.5
18
17.5
17
16.5
4.11
4.6 Sr = 100%
4.24
4.49
4.39
4.7
4.48
3.76
3.39
3.78
4.25
......
Erosion rate Index, I
Modified (core#4)
No erosion (core#4)
Standard (core#4)
'Reduced' (core#4)
Standard (core#20)
Core#4
Core#20
Experimental study in the FLET/CFET (Soils tested)
13. Experimental study in the FLET
> Testing conditions
19 FLETs on
broadly-graded
upstream soils
• DH = 2 m → i = 4
• Di = 12 mm (few with 10 and 16 mm)
• Core#4 (IHET = 4.1)
• Upstream soil: wopt or wopt ± 2% | 95 or 98%
14 FLETs on
granular
upstream soils
• DH = 2, 1.5 or 1 m | Di = 12 mm
• Core#4 (IHET = 4.1)
• Upstream soil: low w | Dr of 100%
33 FLETs 1 year | ~ 3 tons of soil
14. Experimental study in the FLET
> Observed behaviour types | Type F1– Flow limitation (self-healing)
15. Experimental study in the FLET
> Observed behaviour types | Type F2– Flow limitation (non-erodible)
Q = 650 l/h Q = 750 l/h Q = 800 l/h Q = 803 l/h
16. Experimental study in the FLET
> Observed behaviour types | Type F3– Slow down of erosion process
17. Experimental study in the FLET
> Observed behaviour types | Type F4 – Strong progression of erosion
18. Experimental study in the FLET
> Observed behaviour types | Type F4 – Strong progression of erosion
19. Experimental study in the FLET (Conclusions)
> Critical parameters influencing the flow-limiting action
N1 N1wet dry N1opt
N2dry N2opt N2wet
20%
Near
optimum
Dry side Wet side
dw vs pf200_broadly.grf
w wopt (%)
Fines content, FC (%)
40
30
20
10
P1opt
P2dry P2opt98;D10
P2wet98;D16
N3dry
N3dry;98%
N3wet
P2opt P2wet
C#4
P1dry
N3opt
12%
5%
N1wet
N soils | P soils
Type 1
.
Type 3
Type 4
N3wet
Type 1
Type 2
.
Type 4
P2wet98;D16
N2wet
50%
dw vs pc4_broadly.grf
C#4
N1dry N1opt
N2opt
N3opt
P1opt
N2dry
N3dry
N3dry;98%
P1dry
P2dry P2opt98;D10
P2opt P2wet
w wopt (%)
Gravel content, GC (%)
0
10
20
30
40
50
60
70
80
-2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 3
Broadly
graded
soils
• Fines content
• Fines plasticity
• Gravel content
• Water content
Gap-graded
soils
• Unlikely to limit flow
• Low Hydraulic head, DH
• Internal stability
Gravel content (%) Fines content (%)
w –wopt (%)
20. Experimental study in the FLET
> Proposed rules for estimation of likelihood of flow limiting action
Upstream material
Upstream material is cracked Upstream material is not cracked
Water content
Internally unstable
soil with iU
Internally stable
soil with kU (m/s)
Zone
type
Fines content
Fines
type
dry side near opt wet side > 1 < 1 > 10-2 ‡ < 10-5 ‡
I NA NA 0 0 0
II <5% Any ** ** ** VU U VU ***
III 5 to 12% SPF L* N* N* U N U L
HPF‡ N* U* N*
IV >12% SPF L* N* N* N L *** VL
HPF‡ U* N* L*
V 5 to12% NP ** ** ** VU N VU ***
VI >12 to 20%‡ NP L* N* N* U N U L
VII >20 to 30% NP N U U N L *** VL
VIII >30% NP U VU VU N L *** VL
Symbol
/colour
Qualitative
descriptor
Example of
likelihood interval
VL Very Likely 0.98–0.999
L Likely 0.70–0.98
N Neutral 0.30–0.70
U Unlikely 0.02–0.3
VU Very Unlikely 0.001–0.02
21. Experimental study in the CFET
> Testing conditions
4 CFETs on
broadly-graded
upstream soil
• DH = 2 m | Di = 12 mm or 16 mm
• Core#4 (IHET = 4.1) and Filter G (Dr = 100%)
• Upstream soil N1: wopt or wopt ± 2% | 95%
37 CFETs on
granular soils
• DH = 2m | Di = 12 or 16 mm
• Core#4 or Core#20 (IHET = 4.1)
• Filter S or Filter G (Dr about 60%)
• Upstream soil: low w | Target Dr of 100%
41 CFETs 1 ½ year | ~ 4.5 tons of soil
22. Experimental study in the CFET
> Observed behaviour types | Type C1– rapid ‘crack-filling’
23. Experimental study in the CFET
> Observed behaviour types | Type C2a – Filtering after ‘some’ erosion
24. Experimental study in the CFET
> Observed behaviour types | Type C2b – Filtering after ‘excessive’ erosion
25. Experimental study in the CFET
> Observed behaviour types | Type C3 – Continuing erosion
26. Experimental study in CFET (Conclusions)
> Critical parameters influencing the crack-filling action
Broadly
graded
soils
• Highly unlikely to
fill cracks
Gap-graded
soils
• pA0 versus D15F
• 5% fines may
decrease capability
• Core soil with I > 4
has no influence
27. Experimental study in the CFET
> Proposed rules for estimation of likelihood of crack-filling action
Key features of
upstream zone
Embankment zoning in the erosion path at downstream of the core
Rapid crack-filling
Formation of a self-filtering
action
layer
No crack filling nor
filtering mechanism
Fines
content
Effectiveness of
upstream soil
Key feature of the filter Estimate erosion zone by Foster and Fell (2001)
D15F <
Transi
D15F ≥
No
Some
Excessive
2.9 mm
tion
5.1 mm
erosion
erosion
Erosion
Continuing Erosion
<5%
psand > 30% and
no fines content
VL L U VL L N U
Transition L L – U* U L N U U
psand ≤ 20% and
5% of fines
U U VU N U U VU
≥12% NA VU VL L N VU
Symbol
/colour
Qualitative
descriptor
Example of
likelihood interval
VL Very Likely 0.98–0.999
L Likely 0.70–0.98
N Neutral 0.30–0.70
U Unlikely 0.02–0.3
VU Very Unlikely 0.001–0.02
28. Final conclusions
Initial question
‘What is the
influence of
upstream zones? ’
• Answer: Some soils compacted in certain conditions,
when located upstream of a crack in the core, may
effectively provide the flow limiting action or the
crack-filling action.
Proposed
objective
Experimental study
of both actions
• Development of a new test device and of 2 laboratory
tests (FLET and CFET), and their test procedures.
• Identification of the potential behaviour types and of
the influence of critical parameters for each action.
• Proposed rules to aid practical engineers in the design
phase or to estimate probabilities for dams in operation.
29. Future research
FLET
Flow limiting
action
• Extend testing to greater variety of upstream soils prepared
to a wider range of test conditions.
• Evaluate if self-healing ability of very dry upstream
materials also occurs in soils with highly plastic fines.
• Investigate the influence of the permeability of coarse
grained (stable) soils.
CFET
Crack-filling
action
• Extend testing using materials from existing dams.
• Evaluate the relation pA0 versus D15F for other soils.
Both
• Evaluate influence of the erosion rate of the core (for I < 4).
• Evaluate influence of the flow orientation.
30. Progression of internal erosion in the embankment
Tunbridge Dam, Australia
Zoned earth dam (28 m) Source: Jeffery Farrar (2005)
Hanson e Hunt (2007)
Homogeneous dam (25 m) Source: USDA
31. Experimental Investigation on Limitation of the
Progression of Internal Erosion in Zoned Dams
by
Ricardo N.
Correia dos Santos
ricardos@lnec.pt
Under scientific supervision of
Dra. Laura Caldeira
Dr. Emanuel Maranha das Neves
PhD thesis prepared at
PhD thesis defence | 29th October, 2014
Hinweis der Redaktion
Good morning to all.
I will start my presentation with the research question of the thesis, which is
“What is the influence of the upstream zone limiting the progression of erosion through a crack in the core?”
It is believed that the upstream zone may provide the ‘flow limitation action’ or the ‘crack-filling action’ or even both.
The flow limitation is related with the capability of the upstream zone to impose hydraulic head losses that are high enough to limit the flow converging to the crack.
To obtain the isolated influence of this mechanism the best way is to assume that the crack occurs in the core and also the filter.
The worse-case scenario should occur when the upstream zone also cracks.
The crack-filling action occurs when the material eroded from the upstream zone is transported through the crack filling it.
The presence of a filter is, in this case, an absolute requirement.
However there are very few studies on these mechanisms.
With that said, The main objective of the thesis was to experimentally investigate what are the materials that may provide these two actions.
The next obvious step was to search in literature for laboratory tests that could be used to model the flow-limiting and the crack-filling actions.
However, I found that majority of the available tests do not consider the upstream zone.
In fact, many of the tests only consider one material, usually the core.
They are focused on the evaluation of the erodibility of the core, like the Hole Erosion Test or the JET, or on the identification of dispersive clays, like the pinhole test,
Or in the study of the susceptibility to suffusion of soils, like the Upward flow seepage test or the downward flow seepage test.
More complex tests consider the core and the filter.
They are used to evaluate the performance of filters, due to erosion along a hole in the core, like the continuing erosion filter test, or along a crack, like in the crack erosion test.
Professor Maranha das Neves performed a few tests in this device using a uniform fine sand as crack-filler, and concluded that such material was very effective at filling cracks and stopping erosion.
In fact, these were the only tests that I have found, focused on the influence of an upstream zone.
However, to be able to examine coarser soils I felt the need of a larger test device.
It was therefore developed a new test cell, which was completely built at LNEC.
It allows the compaction of up to 3 zones (upstream material, core and filter), and its dimension copes with soils up to the gravel size.
The cell is composed by several aluminum pieces and a Perspex front panel.
The cell was developed to allow performing two distinct tests.
The FLET to study the isolated influence of the flow limitation action, and the CFET to study the crack-filling action.
In the FLET there is no filter, and usually a hole is drilled in the core and in the upstream material.
In the CFET there is a filter and usually the hole is only drilled in the core.
The compaction and drilling of the specimen and the assembly of the cell is made sequentially.
The preparation of the specimen in the CFET is slightly different than in the FLET, because there is the need to consider the filter.
The test setup is similar in both tests.
Water is forced to pass through the specimen, applying a differential between the upstream head and the downstream head.
During test we measure the flow rate and the water level in three piezometers.
The turbidity and the erosion in the outlet chamber are recorded with a digital camera.
After construction of the cell, it was necessary to choose the soil to be tested.
5 broadly-graded soils were selected as upstream material.
These soils were collected during construction of two dams in Portugal.
Soils N1, N2 and N3 (in red) are from Ribeiro Grande dam and have non-plastic fines.
Soils P1 and P2 (in blue) are from Odelouca Dam and have fewer fines but with some plasticity.
They were selected in accordance with guidelines available in literature related with the likelihood of occurrence of the flow limitation action.
There were also selected 2 uniform granular soils and 6 gap-graded granular soils.
Soil A0 is similar to the material tested by prof. Maranha das Neves that proved to be highly effective at filling cracks.
Soils GA1 to GA4 are gap-graded soils formed by mixing soil different percentages of soil A0 with a uniform gravel.
Soil GN has also 5% of non-plastic soils and soil GP 5% of plastic fines.
A total of 13 soils were examined in the upstream zone.
2 soils were selected as Core
Core#4 and a finer one Core#20.
And
2 soils were selected as Filter
Filter S and a coarser Filter G
These filters were intentionally selected to be excessively coarser to be unable to retain effectively both cores.
The next step was to characterize the soils selected.
For that, there were used standard laboratory testing and some theoretical analysis, for example, to evaluate da susceptibility of soils to internal instability.
There were also performed 9 Upward flow seepage tests on the gap-graded soils, and
25 Hole Erosion tests on the core soils
to observe the behaviour of these soils in more simpler test configurations.
In this slide, I show the results of one Upward Seepage test, in particular, on soil GN.
In this type of test, the specimen is subjected to vertical flow (from the bottom to the top), and the hydraulic gradient is slowly increased in steps.
the flow rate is measured, and the average velocity of the flow and the permeability of the soil are estimated.
In some soils it was observed the development of sand boiling, visible on the top of the specimen
When this phenomenon starts there is an increase of the flow rate, velocity and permeability of the soil.
I was expecting that this material could fill the pipe in the core, in the CFET.
Here I show the pieces of the Hole Erosion Test, built at LNEC, for the study of the erodibility of the core soils along a preformed hole.
.. the look of the specimen after assembly of the cell and before starting the test, and during the test, where we can observe the widening of the pipe.
This is how a typical specimen looks like at the end of a test, and this the paraffin mold of the resulting pipe.
These tests allowed to investigate the erosion proprieties of core#4 and of core#20 for different compaction water contents and compaction efforts.
The goal was to compact the core soils in the larger cell with an erosion rate index of about 4 (which corresponds to a moderate erosion behaviour).
There were performed 19 FLETs using the broadly graded upstream materials, for various compaction properties.
and
14 FLETs on the gap-graded
In total there were performed 33 tests, which took about 1 year to perform and involved the compaction of about 3 tons of soil.
These tests allowed the identification of four behaviour types.
In tests showing Type F1, the upstream material erodes at fast rate,
but when its coarser particles start being transported they are retained at the interface with the core,
resulting in a decrease of the flow rate. Eventually the pipe in the upstream material collapses and self-heals.
In behaviour Type 2, the flow rate increases initially but then stabilizes.
This occurs when the upstream material in non-erodible for the testing conditions.
In this time lapse, we see that the flow rate, turbidity and deposition of eroded material all decrease during the test.
After 40 minutes we have 800 liters/hours and after 54 minutes flow is practically the same and there is no evidence of erosion.
Behaviour Type 3 is initially similar to Type 1.
Strong progression of erosion, then coarser particles eroded from the upstream material become blocked at the interface with the core, reducing the flow rate
However, erosion in the core does not stop completely, and suddenly there is a blowout of the coarser particles.
This results in the rapid increase of the flow, which doesn’t decrease anymore.
I have here a small video showing this blowout phenomenon.
Finally, Behaviour type F4 is associated to fast and continuous increase of the flow rate, and strong progression of erosion in the core and mainly in the upstream material.
In this small video, I show an example of a test showing behaviour Type F4.
Flow is always increasing and there is a continuous deposition of material in the outlet chamber.
Ater dissasembly of the cell, usually it can be seen an open pipe in both materials.
This is the parafin mould of the pipe. In this case the pipe in the upstream material is much larger than in the Core.
The results of the FLETs on the broadly graded soils showed that the critical parameters for the flow-limiting action are:
-fines content, gravel content and compaction water content of the upstream soil.
The plasticity of the fines is also an important parameter.
Tests on upstream soils with non-plastic fines are in red and with plastic fines in blue.
Solid symbols correspond to tests in which the flow practically stopped (Type F1),
These have low water content, fines content up to 20% and gravel content higher than 50%.
Hollow symbols are tests with no flow limitation (Type F4),
These typically occur in soils with non-plastic fines, with fines content higher than 30%, and low gravel content.
________________________________________
The results of the FLETs on the gap graded soils showed that these soils are unlikely to limit flow.
Limitation may occur for low hydraulic head losses, or in internally stable soils.
Based on the results of the tests
I have proposed in the thesis some rules for preliminary estimation of the occurrence of the flow limiting action.
These rules depend on the critical parameter identified in testing (fines content, type of plasticity of the fines and water content).
For example, an upstream material with more than 30% of non-plastic fines compacted on the wet side is very unlikely to be able to provide the flow limiting action considering that the upstream material is cracked.
If there is no crack and the material is internally stable then it is very likely to limit flows.
In relation to the tests conducted to evaluate the crack-filling action,
There were performed 4 tests using the broadly graded upstream materials.
and
37 tests on the gap-graded soils
In total there were performed 41 tests, which took about 1 year and a half to perform,
and involved the compaction of about 4,5 tons of soil.
The tests allowed the identification of four behaviour types.
In tests showing behaviour Type C1, the flow rate decreases fast to a practically null value.
There is a formation of a ‘jet of sand’ that is retained by the filter and fills in the pipe.
In behaviour Type C2a, the flow initially increases because the filter is unable to filter the washed in material from the upstream zone (which is spread all over the filter),
but then the filter ends up sealing the flow after some erosion of the core,
I
Behaviour type C2b is similar to C2a,
However, the filtering mechanism occurs only after an excessive erosion of the core.
The filter becomes more contaminated with particles from both materials.
Finally, behaviour Type C3 corresponds to the case where the flow rate never decreases because the filter is unable to retain the soils eroded from the upstream material and from the core.
The results of the tests on the gap-graded soils showed that the occurrence of the crack filling action depends on the relation between sand fraction of the upstream material that can be eroded and of the D15 of the filter.
The higher the sand content of the upstream material and the lower the D15 of the filter more likely is crack-filling to occur.
Fines content up to 5% may decrease potential of soils to provide the crack-filling action.
Core soils with index higher than 4 should not have great influence on the crack filling action, because this phenomenon is very rapid.
The tests on the broadly graded soils showed that upstream soils that are able to support an open pipe are highly unlikely to fill cracks.
Based on the results of the tests on the CFET
I have proposed some rules for preliminary estimation of the occurrence of the crack action.
These rules use the critical parameter identified in the tests.
For example: If we have an upstream soil with sand content higher than 30% without fines, and a filter with D15 lower than 2,9 mm,
Then crack-filling action is very likely to occur.
Final conclusions.
In relation to the initial question about “what is the influence of upstream zones”?
The answer is that, in fact, some soils compacted in certain conditions, when located upstream of a crack in the core, may effectively provide the flow limiting action or the crack-filling action.
I believe that the initial proposed objective was fully achieved
With the Development of a new test device and of 2 laboratory tests, and their test procedures
With the identification of the potential behaviour types and of the critical parameters influencing each action
With the proposed rules to aid pratical engineer in the design phase or to estimate probabilities for dams in operation.
There are multiple possibilities to extend this research,
For example, using the FLET,
- Extend testing to greater variety of upstream soils prepared to wider range of test conditions.
- Evaluate if self-healing ability of very dry upstream materials also occurs in soils with highly plastic fines.
- Investigate the influence of the permeability of coarse grained (stable) soils
Using the CFET,
- Extend testing using materials from existing dams
- Evaluate the relation between sand content and D15 for other filters
Using both,
Evaluate the influence of the erosion rate of the core (for soils more erodible than those tested)
And evaluate the influence of the flow orientation of the flow.