Laboratory Plate Load Testing of Non-Segregating ...

Pinheiro, M.; Proskin, S. & Li, B. (2017) Laboratory plate load testing of non-segregating tailings. In: Proc. 21st Intl. Conf. Tailings & Mine Waste, Eds. G.W. Wilson, D.C. Sego & N.A. Beier, 5-8 November 2017, Banff, AB, Canada.

Laboratory Plate Load Testing of Non-Segregating Tailings M. Pinheiro & S. Proskin Thurber Engineering Ltd., Calgary, Alberta, Canada

B. Li Canadian Natural Resources Ltd., Calgary, Alberta, Canada

ABSTRACT: Non-segregating tailings (NST) are tailings produced by mixing mature fine tailings or thickened tailings with cyclone underflow coarse tails, and a coagulant to accelerate solids settling and densification. NST is prepared at a target sand-to-fines ratio of about 4 or more, thereby producing a predominantly frictional material. A research project was setup to measure the bearing capacity of NST, in order to assess its readiness for capping or landform construction activities. Five plate load tests were carried out on NST dewatered to different solids content. Index testing, and strength testing (using pocket penetrometer, handheld field vane and a Rimik® cone device) were also conducted to further characterize the tailings. The results revealed that the ultimate bearing capacity of NST increases with plate settlement, as expected for a frictional material whose shear strength depends on stress and confinement levels. For a small settlement (20 mm), the ultimate bearing capacity varied from 3–16 kPa. For larger settlements (60 mm), ultimate bearing capacities reached nearly 30 kPa for the NST at 80.7% solids content. 1 INTRODUCTION In 2009, the Alberta Energy Regulator (AER), then known as Energy Resources Conservation Board (ERCB), issued “Directive 074: Tailings Performance Criteria & Requirements for Oil Sands Mining Schemes.” Under this directive, oil sands operators were required to come up with strategies to meet specific targets, such as having a trafficable tailings deposit with a minimum shear strength of 10 kPa within five years after tailings has been deposited in a dedicated disposal area. Directive 074 proved itself to be difficult to comply with, and has since been replaced in 2016 by “Directive 085: Fluid Tailings Management for Oil Sands Mining Projects” (AER, 2009, 2016). Directive 085, which is currently being completed and enhanced, does not set out a particular strength target for trafficability. Instead, it sets out a tailings management framework in which performance criteria are first developed by the operator for each tailings deposit. The proposed performance criteria are then assessed by the regulator for approval. Each criterion must identify indicators (e.g. material properties, residual settlement, trajectory to trafficability) and measures (e.g. solids content, sand-to-fines ratio, cone penetration testing) that will be used to assess and track progress towards the various stages of reclamation – trafficability being one of them. Non-segregating tailings (NST) are tailings that have been significantly dewatered by using thickeners and cyclones. They are produced by mixing either mature fine tailings or thickened tailings with sand derived from cyclone underflow coarse tails. Coagulants (e.g. CO2, CaO) are added to accelerate solids settling and densification. NST is typically prepared at a target sandto-fines (SFR) ratio of about 4 or more, thereby producing a predominantly frictional material. Plate load testing and ultimate bearing capacity are here investigated as an alternative gauge to assess the shear strength of NST. The underlining concept would be to conduct plate load tests on NST samples prepared at different solids contents by weight (SBW), and estimate the ultimate bearing capacity of each sample. Solids content would then be used as a measure to as-

sess the deposit’s trafficability and readiness for capping or landform construction activities and monitor progress towards reclamation. A research project was setup, and the scope of work entailed the following tasks:  Develop procedures for the index testing, strength characterization, and plate load testing.  Design and fabricate a plate load assembly for testing the NST under laboratory conditions.  Conduct plate load tests.  Conduct strength characterization tests on the NST samples. a. Cone penetration testing using a hand-held electronic cone penetrometer. b. Shear vane testing using a hand-held shear vane device. c. Pocket penetration testing using a hand-held pocket penetrometer.  Collect samples for geotechnical index testing, including Atterberg limits, solids content, bitumen content, and particle size distribution.  Analyze results from index, strength characterization and plate load testing programs.  Determine the ultimate bearing capacity of the NST material by means of analytical methods. 2 METHODOLOGIES The following sections briefly describe the methodologies employed to prepare the NST and to characterize it in a laboratory setting. NST production and testing, including plate load testing were all carried out at Canadian Natural’s “Bitumen Production / Applied Process Innovation Centre,” located at Horizon Oil Sands mine, just north of Fort McMurray, Alberta, Canada. 2.1 NST Material Preparation The NST material was produced at a SBW of 64.4% and at a SFR of about 5. The tailings material was then discharged into four 1.0 m × 1.0 m × 0.7 m (length × width × height) aluminum pans: Pans #12, #11, #09 and #08 (Figure 1). The initial plan was to prepare batches of NST at solids contents ranging from around 65% to 85%, and discharge each batch into separate pans. However, this plan was found to be impractical – it would take more time and resources to prepare several batches at distinct solids content than to prepare a single homogeneous material. Moreover, NST produced at SBWs above 80% would not be representative of field conditions. The tailings in each pan was then subject to different dewatering conditions so that by the end of two weeks, they were expected to have reached target solids contents of 65%, 70%, 75% and 85%. The dewatering strategies involved:  Laying a 0.10 m thick layer of dry and loose beach sand at the bottom of the pan to provide bottom drainage, at least to the maximum storage capacity of this sand layer.  Installing a French drain at the sand layer to allow the under-seepage water to drain out, and enhance the bottom drainage.  Decanting the “free” water accumulated on the tailings surface.  Mounting lights, and subjecting the surface of the tailings material to varying light schemes and intensity (i.e. number of light bulbs on), to enhance evaporation.  Setting up fans to blow air moisture away from the tailings surface, and enhance evaporation. Table 1 summarizes the various dewatering conditions applied to the NST in each pan. Table 1. Pan setup and dewatering conditions. Average target SBW Sand beach layer at the bottom of pan French drain Removal of decanted water Lights / Light intensity Fan

Pan #12 65% Yes No No No No

Pan #11 70% Yes No Yes No No

Pan #09 75% Yes Yes Yes Yes / Medium No

Pan #08 80% Yes Yes Yes Yes / High Yes

Figure 1. (a) Overview of pans and lab setup. (b) Overall plate load test setup showing pan, reaction frame, hydraulic ram, load cell, LVDT, data acquisition equipment and laptop.

All four pans were initially placed on top of Desna scales to monitor the change in weight (in kilograms) of the tailings material with time prior to plate load testing. Figure 1a shows the pans, light stands and Desna scales used in this research project. Figure 1b depicts the plate load testing setup, which will be described in the next subsection. NST dewatering occurred over a two-week period. Weights were recorded at irregular intervals, usually during weekdays, as no one worked full time in the lab. The results of the dewatering program are presented in Figure 2, in terms of solids content. The NST material dewatered rapidly after deposition as indicated by the sharp jump in solids content from 64.4% to almost 75% in Pans #11, #09 and #08. The NST material in Pan #12 also underwent similar rapid dewatering; however, this is not readily observed in Figure 2 because the decant water was not pumped out from this pan. Figure 2 also shows target (from Table 1) and actual average SBW values achieved prior to plate load testing. The solids content of the tailings in Pan #08 was on target, but the others were not. Anyway, an NST at a solids content below 75% would have been too soft/weak to be tested, as discussed in Section 2.2. 0.29 Pan #12 [Target 65%] Pan #11 [Target 70%] Pan #09 [Target 75%] Pan #08 [Target 80%]

85

plate load test on Pan #08

0.46

plate load test on Pan #09

80

80.7, 0.63

0.65

80.3, 0.64 76.3, 0.81

75 decant water pumped out from Pans #11, #09, #08

70

plate load test on Pan #11

0.87

75.4, 0.85

1.11

65 60

1.40

1

2

3

4

5

6

7

8 9 10 11 12 Days of Measurement

13

14

15

16

17

1.73

Figure 2. Average solids by weight of the NST in each pan over an approximately two-week period.

Average Void Ratio

Average Solids Content by Weight (%)

90

The NST in Pans #09 and #08 developed millimetre-to-centimetre thick cracks. Desiccation was not uniform because of uneven wind and light exposure (some of the lightbulbs fixed to the stand were burnt out, and air was blown from a fan located at one of the corners of the pan). Unsaturation in the driest areas may have reached down a couple centimetres. The topmost surface of the NST was notably glossy, likely due to bitumen migration during deposition and sedimentation. The topmost layer was also visibly richer in fine particles than the underlying one. 2.2 Plate Load Testing The plate load testing equipment included the following:  Custom-made load frame.  Custom-made 203.2 mm (8-inch) diameter steel plate.  Custom-made adapters and extension rods.  4-ton capacity hydraulic Porta-Power ram assembly.  1-ton capacity electronic load cell.  Linear Variable Differential Transformer (LVDT).  Data acquisition system and laptop. The load frame was especially designed for this project to accommodate the dimensions of the pans, the expected maximum loads, as well as transportation requirements. The load frame, steel plate, adapters and extension rods were all manufactured in Calgary. The hydraulic ram, the disassembled frame and companion parts were shipped to Horizon mine, and reassembled on site, as shown in Figure 1b. The plate load test consisted of pushing the steel plate onto the tailings material, and recording both plate displacement (in millimetre) and load (in kilogram). Any desiccated/crusted tailings material within the testing area was removed, and the exposed surface evened out with a spatula before laying down the steel plate. The data acquisition system was set to automatically collect and store readings of displacement and load every second. These readings were plotted on a laptop in real-time for quick interpretation of the tailings behaviour. The first displacement increments were very small (0.01 – 0.05 mm) because the initial NST response was relatively stiff. As the NST deformed and softened, the imposed displacement increments were increased in steps up to a maximum value of about 1 – 3 mm. The recorded total displacement and load values were transferred to an Excel spreadsheet to calculate the relative total displacement of and applied pressure under the plate. These were then plotted to help assess the results and determine when to terminate the test. Plate load tests were completed on Pans #11, #09 and #08. No test was carried out on Pan #12 due to time constraints and because its NST was still very soft. Moreover, the bearing capacity of the tailings material in Pan #12 might have been below the sensitivity of the load cell. The results of the plate load testing are discussed in Section 3.3. 2.3 NST Strength Characterization Strength properties of the NST material were also indirectly assessed by other means based on pocket penetrometer test, field shear vane test and cone penetration test. The methodology of each test is briefly described below. Desiccated and cracked material was removed and, the exposed surface levelled before testing. The pocket penetrometer test consisted of pushing the piston of a Humboldt pocket penetrometer device onto the NST, and recording the unconfined compressive strength, Sq (in tons/ft2 or kg/cm2) directly from a scale indicator. An adaptor foot was attached to the piston to increase the device’s accuracy in the lower range of measurement. The field shear vane test consisted of pushing a Humboldt vane into the NST material to the desired depth, slowly turning the handle clockwise at constant speed until failure, and recording the value, Sv, registered on the graduated scale of the shear vane device. A shaft rod was used to prevent friction between NST and extension rod. A large 50.8 mm × 101.6 mm four-bladed vane size was used to measure shear strengths in the range of 0 to 8.125 kPa. The cone penetration test consisted of pushing a Rimik® CP140II cone vertically into the NST material at a constant speed until the tip of the cone reached the bottom of the pan. The

cone was pushed at three locations. The Rimik assembly comes with two size cones (areas: 130 mm2 and 323 mm2). The larger size cone was used to improve the device’s accuracy in the lower range of measurement. The Rimik device automatically tracks insertion depth, speed and load. It then averages the recorded data over a 25-mm range, and converts the load into a cone resistance index, qc (in kPa). The limitations of these tests and their applicability to NST-like material is well understood. Here, they were used as qualitative tools to improve our understanding of the NST material, to help interpret the plate load test results, and to compare the NSTs prepared at different solids contents – they provide a relative measure of resistance when comparing these materials. 2.4 NST Index Characterization Two 4-litre pails (pails #1 and #2) were filled with the NST prepared on site at 64.4% SBW, and shipped to Calgary for index characterization, including Atterberg limit tests, solids content and bitumen content determination, and sieve-hydrometer analyses. These tests were conducted according to the following methodologies:  Atterberg limit tests were completed according to ASTM D4318 standards.  Solids contents were measured using the conventional oven technique, as per ASTM D2216 standards.  Bitumen content was determined by means of Dean-Stark extraction method.  Sieve-hydrometer analyses were carried out on non-bitumen extracted samples according to ASTM E11-09 and D422-63 standards, respectively. Mining fines content, defined as the mass of fines (< 44 μm) divided by the mass of mineral solids and bitumen, was obtained directly from the particle size distribution curves. The solids contents of the NST in Pans #11, #09 and #08 were also measured on site using the quick oven technique, as per ASTM D4959 standards. NST samples were collected at various depths by pushing a 30.5 cm (1-foot) long, 2.5 cm (1-inch) diameter cylinder sampler into the tailings, then digging around the cylinder, and placing a cap at the top of the cylinder to minimize water loss when carrying the sample for quick oven testing. 3 RESULTS 3.1 NST Index Characterization The results of the index testing completed on the fresh NST tailings (before any dewatering took place) are summarized below:  Atterberg limits could not be measured because of the high sand content (about 80%). The NST is classified as a cohesionless non-plastic material.  The solids contents of the NST in pails #1 and #2 were 66.3% and 65.6%, respectively. These values were determined using the conventional oven drying method, and are slightly higher than 64.4%, measured on site using the quick oven method.  Bitumen content, determined by Dean-Stark, ranged from 0.12% to 0.15%.  Particle size distribution curves are shown in Figure 3 for pails #1 and #2. Mining fines content is about 16%, and SFR is about 5.2. The solids content of the NST material in Pans #11, #09 and #08 was again determined immediately prior to plate load testing, as described in the previous section. The measured solids contents agree with the average values estimated based on the weights of the pans. Averaged void ratio values were around 0.65 or higher. Interestingly, Robertson et al. (2011) pointed out, based on field measurements on Suncor’s composite tailings in Pond 5, that effective stress starts to develop at a critical void ratio of approximately 0.65. The strength characterization results presented in the following section corroborate their findings.

clay

silt

80

sand

mining fines < 44 μm

Percent Finer (by Weight)

90 70 60 50 40 30 20

geotechnical fines < 75 μm

100

pail #1 pail #2

10 0 0.001

0.01

0.1 Particle Size (mm) Figure 4. Particle size distribution curves of NST from pails #1 and #2

1

3.2 NST Strength Characterization Both pocket pen and field shear vane tests generate readings that are directly correlated with shear strengths. The correlations depend on the testing device employed. For the Humboldt pocket pen, the shear strength, Sp (in kPa) was calculated using the following equation:

Sp 

Sq  1  95.76   16  2

(1)

where Sq is the unconfined compressive strength; “95.76” is a factor that converts tons/ft2 into kPa; and “16” is to account for the foot adaptor, which has an effective area sixteen times greater than the piston (Humboldt, 2010). For the Humboldt shear vane, the shear strength, Sf (in kPa), was calculated using the equation below: Sf  0.625  Sv

(2)

where Sv is the value registered on the graduated scale of the vane device (Humboldt, 2011). Shear strengths obtained from field shear vanes are normally corrected to account for the plasticity and liquidity of the material being tested (Bjerrum, 1972). Such correction is not required here, as the NST is non-plastic. The cone resistance index (qc) has been correlated with several soil properties, including shear strength and friction angle (Kulhawy & Mayne, 1990). For the hand-held Rimik cone, the following equations were used for the shear strength, Sc (in kPa), and friction angle, ϕ (in degrees), respectively:

Sc 

qc  σv N kt

q  and tan ϕ  0.10  0.38  log  c   σv 

(3)

where σv is the total vertical stress; σv' is the effective vertical stress; Nkt is the cone bearing factor. Nkt is an empirical factor that ranges from 10 to 20 with an average of 15, the value assumed here (Eid & Stark, 1998). Figure 4a shows the shear strengths calculated from the pocket pen and field shear vane measurements. We refrained from using the term “undrained” as we are not sure of the actual drainage conditions of the tests carried out for the project. For Pan #11, shear strengths increase gradually with depth, whereas for Pans #09 and #08, much higher shear strengths are encountered at the surface due to desiccation; however, at about 25 cm from tailings surface, shear strength values are almost as low as those measured in Pan #11. Figures 4b,c display the shear strengths and friction angles estimated from the Rimik cone. For Pan #11, shear strengths are relatively constant. For Pans #09 and #08, much higher shear strengths are encountered at the surface of the NST due to desiccation – and near the contact be-

tween NST and beach sand layer due to under drainage. Inferred friction angles vary from about 18º–24º for the very loose NST in Pan #11 to about 27º–30º for the loose to medium dense NST in Pans #09 and #08. The desiccated NST, as expected, shows higher friction angles, above 40º in average. The results from the Rimik cone, pocket pen and vane shear tests are consistent with one another. 0.6

(b) Rimik

(a) pocket pen

Distance from Bottom of Pan (m)

0.5

(c) Rimik

NST (top)

0.4 vane shear 0.3 Pan #11 Pan #09 0.2

Pan #08

NST (bottom)

0.1

beach sand 0.0 0

3 6 9 Shear Strength (kPa)

12

0

3 6 9 Shear Strength (kPa)

12

10

20 30 40 Fricion Angle (deg)

50

Figure 4. Profiles of (a) shear strength from pocket pen and vane shear tests, (b) shear strength from Rimik tests and (c) friction angle from Rimik tests for Pans #11, #09 and #08.

3.3 Ultimate Bearing Capacity Five plate load tests were completed within two days: one in Pan #11, two in Pan #09 and two in Pan #08. The results of the plate load testing are presented in Figure 5, in terms of averaged pressure (force divided by plate area) versus displacement curves. These curves portray the typical plate load test response of loose to medium cohesionless soils (Lambe & Whitman, 1969). That is, for the loose to medium NST in Pans #09 and #08, the curves show an initial stiffer mechanical response followed by a softer behaviour after a sharp bend in the curve, corresponding to a local shear failure. The initial stiffer response is likely governed by the presence of desiccated material at the surface of the NST. For the very loose NST in Pan #11, the shear zones at the sides of the footing are not well defined and no surface heave was observed. This is termed “punching failure”, much like that observed in Pans #09 and #08 (Figure 6). Typically, the “bearing capacity” (ql) of the soil is defined as the pressure at which there is a pronounced change in slope, corresponding to the “knee” or bend in the pressure-displacement curve. The “ultimate bearing capacity” (qult) is the bearing pressure that causes a sudden catastrophic settlement of the foundation (Lambe & Whitman, 1969). This catastrophic settlement was not observed in our tests for two reasons: (a) the NST was loaded at a displacementcontrolled mode, and (b) the shear strength of sand-dominated materials depends on stress and confinement levels; that is, when the load increases, so does the stress level and therefore the shear strength. As such, the ultimate bearing capacity of sands increases as more load is applied, and failure is not clearly defined. For that reason, some authors have defined the ultimate bearing capacity of coarse-grained soils as the bearing pressure for a settlement equivalent to 10% of the plate diameter (Briaud, 2013).

0

Vertical Pressure (kPa) 10 20 30

40

0

(a)

10

(b)

Pan #11 Pan #09 Pan #08

30

1m

40 [2-1]

Pan

Test Test Location Location [2] [1] Load Frame

[1]

60

diam. 0.2 m [2]

70

[1] [2]

80 90

0.4 m

Vertical Displacement (mm)

20

50

1m

displacement at 10% plate diam.

0.3 m [2-2]

0.3 m

test locations

100

Figure 5. Plate load testing (a) results and (b) locations.

Figure 6. Punching failure observed on the NST in (a) Pan #11 and (b) Pan #08

In light of the above discussion, the bearing capacity of the NST material varied from about 1 kPa for the 76.3% SBW material to around 9–12 kPa for the 80.3–80.7% SBW material. If one resorts to Briaud’s definition of ultimate bearing capacity, then qult varied from about 3 kPa for Pan #11 to around 12–16 kPa for Pans #09 and #08. Evidently, if larger settlements can be tolerated (e.g. in case of construction equipment), then higher ultimate bearing capacities can be expected. For example, for a plate displacement of 60 mm, an ultimate bearing capacity of 20-30 kPa is obtained. Results of field plate load tests can be used to estimate the ultimate bearing capacity of actual footings (track or wheel footprint). For cohesionless soils, the ultimate bearing capacity of actual footings is proportional to the ultimate bearing capacity obtained from a plate load test by the ratio of footing size to plate size (Das, 2010).

4 SIMPLIFIED BEARING CAPACITY ANALYSIS Several formulations are available to calculate the ultimate bearing capacity (qult) of soils. Terzaghi’s, Meyerhof’s, Hansen’s and Vesic’s methods are among the most widely used ones (Bowles, 1988). Meyerhof’s seems to be the best method for frictional soils, and will be applied here. For concentrically loaded circular footings, Meyerhof’s formulation (in SI units) reduces to the following equation: q ult  c  N c  s c  d c  q  N q  s q  d q  0.5 γ  B  N γ  s γ  d γ

(5)

where:  c (in kPa) is the cohesion (or shear strength) of the soil.  ϕ (in degrees) is the friction angle of the soil.  γ' (in kN/m3) is the effective unit weight of the soil.  q (in kPa) is an external load (equal to γ'D for loading plate embedded into the soil).  B (in m) is the plate diameter; D (in m) is the embedment depth of the plate.  Kp is the coefficient of passive resistance (= tan2(45º + ϕ/2)).  Nc, Nq and Nγ are dimensionless bearing capacity factors; Nc = (Nq – 1) cot(ϕ), Nγ = (Nq – 1) tan(1.4ϕ), Nq = eπ tanϕ Kp.  sc, sq and sγ are dimensionless shape factor; sc = 1 + 0.2Kp; sq = sγ = 1 + 0.1Kp for ϕ > 10º.  dc, dq and dγ are dimensionless depth factor; dc = 1 + 0.2√Kp (B/D); sq = sγ = 1 + 0.1√Kp (B/D) for ϕ > 10º. Table 2 compares measured and calculated ultimate bearing capacities under various modelling assumptions. The mechanical properties of the NST (i.e. shear strength and friction angle) were obtained from the Rimik cone data. The saturated unit weights were calculated directly from the solids contents in Figure 3. In all cases, Meyerhof’s method overestimates the ultimate bearing capacities, with the best estimates being based on shear strength values. This suggests that the friction angles in Figure 4 are slightly overestimated. In Meyerhof’s formulation, the soil was assumed homogenous. This may not apply to Pans #09 and #08 because of the upper 5–10 cm NST material. Employing Meyerhof’s method for a strong soil layer over a weak soil layer profile (Hanna & Meyerhof, 1980) yields ultimate bearing capacity estimates that are still comparable to the homogeneous case with the loading plate at surface. Table 2. Measured versus calculated ultimate bearing capacities based on Meyerhof’s method Pan #11 Pan #08 Pan #09 76.3 80.3 80.7 SBWave (%) γs(ave) (kN/m3) 18.6 19.8 19.8 Measured qult for plate 20 mm and 60 qult-20mm (kPa) 3 16 12–14 mm below NST surface qult-60mm (kPa) 20–24 8 28–32 Plate at NST surface and 60 mm below qult-surface (kPa) 4 17 30 surface (ϕ from Rimik data) qult-60mm (kPa) 10 32 51 ϕave (deg) 22º 29º 32º Plate at NST surface / NST as a coheqult (kPa) 4 13 20 sive material (S from Rimik data) Save (kPa) 0.7 2.1 3.2 Plate at NST surface / Two-layered qult-surface (kPa) NST in Pan 21 30 frictional material (ϕ from Rimik data) ϕave-top10cm (deg) #11 is homo40º 41º ϕave-elsewhere (deg) geneous 27º 30º

5 CONCLUSIONS The bearing capacity of the NST material varied from as low as 1 kPa to about 9–12 kPa. The ultimate bearing capacity was found to increase with plate settlement, as expected for a sanddominated frictional material, whose shear strength depends on stress and confinement levels. Based on the plate load test results, it was also evident that the ultimate bearing capacity of NST increases with solids content. However, because we tested three pans effectively at two solids

content (Pan #11 at about 76% SBW and Pans #09 and #08 at about 80-81% SBW), there is insufficient data to derive a reliable relationship between bearing capacity and solids content. This relationship is very likely to be non-linear, as is the case for undrained shear strength and void ratio (Sobkowicz et al., 2013; Moore et al., 2014). While considering the above, it is also important to note that we have presented a preliminary evaluation based on laboratory measurements using about 0.50 m thick layers of laboratory prepared and deposited NST with a loading surface of 0.20 m diameter plate. Inferring field estimates of bearing capacity must consider the geotechnical issues associated with extrapolating laboratory data and empirical relationships to field scale designs. This NST was prepared under controlled lab conditions. The NST in the pans were exposed to dewatering conditions, and both may differ significantly from field deposits. Furthermore, field bearing capacity must consider the larger volume and depth of soil experiencing higher stresses imposed by larger scale loads. The NST material was at a very loose state after deposition and initial dewatering. Liquefaction of loose, saturated sands may be caused by either static or dynamic, undrained loading. The assessment of static and seismic liquefaction potential is certainly advised as part of the design of NST deposits. A similar investigation is advisable if there were potential for deposits to be trafficked by equipment applying dynamic/cyclic loads. 6 ACKNOWLEDGEMENTS The authors would like to acknowledge the contributions of Messrs. Vincent Gao, Chenxi Zhang, Iain Gidley and Trempess Moore for this research project, as well as the support from a number of staff within Canadian Natural. 7 REFERENCES Alberta Energy Regulator (AER). 2009. Directive 074: Tailings Performance Criteria and Requirements for Oil Sands Mining Schemes. February 3, 2009. Alberta Energy Regulator (AER). 2016. Directive 085: Fluid Tailings Management for Oil Sands Mining Projects. July 14, 2016. Bjerrum, L. 1972. Embankments on soft ground. In Proc. Specialty Conf. of Earth and Earth Supported Structures: 1-54, Vol. 2, ASCE, New York. Bowles, J.E. 1988. Foundation Analysis and Design. 4th edition, McGraw-Hill. Briaud, J.-L. 2013. Geotechnical Engineering: Unsaturated and Saturated Soils. Wiley. Das, B.M. 2010. Principles of Geotechnical Engineering. 7th edition. Stamford: Cengage Learning. Eid, H.T. & Stark, T.D. 1998. Undrained shear strength from cone penetration test. In: Geotechnical Site Characterization, P.K. Robertson & P.W. Mayne (Eds.), Rotterdam: Balkema. Hanna, A.M. & Meyerhof, G.G. 1980. Design charts for ultimate bearing capacity of foundations on sand overlying soft clay. Canadian Geotechnical Journal, 17, 300-303. Humboldt. 2010. H-4200 & H-4200F Product Manual. Soil Penetrometer. Illinois, US. Humboldt. 2011. H-4227 Product Manual. Vane Inspection Kit. Illinois, US. Kulhawy, F.H. & Mayne, P.W. 1990. Manual on Estimating Soil Properties for Foundation Design. Report EL-68000, Electric Power Research Institute, EPRI, August 1990. Lambe, T.W. & Whitman, R.V. 1969. Soil Mechanics. New York: John Wiley & Sons. Moore, T., Zhang, C. & Pinheiro, M. 2014. Geotechnical performance of ATA™-treated fluid fine tailings. In Paste 2014, Vancouver, BC. Robertson, A.M., Caldwell, J., Wells, P.S. & Barnekow, U. 2011. Advances from the Wismut project on soft tailings cover technologies. Retrieved from: http://www.infomine.com/publications/docs/Robertson2011.pdf Sobkowicz, J.C., Pinheiro, M., Moore, T.W. & Moffett, R.H. 2013. Predictive model for thick lift placement of silica-treated fluid fine tailings. In Tailings & Mine Waste 2013, Banff, AB, Canada.