CYCLIC CONSTANT NORMAL STIFFNESS TESTS ON SAND

Pile-supported jacket structures were used for the Block Island Wind Farm project, the first offshore wind farm in the United States (DEEPWATERWIND 2012). Due to the significant length of the piles (60 m in some cases), the frictional resistance along the sides of these piles, termed shaft friction, is the main component of axial capacity. Shaft friction can be degraded due to cyclic loading from wind and waves, and understanding this behavior is critical for safe design. It is hypothesized that degradation of shaft friction can occur due to the contraction of a thin layer of soil, called the shear band, immediately in contact with the pile (DeJong, White & Randolph 2006). In the laboratory, shaft friction is often modeled by performing interface shear tests with soil and the pile material. Interface shear tests, whether monotonic or cyclic, are commonly performed under constant normal load (i.e. direct shear). This boundary condition does not accurately model the shear behavior of piles as they do not account for changes in the normal stress during shearing. LeHane and White (2004) showed that, during loading, contraction of the shear band caused a reduction in the normal stress acting on a pile. In contrast, dilation of the shear band caused an increase in the normal stress. This behavior can be recreated in laboratory interface shear tests by imposing a constant normal stiffness condition (CNS) (e.g. with a spring) on the sample. The primary objective of this thesis was to modify an existing cyclic simple shear device to be able to perform cyclic shear tests under constant normal stiffness (CNS) conditions. A second objective was to perform a series of monotonic and cyclic tests in support of a research project funded by the United States Bureau of Safety and Environmental Enforcement (BSEE) to understand the behavior of piles under cyclic loading. Both monotonic as well as cyclic CNS tests were performed on samples of Monterey Sand at various densities and values of normal stiffness. In the monotonic tests, dilation caused an increase and contraction caused a decrease in the shear strength of the samples compared to constant normal load (CNL) tests. Contraction of the soil along the interface occurred in all the cyclic tests, resulting in a decrease in shear resistance with each cycle of loading. The influence of initial normal stiffness and displacement amplitude was also investigated.


A A A ABSTRACT BSTRACT BSTRACT BSTRACT
Pile-supported jacket structures were used for the Block Island Wind Farm project, the first offshore wind farm in the United States (DEEPWATERWIND 2012). Due to the significant length of the piles (60 m in some cases), the frictional resistance along the sides of these piles, termed shaft friction, is the main component of axial capacity. Shaft friction can be degraded due to cyclic loading from wind and waves, and understanding this behavior is critical for safe design. It is hypothesized that degradation of shaft friction can occur due to the contraction of a thin layer of soil, called the shear band, immediately in contact with the pile (DeJong, White & Randolph 2006).
In the laboratory, shaft friction is often modeled by performing interface shear tests with soil and the pile material. Interface shear tests, whether monotonic or cyclic, are commonly performed under constant normal load (i.e. direct shear).
This boundary condition does not accurately model the shear behavior of piles as they do not account for changes in the normal stress during shearing. LeHane and White (2004) showed that, during loading, contraction of the shear band caused a reduction in the normal stress acting on a pile. In contrast, dilation of the shear band caused an increase in the normal stress. This behavior can be recreated in laboratory interface shear tests by imposing a constant normal stiffness condition (CNS) (e.g. with a spring) on the sample.
The primary objective of this thesis was to modify an existing cyclic simple shear device to be able to perform cyclic shear tests under constant normal stiffness (CNS) conditions. A second objective was to perform a series of monotonic and cyclic tests in support of a research project funded by the United This year has given me a series of amazing memories and personal connections that I will cherish for a very long time.
Lastly, I would like to thank my family for the extraordinary support given that you give me day by day. My parents, for their efforts in providing for us an v exceptional life filled with affection and opportunities to grow. My brother, for being the person that throughout my whole life has stood by my side the longest. vi   TABLE OF CONTENTS  TABLE OF CONTENTS  TABLE OF CONTENTS  TABLE OF  Step       It is hypothesized that one of the major mechanisms responsible for a reduction in axial capacity of offshore piles under cyclic loading is contraction of the soil along interface between the pile and soil (termed the shear band).
Contraction along the interface reduces the normal stress acting on the pile, which will cause the shear strength of the soil surrounding the pile to decrease because soil strength is frictional in nature (i.e. stress dependent).
The shaft resistance of piles can be estimated in the laboratory by performing interface shear tests. Commonly, these tests are performed under constant normal load conditions even though this boundary condition does not allow for a change in normal stress due to contraction or dilation at the interface By imposing a constant normal stiffness boundary condition on the sample instead of a constant normal load, it is possible to account for the changes in the normal stress due to contraction or dilation. These tests are called constant normal stiffness shear tests, but are never done in practice and rarely done in research labs.
The primary objective of this thesis is to modify an existing cyclic simple    The direct shear test consists of two phases. The first phase is the consolidation phase, in which the normal stress is applied by the load frame and the sample consolidates. The normal stress usually ranges between 50 kPa and 500 kPa and is held constant throughout the test.
After the soil sample has consolidated the shearing phase occurs. Here a horizontal motion is applied, which can be either strain-controlled (preferred) or stress controlled.
The strain-controlled variant occurs at a constant rate of horizontal displacement that moves the lower half of the shear box. A load cell attached to the upper half records the shear stress on the horizontal plane. The shear stress will increase with increasing horizontal displacement until the sample fails.
Alternatively, the stress-controlled variant involves increasing the shear stress on the sample in steps and measuring the resulting horizontal displacement.
This is rarely done in modern geotechnical testing.  One of the first to use direct shear tests to determine soil properties was Alexandre Collin (Skempton 1949). He studied the stability of clay slopes and published in 1846 his findings in "Recherches Experimentales sur Quelques Principes de la Mecanique Terrestre". A description of a shear box-type apparatus was included in the publication. The publication did not become relevant until 70 years after its publication, as the topic of slope stability regained importance within the scientific community.
The direct shear apparatus, as it is known today, was devised by Glennon Gilboy in 1936. Gilboy's apparatus was the first to feature a strain-controlled approach. The apparatus presented measurements of the shear load and led the soil sample contract or dilate freely while keeping a constant load. A picture and a drawing of the apparatus are shown in Figure 5 and Figure 6. Current devices are not significantly different in principle to Gilboy's apparatus.  (Gilboy 1936)  Results of 4 performed direct shear tests are shown in Figure 7, Figure 8 and Figure 10. The tests were performed on samples of Monterey sand, which is a commonly tested sand for laboratory earthquake studies. Properties of the Monterey sand were determined by Morales (2014) and are listed in Table 1. The shear displacement rate was 1 mm/min. The tests were performed at 4 different normal stresses: 100 kPa, 200 kPa, 300 kPa, and 400 kPa. The relative density of the soil samples ranged from 10% to 20%, and thus the samples can be considered to be in a loose condition.   Figure 8 shows the relationship between vertical displacement and horizontal displacement during shear. When a soil is sheared it will either dilate or contract. Dilation means that the soil will increase in volume; contraction means that the volume decreases. If the soil state happens to be close to the critical state line there will be little to no change in volume (see Figure 9). In a direct shear test, the change in volume due to contraction or dilation is manifested as a change in height of the sample.
The critical state line is useful for understanding the dilative/contractive behavior of soils. For different samples with the same void ratio (or relative density), the confining stress determines if a sample dilates or contracts during shear, and the relative amount is governed by the distance from the critical state line. Samples will dilate most when confining stresses are low and the density is high. With increasing confining stresses, the amount of dilation decreases until the sample becomes contractive (note that in direct shear testing the confining stress is the normal stress).
Dilative samples will commonly contract slightly at the beginning of shearing before dilation begins (e.g. tests at 100 kPa and 200 kPa in Figure 8)  The stresses at failure as represented by the Mohr-Coulomb failure envelope are shown in Figure 10. This envelope consists of the normal and shear stresses on the failure plane (i.e. horizontal in the direct shear test) at failure. The envelope should approximate a straight line although the true failure envelope is curved due to dilation (at low confining stresses) and contraction (at high confining stresses). For granular soils (sands, gravel) under drained loading conditions, the envelope will start at zero. Cohesive soils and soils under undrained loading conditions possess some shear strength at zero normal stress (termed "cohesion"), implying that the failure envelope will start at a value greater than zero in the shear stress axis. Cohesion is commonly represented with the letter "c" where: • : Effective normal stress The interface friction angle is typically described as a percentage of the soil's friction angle. The interface friction angle for concrete, for example, ranges between 0.5 and 0.66φ (depending on the roughness), and for steel the interface friction angle is often assumed to be 0.66φ (approximately 20 o -25 o ). For the current study, it was possible to utilize a base plate built by Hildebrandt (2018) (Figure 11) to perform shear tests on a sand-steel surface.
Four, monotonic interface shear tests were performed on samples of Monterey sand at normal stresses of 100 kPa, 200 kPa, 300 kPa, and 400 kPa.     Compared to offshore loading, earthquake loading is typically much higher in frequency, resulting in undrained loading that can lead to liquefaction in saturated sands, silts, and even some gravels. Cyclic loading is characterized in the laboratory by either cyclic triaxial tests or cyclic simple shear tests. The latter is comparable to direct shear tests and will be discussed further here.
In a cyclic simple shear test the soil sample is confined by a stack of rings or a wire-reinforced membrane that allows for Ko consolidation (zero lateral strain) and simple shear ( Figure 15). This allows for the development of shear strain (γ), which is the horizontal displacement divided by the height of the sample, from the application of shear stress. generated; if the sample dilates negative pore water pressure is generated.
In Figure 17, for example, near the 30 th cycle, the effective stress decreased to the point that the shear strain started increasing rapidly. Results of a typical strain-controlled cyclic simple shear tests are shown in Figure 19. In this case, the applied cyclic shear strain stays constant. With increasing cycles, the shear stress decreases until a residual shear stress state is reached. As with stress-controlled tests, excess pore water pressures increase with each cycle. The advantage of simple shear testing over direct shear testing is that in simple shear the failure plane is not forced. In the case of direct shear, the failure plane is forced to be the interface between the two sliding halves.
A disadvantage of simple shear testing, in general, is that the stack of rings precludes the possibility of performing interface tests. Tests performed by Boulon and Foray (1986) indicated that the shear stress along piles is concentrated in a thin zone about ten times the grain diameter. This zone, referred to as the interface, is where the dilation and contraction occurs. This interface is constrained by the soil surrounding it; it can be thought of as an infinite number of uncoupled springs along the interface. These springs have an initial deflection that represents the earth pressure (normal stress) acting on the pile. If dilation occurs is the spring loaded, increasing thus the normal stress acting on the pile. If contraction occurs the spring will be unloaded, decreasing the normal stress. This is termed the constant normal stiffness (CNS) boundary condition. The CNS condition for pile-soil interfaces mentioned above is shown in Figure 20. " " is the change in interface thickness due to dilation or contraction.
The initial normal load acting on the pile "σh" will increase or decrease due to dilation or contraction by Δσh: where: • : Shear modulus of the soil around the pile is not the intention for this thesis, therefore, won't be compared with the 3 methods mentioned above regarding the increase in the lateral pressure. The results presented in Figure 21 show clearly that an increase in lateral pressure due to dilation occurs when the pile is sheared. In the tests performed by Lehane and White (2004) only dilation occurred; however, it can be assumed by logic that contraction would lead to a decrease in the lateral pressure.  In this chapter are presented three different CNS direct shear apparatuses.
They account in different ways for the same behavior: during shearing, dilation and contraction of the interface leads to a variation in the normal stress.
Additional to schematic drawings are presented some results of tests performed with the respective apparatuses. This should give a good idea into how the CNS condition is achieved and how it reflects itself in the actual testing. Lam and Johnston (1982) The first authors known to devise a CNS direct shear test apparatus were Lam and Johnston in 1982. Their intention, however, was to account for a different phenomenon. When a socketed pile foundation experiences axial displacement, dilation of the socketed walls occurs due to the roughness of the interface between pile and rock. This dilation occurs against the stiffness of the rock. Figure 22 illustrates this behavior. If the pile is moved axially up or down, an increase in the lateral stress will occur due to the roughness of the rock-pile interface. The increase in the normal stress will be proportional to the amount of dilation and the stiffness of the rock.  (Lam & Johnston 1982) With this in mind, Lam and Johnston (1982) devised a direct shear apparatus that imposed a constant normal stiffness on the rock-concrete sample.
The principle of the shear apparatus is shown in Figure 23. The bottom rock box is only allowed to move horizontally whereas the upper concrete box, which models the surface of the pile, is only allowed to move vertically. The roughness of the surface of the concrete and rock is represented in the drawing with 3 peaks.
Due to this roughness, any horizontal displacement (shear displacement) of the rock box results in a vertical displacement in the concrete box. The vertical displacement will trigger the stiffness of the spring (kCNS) which represents the stiffness of the rock. The spring is in the form of a steel bar that bends; thus the bending stiffness of this bar is the stiffness of the spring.
A photograph of the CNS direct shear apparatus is shown in Figure 24.  (Lam & Johnston 1982) Some results are presented in Figure 25. Tests 4 and 5 were performed with a stiffness of 300 kPa/mm, and tests 6 and 7 with a stiffness of 950 kPa/mm.
The stiffness is reported as the change in normal stress (i.e. force divided by the sample area) acting on the sample divided by the vertical displacement. Dilation of the sampe triggered the spring which led to an increase of the normal stress.
The higher normal stress increased the shear strength of the probe. It can also be seen that tests performed with higher constant normal stiffness have denoted higher increases in the normal stress. Results of the static shear tests are shown in Figure 27. Medium dense samples (relative density = 60%), which have the tendency to dilate, showed a notable increase in the normal stress. The loose samples (relative density ≈ 0%) contracted, which led to a decrease in the normal stress. There is also a clear trend when the stiffness of the bar is increased: higher stiffness's lead to higher changes in the normal stress.
CNS tests performed on dilative samples exhibited higher shear strengths than CNL tests with the same density, CNS tests that contracted exhibited lower shear strengths than their CNL counterparts.  (Tabucanon et al. 1995) Results of cyclic direct shear tests on loose (relative density ≈ 0%) and dense (relative density = 60%) sand samples are shown in Figure 28 and Figure   29. The tests were strain-controlled with a displacement amplitude of 1 mm.
In the loose samples ( Figure 28), contraction during the first cycles due to the large displacement lowered the normal stress to 0 (i.e. unloading of the spring; right figure). This significantly decreased the mobilized shear stresses (left figure).  (Tabucanon et al 1995) The dense samples ( Figure 29) experienced less contraction and thus a lower decrease in the normal and shear stresses. Where: • Δσh: Increment of normal stress • Δt: Vertical displacement • kn: Constant normal stiffness When dilation or contraction occurred, the normal stress is automatically increased or decreased, thus providing a constant normal stiffness condition. A schematic of the direct shear apparatus is provided in Figure 30.  (Porcino et al. 2003) The CNS tests were performed on an aluminum-sand interface (interface CNS test). The 3 sands used were natural silica sand with properties listed in Table 2. Different aluminum surfaces were also used; ranging from smooth to rough. Smooth surfaces will make a sample more contractive whereas rougher surfaces will make a sample more dilative.  (Porcino et al. 2003) Results of a test performed with a rough aluminum surface on a dense sample of Ticino sand are shown in Figure 31. Note that in the graphs the symbol is the constant normal stiffness !"# . A value of = 0 kPa/mm refers to a CNL test. The initial normal stress applied was $ = 150 kPa.   Figure 2). For the initial testing, the original reaction bar provided by GEOCOMP, with a thickness of 38.5 mm, was replaced by a significantly smaller aluminum bar with a height of 6.9 mm ( Figure 32). The theoretical bending stiffness of the plate was calculated using equations 4 and 5. These calculations were done in order to have an initial idea of the value of the stiffness that was being employed.

CHAPTER III CHAPTER III CHAPTER III CHAPTER III ----CNS TESTING M CNS TESTING M CNS TESTING M CNS TESTING METHODOLOGY ETHODOLOGY ETHODOLOGY ETHODOLOGY
The actual stiffness of the system should be between the calculated stiffness values from equations 4 and 5, where the first represents the stiffness of an ideal simply supported beam and the second of an ideal completely fixed beam. where • & = Young's modulus of elasticity (69 GPa for aluminum) • ' = Moment of Inertia (ℎ ) ./12) • ( = Span length (200 mm)

• = bending stiffness in kPa/mm
Using equations 4 and 5, the 6.9 mm high aluminum bar resulted in a stiffness ranging from 127 kPa/mm to 510 kPa/mm.
The actual stiffness of the system was determined through bending tests in the direct shear apparatus. The set-up of the tests is shown in Figure 33. The shear box was replaced by a piece of steel. Loads were applied onto this steel piece.
Since it is much stiffer than the system it was assumed the steel piece did not deform under the applied load, meaning that all of the measured deformations occurred in the bar.
The vertical deformation in the center of the aluminum bar was measured along with the load in the vertical load cell. Note that the measured vertical deformation is the combined deformation of multiple components of the system as a whole. The deformation includes the elastic extension of components such as the threaded rods and the steel block. The results of 10 3-point bending stiffness tests are shown in Figure 34, with an average stiffness of around 225 kPa/mm. This was the value of constant normal stiffness ( !"# ) assumed for tests performed with the 6.9 mm high aluminum plate.  CNS tests were performed by modifying the ShearTrac-II. As already mentioned above, the original bar was replaced by a significantly smaller one. The consolidation phase was conducted in the same way as a CNL test; the loading frame lowers until the desired normal stress is reached. Since the bar used for CNS testing is considerably less stiff it will bend noticeably in a CNS test (see Figure 36). During the shear phase in CNS tests, dilation or contraction needs to occur against a constant normal stiffness. For this to happen, the vertical load frame must not move during shear. All the vertical displacement must occur in the cross bar (i.e. spring) to maintain constant stiffness. This can be achieved in the ShearTrac-II by setting the P-Gain to "0" in the software. This effectively locks the cross arm during shear. Note that this must be done after the consolidation phase is concluded and before the shearing phase is started.
Step by step instructions for performing Monotonic and cyclic CNS tests can be found in Appendix B.

Cyclic CNS Testing Cyclic CNS Testing Cyclic CNS Testing Cyclic CNS Testing
Cyclic CNS tests were performed in a modified Cyclic ShearTrac-II system manufactured by the GeoComp Corp. In its original form, this apparatus is made for running cyclic simple shear tests, which by definition are constant volume tests using either stacked rings or wire-reinforced membranes. As a result, the existing Geocomp system was modified in two ways to perform cyclic direct shear tests under CNS conditions. First, a steel angle was attached to the base of the restraining arm to hold the upper half of the shear box in place during the cyclic shearing. The second modification was to replace the cross arm with bars of various thicknesses (i.e. various stiffnesses). After the consolidation phase, the loading frame is locked by setting the normal control to "no control" in the cyclic table of the software.
Detailed instructions for setting up the apparatus for running CNS direct shear tests can be found in Appendix C.

Sample Preparation Sample Preparation Sample Preparation Sample Preparation
For comparing different tests results, it is important for samples to be prepared in a uniform fashion. In this subchapter will be explained the methods in which the different samples were prepared.

Monotonic Tests
The loose monotonic tests were prepared by filling a scoop with the sand sample and pouring it as fast as possible in the shear box. The exceeding sand was then scraped off so that the sample had an even surface. The excess sand was cleaned off the shear box and subsequently was weighed the sample. For soil on soil tests was the weight of the loose samples between 208 and 210 gr. For in interface tests was weight between 122 and 124 gr.
The dense monotonic tests were prepared by pouring sand in the shear box through a funnel until it was half full and stomping it with a pestle. When using the funnel is important to maintain a constant height. The same procedure was used for the other half of the shear box. The excess sand was cleaned off the shear box and subsequently was weighed the sample. The weight of the samples was between 220 and 223 gr for sand-sand tests and between 122 and 124 gr for the interface tests.

Cyclic Tests
The cyclic tests were prepared using an exact predetermined amount of sand. In the case of the loose tests, this amount was 115.3 gr (Dr = 10%) and for the dense tests it was 122.5 gr (Dr = 60%). Note that the shearing occurs within the interface. This means that the density in the interface will be decisive for the overall behavior occurring in the shear tests. The void ratio or relative density that is presented is calculated for the whole volume of the shear box and could be not representative of the relative density in the interface.

RESULTS RESULTS RESULTS
This chapter presents the results of the monotonic CNS tests. Results of CNL tests will also be provided so that the effect of the CNS condition can be compared to traditional direct shear test results. All Tests were performed with Monterey Sand (see Table 1 Figure 42). All 6 tests were consolidated to a vertical effective stress of 100 kPa and the stiffness, kCNS, of the CNS tests was 225 kPa/mm.
The relative density of the tests was between 20 and 30%. All the CNS tests showed considerably higher shear strength than the CNL test. This is in accordance with the results issued by other authors shown herein. Figure 38 shows 5 CNS tests performed with almost equal conditions (same starting normal stress and similar relative density), yet there is some variability.
This variability is likely due to several factors: differences in the relative density, differences in the sample preparation, inhomogeneity of the sand, inaccuracies in the shear apparatus, etc. These dissimilarities are all inherent to direct shear testing, CNL as well as CNS. In the case of CNS testing any dissimilarity will cause the normal stress to vary differently in each test; this does not happen in CNL testing. Because of this will be the variability in CNS testing higher than in CNL testing.  For the CNS tests, the variation of the normal stress during shearing is shown in Figure 41. There is a considerable spread in the change in the normal stress. It can also be noted that at the beginning of shearing the normal stress decreases; this is due to the initial contraction of the sand sample during shearing. In Figure 42 is presented the vertical displacement with the horizontal displacement. All tests dilated.  A general pattern can be seen when looking at Figure 43 to Figure 54: In CNS testing, dilative samples will have higher shear strengths, whereas contractive samples will have lower shear strengths when compared to CNL tests under same conditions.
Dilative CNS tests dilated consistently less than their CNL counterparts.
This is due to the increase in the normal stress during shearing (higher normal stress leads to lower dilations). The opposite happens to contractive CNS tests: lower normal stresses lead to higher contractions.
Dilative CNS tests reached their peak shear stress at larger shear displacement than the CNL tests. The shear displacement at which the contractive CNS tests reached the peak shear stress did not differ much from the CNL tests. The results of the 16 monotonic CNL and CNS shear tests are summarized in this subchapter. Figure 55 shows the Mohr-Coulomb failure envelopes for those tests and the results are summarized in Table 4. The medium dense CNS tests present the highest values of shear stress whereas the loose CNS tests the lowest.
The reason for this is that the medium dense samples exhibited dilation which led to an increase of the normal stress, and so increasing the shear strength of the sample. The values of the correlated angles of friction are listed in Table 4.
Note that the peaks of the CNS tests are not at the starting normal stresses of 100, 200, 300 and 400 kPa. This is because the normal stress changed due to the CNS condition and thus the peaks occurred at a different normal stress.    Table 5.
All CNS tests contracted; the reason for this is the smooth steel surface and relatively high normal stresses. This led to a decrease in the normal stress in all the CNS tests. Dense CNS tests had a lower contraction and thus lower decreases in the normal stress.
Dense CNS tests consistently presented higher peak shear strengths than the dense CNL tests. The value of the residual (i.e. large displacement) shear strengths of the dense CNS and CNL tests were close to each other.
The loose samples contracted significantly. This led to a large decrease in the normal stress in the CNS tests, and these tests had the lowest shear strengths.
The values of peak shear strength were reached at low horizontal displacements of approximately 0.5 mm. Increasing the confining pressure did not cause a significant change in the horizontal displacement at which the peak is reached.    Coulomb failure envelopes for  Coulomb failure envelopes for  Coulomb failure envelopes for Interface Test  Interface Test  Interface Test  Interface Test   Table 6 and Figure 68 presents the Mohr-Coulomb failure envelope for the Mohr-Coulomb failure envelope and resulting friction angle for the interface tests.
The medium dense CNS tests presented the highest shear stresses whereas the loose CNS tests the lowest. This is consistent with the Sand-Sand tests.
Note that the peaks of the CNS tests are not at the starting normal stresses of 100, 200, 300 and 400 kPa. This is because the normal stress changed due to the CNS condition and thus the peaks occurred at a different normal stress. Tests were performed with Monterey Sand (see Table 1) on a smooth steel interface with a medium dense (Dr = 60%) and a loose relative density (Dr = 10%).
The cyclic period was 10 s. All tests presented in this subchapter were run with a constant normal stiffness of 225 kPa/mm. Different horizontal displacement amplitudes were analyzed: 0.25 mm, 0.5 mm, 0.75 mm, 1 mm and 2 mm. This was done to have an idea about how many cycles to expect until the normal stress reached zero depending on the horizontal displacement amplitude.
The starting normal stress was 100 kPa. All the tests exhibited contractive behavior and thus the normal stress decreased throughout shear. The tests were ended when the normal stress reached a value of 0 kPa. Table 7 shows the test matrix for these tests, including levels of cyclic displacement amplitude, initial normal stress, initial void ratio, initial relative density and the value of constant normal stiffness. Figure 69 to Figure 78 show the test results in terms of shear stress vs. horizontal displacement, shear stress vs.
normal stress (i.e. stress path), vertical displacement vs. shear displacement, and degradation of shear stress and normal stress vs. cycles of loading.  The results are summarized by the loss in the peak-to-peak shear stress vs. the number of cycles in Figure 79 and Figure 80.     The amount of change in the normal stress during shear depends very strongly on the value of the normal stiffness and the amount of contraction or dilation of the sample. Higher values of normal stiffness will cause a more rapid change in the normal stress. This is corroborated by the results displayed in Figure   83. It can also be seen in Figure 83 that the starting peak-to-peak shear stress of the 3 tests is different. This is due to the reduction of normal stress within the first cycle, with the amount differing depending on the stiffness that was used. This is visualized in Figure 85.  and sample collections at 6 points (see Figure 86, URI 1 to 6).   Table   8.   Table 9. The values of normal stress, constant normal stiffness, and density are averages of the different SCPT and CPT locations. After consulting with NGI, it was determined that the displacement amplitudes employed in the cyclic tests should be determined from the results of the monotonic tests. A displacement amplitude was to be chosen that represented a shear stress prior to reaching the peak shear stress, and another displacement amplitude just past the peak shear stress.
All the samples were prepared to the same density with the intention of making dense specimens. No information on the minimum and maximum densities was available for this study.
The test matrix that was used for the actual testing is shown in Table 10.  Table 1). The cyclic period was set to 10 s. The interface consisted of a sand -smooth steel surface. Shear tests performed at low normal stresses dilate the most. In the results presented in Figure 87, the dilation led to an increase in the normal stress and thus in the shear stress. Due to this it is difficult to choose a displacement amplitude for the cyclic shear testing. Hence it was decided to run a monotonic CNL test. The results are shown in Figure 88.
Based on these results displacement amplitudes of +/-0.3 mm and +/-0.6 mm were chosen for the strain-controlled cyclic testing. It was of interest for these tests choosing displacement amplitudes that were, based on the monotonic tests, before and after reaching the peak shear stresses. For the stress-controlled cyclic testing was chosen a stress amplitude of 5 kPa.    It was of interest for these tests choosing displacement amplitudes that were, based on the monotonic tests, before and after reaching the peak shear stresses.
The cyclic results are shown in Figure 93 through Figure 95.  Based on the results of these tests, the following conclusions can be made: • Dilation causes an increase in normal stress and contraction causes a reduction in the normal stress.
• In both monotonic and cyclic tests, the value of the constant normal stiffness plays a significant role in the overall behavior. Higher stiffness causes the changes in the normal stress to be faster. This is especially significant for contractive sands, where after a few cycles a major reduction in the normal stress occurs.
• In cyclic testing, the displacement amplitude has a significant effect in the reduction of the normal stress. When the target value of !"# is known the next step is to choose a bar with a similar value of stiffness. As part of this thesis, a series of different cross arms of varying bending stiffness were produced.
If the stiffness of a bar is unknown, it can be determined using the direct shear apparatus.
The first step is to take out the shear box water bath and replace it with a solid piece of metal that will not deform.
Place on top of the metal piece the metal cap with the concavity in the middle that fits the steel ball. The set-up should look like in Figure 100.  Set the horizontal LVDT on the top of one of the threaded rods of the loading frame. You can use for this a connector like the one shown in Figure 101. After doing so the direct shear apparatus should look like in Figure 102.
The next step is to run consolidation tests. There should be run for multiple steps with increasing normal stress. Between each step should be let the normal stress stabilize before starting the next one.
The amount of bending in the bar will be the value of the horizontal strain gauge minus the value of the vertical strain gauge (Use only the absolute values).
An example of this in an excel worksheet is shown in Figure 103.
The last step is to plot the normal stress or load with the bending of the bar ( Figure 104). The trend of the curve is the bending stiffness of the bar. The bending stiffness of the bar is the constant normal stiffness when using this specific bar.   If a specific !"# is desired but the required height of the bar is unknown, Figure 105 can be used. The 2 lines represent the bending stiffness of a bar with increasing height for a simply supported and completely fixed system. The green point represents the stiffness of a plate 6.9 mm thick. The yellow point represents the stiffness of a plate 10 mm thick.

Heigth [mm]
Simply supported Fixed 6.9 mm Bar 10 mm Plate The nuts that fix the new bar to the threaded rod of the loading frame should be, if possible, tightened to 10 lb. ft or 13.55 Nm.
In the software of ShearTrac-II can now be set the desired value of initial normal stress for the consolidation phase.
Begin the consolidation phase. When the normal stress stabilizes the 4 lifting screws must be put in contact the lower half of the box and be turned 180 degrees. This will create a gap between the two halves of the shear box in order to prevent any additional friction. Now the 2 plastic bolts that hold the direct shear box together can be removed and subsequently the 4 lifting screws need to be unscrewed.
Before the shearing phase is started, the vertical load frame must be locked by setting the velocity limit to 0. This can be done under Options/PID/Vertical Load (see Figure 106). Set the value of the velocity limit to 0, click "apply" and "Ok".  Cyclic CNS tests must be performed in the cyclic simple shear apparatus and the equipment must be set up as shown in Figure 107. In its original form, was this apparatus conceived for cyclic simple shear tests. For running cyclic direct shear tests, modifications had to be done. The first modification was attaching an angle on the top of the base of the steel arm. At the end of this angle is attached a metal rod ( Figure 108). The function of this metal rod is to prevent the upper half of the shear box from moving horizontally. If the apparatus looks like that in Figure 107, can be followed the subsequent step by step instructions for performing a CNS cyclic direct shear test.
Step by Step Step by Step Step by Step Step

by Step Instructions Instructions Instructions Instructions
Replace the original black plate provided by GEOCOMP with a plate with the bending stiffness of interest. Raise the metal arm. Make sure that the angle is attached to the base of the metal arm and that the metal rod is set up like in Figure   109.  In cyclic testing, it is important that the lower half of the shear box is tightly fixed. For doing so, use the spacers as shown in Figure 111, and firmly tighten the lower half shear box. Open the Shear-Trac software CDSS. If a warning indicates that there is no connection between the equipment and the computer attached to it, restart the computer while the shear apparatus is turned on (restart; do not turn off and on).
For cyclic testing, it is very important to set in the software an accurate value of the sample height. This due to that the amount of shear displacement is dependent on the value of the sample height. This is done in the specimen table, shown in Figure 112. It is also important to input the value of the diameter of the shear box.
The cyclic period of the loading can also be input.

Figure 113 -Cyclic table
It is recommended to input the rest of the settings in the cyclic table as they are in Figure 113. If tests are run with low strain amplitudes, the desired response gain can be increased to 6 or 7. This will make the strain amplitude curve during testing more accurate.
In the consolidation table, set the value of the normal stress. Note that the rod of the loading frame is not attached to the load cell, thus is this weight acting on the soil probe. This amounts to roughly 2 kPa.
For running CNS tests, the cyclic table, the "Normal Control" must be set to no control. This will prevent the loading frame from moving during shearing.
Make sure that all screws are tight before starting. The test can now be started. After the consolidation phase is concluded the two bolts holding the two halves of the shear box together must be removed.
Begin the cyclic shear phase. After the test is concluded the results can be reviewed under the report tab ( Figure 114). If you want to extract the results of an excel table, you can click while viewing the results on view/export to export the data to an excel file.  The rubber tests also lead to the conclusion that the jumps are caused by the velocity regulating mechanism of the cyclic shear apparatus and not the shear box.   Table 11 shows the Test Matrix for the tests performed on the Alan Harbor Sand.
The sand samples were chosen to be the closest to the targeted depths.