ADVANCEMENTS IN STATNAMIC DATA REGRESSION TECHNIQUES
Gray Mullins1, Christopher L. Lewis2, Michael D. Justason3
ABSTRACT
Until recently, the analysis of Statnamic test data has typically incorporated the “Unloading
Point Method” (Middendorp et. al, 1992) to determine an equivalent static capacity. The UP method
requires that the foundation move as a rigid body, thus excluding stress wave phenomenon from the
analysis. If this requirement is met the foundation capacity can be determined using this simplified
method. However, many foundations do not meet the UP criteria (e.g. fixed end or relatively long
piles) and have proven difficult to analyze without more complex techniques. This paper presents
a new analysis method that uses measured strain data as well as the standard Statnamic test data to
determine foundation capacity. This new method discretizes the foundation into smaller segments
that each meet the rigid body criteria of the UP method. Thereby, a more refined inertia and viscous
damping evaluation can be implemented that individually determines the contributions from the
various segments. This approach, termed the “Segmental Unloading Point” (SUP) method, is
developed herein and then demonstrated with results from full-scale Statnamic test data.
INTRODUCTION
Since its inception in1988, Statnamic testing of deep foundations has gained popularity with
many designers largely due to its time efficiency, cost effectiveness, data quality, and flexibility in
testing existing foundations. Where large capacity static tests may take up to a week to set up and
conduct, the largest of Statnamic tests typically takes no more than a few days. Further, multiple
smaller-capacity tests can easily be completed within a day. The direct benefit of this time efficiency
is the cost savings to the client and the ability to conduct more tests within a given budget.
Additionally, this test method has boosted quality assurance by giving the contractor the ability to
test foundations thought to have been compromised by construction difficulties without significantly
affecting production.
1
Geotechnical Engineer, Applied Foundation Testing, Green Cove Springs / Tampa, FL.
2
Assistant Professor, University of South Florida, Department of Civil and Environmental
Engineering, Tampa, FL.
3
Senior Engineer, Berminghammer Foundation Equipment, Hamilton, ON.
Statnamic testing is designated as a rapid load test that uses the inertia of a relatively small
reaction mass instead of a reaction structure to produce large forces. Rapid load tests are
differentiated from static and dynamic load tests by comparing the duration of the loading event with
respect to the axial natural period of the foundation (2L/C). Test durations longer than 1000 L/C are
considered static loadings and those shorter than 10 L/C are consider dynamic, where L represents
the foundation length and C represents the strain wave velocity (Janes et al., 2000; Kusakabe et.al,
2000). Tests with a duration between 10L/C and 1000 L/C are denoted as rapid load tests. The
duration of the Statnamic test is typically 100 to 120 milliseconds, but is dependant on the ratio of
the applied force to the weight of the reaction mass. Longer duration tests of up to 500 milliseconds
are possible but require a larger reaction mass.
The Statnamic force is produced by quickly-formed high pressure gases that in turn launch
a reaction mass upward at up to twenty times the acceleration of gravity. The equal and opposite
force exerted on the foundation is simply the product of the mass and acceleration of the reaction
mass. It should be noted that the acceleration of the reaction mass is not significant in the analysis
of the foundation; it is simply a by-product of the test. Secondly, the load produced is not an impact
in that the mass is in contact prior to the test. Further, the test is over long before the masses reach
the top of their flight. The parameters of interest are only those associated with the movement of the
foundation (i.e. force, displacement, and acceleration).
Typical analysis of Statnamic data relies on measured values of force, displacement and
acceleration. A soil model is not required, hence, the results are not highly user dependent. A new
method of analysis is introduced that extends present methods by incorporating additional measured
values of strain at discrete points along the length of the foundation. In the ensuing sections a
discussion of analysis methods and their applicability will be presented. Full details on the
development of this method can be found elsewhere (Lewis, 1999).
PRESENT ANALYSIS PROCEDURES
The Statnamic forcing event induces foundation motion in a relatively short period of time
and hence acceleration and velocities will be present. The accelerations are typically small(1-2 g’s),
however the enormous mass of the foundation when accelerated resists movement due to inertia and
as such the fundamental equation of motion applies, Equation 1.
Equation 1
where, F = forcing event
m = mass of the foundation
a = acceleration of the displacing body
v = velocity of the displacing body
c = viscous damping coefficient
k = spring constant of the displacing system
x = displacement of the body
2
The equation of motion is generally described using four terms: forcing, inertial, viscous
damping, and stiffness. The forcing term (F) denotes the load application which varies with time
and is equated to the sum of remaining three terms. The inertial term (ma) is the force which is
generated from the tendency of a body to resist motion, or to keep moving once it is set in motion
(Young,1992). The viscous damping term (cv) is best described as the velocity dependant resistance
to movement. The final term (kx), represents the classic system stiffness, which is the static soil
resistance.
When this equation is applied to a pile/soil system the terms can be redefined to more
accurately describe the system. This is done by including both measured and calculated terms. The
revised equation is displayed below:
Equation 2
where, FStatnamic is the measured Statnamic force, m is the calculated mass of the foundation, a is the
measured acceleration of the foundation, c is the viscous damping coefficient , v is the calculated
velocity, and Fstatic is the derived pile/soil static response.
There are two unknowns in the revised equation Fstatic and c, thus the equation is under
specified. Fstatic is the desired value, so the variable c must be obtained to solve the equation.
Middendorp (1992) presented a method to calculate the damping coefficient referred to as the
Unloading Point Method (UP). With the value of c known, the static force can be calculated. This
force, termed “Derived Static,” represents an equivalent soil response to that produced by a
traditional static load test.
UP DESCRIPTION
The UP is a simple method which allows the equivalent static resistance to be derived from
the measured Statnamic quantities. It uses a simple single degree of freedom model to represent the
foundation/soil system as a rigid body supported by a non-linear spring and a linear dashpot in
parallel (see Figure 1). The spring represents the static soil response (FStatic) which includes the
elastic response of the foundation as well as the foundation/soil interface and surrounding soil
response. The dashpot is used to represent the dynamic resistance which depends on the rate of pile
penetration (Nishimura, 1995).
The UP makes two primary assumptions in its determination of “c.” The first is the static
capacity of the pile is constant when it plunges as a rigid body. The second is that the damping
coefficient is constant throughout the test. By doing so a window is defined in which to calculate
the damping coefficient. The first point of interest (1) is that of maximum Statnamic Force. At this
point the static resistance is assumed to have become steady state, for the purpose of calculating “c”.
Thus, any extra resistance is attributed to that of the dynamic forces (ma and cv). The next point of
interest (2) is that of zero velocity which has been termed the “Unloading Point.” Figure 2 shows
a typical Statnamic load-displacement curve which denotes points (1) and (2). At this point the
3
foundation is no longer moving and the resistance due to damping is zero. The static resistance, used
to calculate “c” from point (1) to (2), can then be calculated by the following equation:
Equation 3
where, F Statnamic, m, and a are all known parameters; F Static U P is the static force calculated at (2) and
assumed constant from (1) to (2).
Figure 1 Single D.O.F. Model (After Das 1994)
Figure 2 UP time window for C determination.
4
Next, the damping coefficient can be calculated throughout this range, from maximum force (1) to
zero velocity (2). The following equation is used to calculate c:
Equation 4
Damping values over this range should be fairly constant. Often the average value is taken as the
damping constant, but if a constant value occurs over a long period of time it should be used (see
Figure 3). Note that as v approaches zero at point (2), values of c can be different from that of the
most representative value and therefore the entire trend should be reviewed. Finally the derived
static response can be calculated as follows:
Equation 5
Figure 3 Variation in C between times (1) and (2).
Currently software is available to the public that can be used in conjunction with Statnamic
test data to calculate the derived static pile capacity using the UP Method (Garbin, 1999). This
software was developed by the University of South Florida and the Federal Highway Administration
and can be downloaded from www.eng.usf.edu/~gmullins under the Statnamic Analysis Workbook
(SAW™) heading.
UP SHORTCOMINGS
The UP has proven to be a valuable tool in predicting damping values when the foundation
acts as a rigid body. However, as the pile length increases an appreciable delay can be introduced
between the movement of the pile top and toe, hence negating the rigid body assumption. This
5
occurrence also becomes prevalent when an end bearing condition exists; in this case the lower
portion of the foundation is prevented from moving jointly with the top of the foundation.
Middendorp (1995) defines the “Wave Number” (Nw) to quantify the applicability of the UP.
The wave number is calculated by dividing the wave length (D) by the foundation depth (L). D is
obtained by multiplying the wave speed c in length per second by the load duration (T) in seconds.
Thus, the wave number is calculated by the following equation:
Equation 6
Through empirical studies Middendorp determined that the UP would accurately predict
static capacity, from Statnamic data, if the wave number is greater than 12. Nishimura (1995)
established a similar threashold at a wave number of 10. Using wave speeds of 5000 m/s and
4000m/s for steel and concrete respectively and a typical Statnamic load duration, the UP is limited
to piles shorter than 50 m (steel) and 40 m (concrete). Wave number analysis can be used to
determine if stress waves will develop in the pile. However, this does not necessarily satisfy the
rigid body requirement of the UP.
Statnamic tests cannot always produce wave numbers greater than 10, and as such there have
been several methods suggested to accommodate stress wave phenomena in Statnamically tested
long piles (Middendorp, 1995). Due to limitations on paper length these methods are not presented.
MODIFIED UNLOADING POINT METHOD
Given the shortcomings of the UP, users of Statnamic testing have developed a remedy for
the problematic condition that arises most commonly. The scenario involves relatively short piles
(Nw>10) that do not exhibit rigid body motion, but rather elastically shorten within the same
magnitude as the permanent set. This is typical of rock-socketed drilled shafts or piles driven to
dense bearing strata that are not fully mobilized during testing. The consequence is that the top of
pile response (i.e. acceleration, velocity, and displacement) is significantly different from that of the
toe. The most drastic subset of these test results show zero movement at the toe while the top of pile
elastically displaces in excess of the surficial yield limit (e.g. upwards of 25 mm). Whereas with
plunging piles (rigid body motion) the difference in movement (top to toe) is minimal and the
average acceleration is essentially the same as the top of pile acceleration; tip restrained piles will
exhibit an inertial term that is twice as large when using top of pile movement measurements to
represent the entire pile.
The Modified Unloading Point Method (MUP), developed by Justason (1997), makes use
of an additional toe accelerometer that measures the toe response. The entire pile is still assumed
to be a single mass, m, but the acceleration of the mass is now defined by the average of the top and
toe movements. A standard UP is then conducted using the applied top of pile Statnamic force and
the average accelerations and velocities. The derived static force is then plotted against top of pile
displacement as before. This simple extension of the UP has successfully overcome most
problematic data sets. Plunging piles instrumented with both top and toe accelerometers have shown
6
little analytical difference between the UP and the MUP. However, MUP analyses are now
recommended whenever both top and toe information is available.
NEED FOR ADVANCEMENT
Although the MUP provided a more refined approach to some of the problems associated
with UP conditions, there still exists a scenario where it is difficult to interpret Statnamic data with
present methods. This is when the wave number is less than ten (relatively long piles). In these
cases the pile may still only experience compression (no tension waves) but the delay between top
and toe movements causes a phase lag. Hence an average of top and toe movements does not
adequately represent the pile.
SUP METHOD DESCRIPTION AND GENERAL PROCEDURE
The fundamental concept of the SUP is that the acceleration, velocity, displacement, and
force on each segment can be determined using strain gage measurements along the length of the
pile. Individual pile segment displacements are determined using the relative displacement as
calculated from strain gage measurements and an upper or lower measured displacement. The
velocity and acceleration of each segment are then determined by numerically differentiating
displacement then velocity with respect to time. The segmental forces are determined by calculating
the difference in force from two strain gage levels.
Typically the maximum number of segments is dependent on the available number of strain
gage layers. However, strain gage placement does not necessitate assignment of segmental
boundaries; as long as the wave number of a given segment is greater than 10, the segment can
include several strain gage levels within its boundaries. The number and the elevation of strain gage
levels are usually determined based on soil stratification; as such, it can be useful to conduct an
individual segmental analysis to produce the shear strength parameters for each soil strata. A
reasonable upper limit on the number of segments should be adopted because of the large number
of mathematical computations required to complete each analysis. Figure 4 is a sketch of the SUP
pile discretization.
The notation used for the general SUP case defines the pile as having m levels of strain gages
and m+1 segments. Strain gage locations are labeled using positive integers starting from 1 and
continuing through m. The first gage level below the top of the foundation is denoted as GL1 where
the superscript defines the gage level. Although there are no strain gages at the top of foundation,this
elevation is denoted as GL0. Segments are numbered using positive integers from 1 to m+1, where
segment 1 is bounded by the top of foundation (GL0) and GL1. Any general segment is denoted as
segment n and lies between GLn-1 and GLn. Finally, the bottom segment is denoted as segment m+1
and lies between GLm and the foundation toe.
7
Figure 4 Segmental Free Body Diagram
CALCULATION OF SEGMENTAL MOTION PARAMETERS
The SUP analysis defines average acceleration, velocity, and displacement traces that are
specific to each segment. In doing so, strain measurements from the top and bottom of each segment
and a boundary displacement are required. Boundary displacement may come from the Statnamic
laser reference system (top), top of pile acceleration data, or from embedded toe accelerometer data.
The displacement is calculated at each gage level using the change in recorded strain with
respect to an initial time zero using Equation 7. Because a linearly-varying strain distribution is
assumed between gage levels, the average strain is used to calculate the elastic shortening in each
segment.
Level displacements
Equation 7
where
x n = the displacement at the nth gage level
), average seg n = the average change in strain in segment n
L seg n = the length of the nth segment
To perform an unloading point analysis, only the top-of-segment motion needs to be defined.
However, the MUP analysis, which is now recommended, requires both top and bottom parameters.
The SUP lends itself naturally to providing this information. Therefore, the average segment
movement is used rather than the top-of-segment; hence, the SUP actually performs multiple MUP
analyses rather than standard UP. The segmental displacement is then determined using the average
of the gage level displacements from each end of the segment as shown in the following equation:
Equation 8
where x seg n is the average displacement consistent with that of the segment centroid.
8
The velocity and acceleration, as required for MUP, are then determined from the average
displacement trace through numerical differentiation using Equations 9 and 10, respectively:
Equation 9
Equation 10
where vn = the velocity of segment n
an = the acceleration of segment n
)t = the time step from time t to t+1
It should be noted that all measured values of laser displacement, strain, and force are time
dependent parameters that are field recorded using high speed data acquisition computers. Hence the
time step, )t, used to calculate velocity and acceleration is a uniform value that can be as small as
0.0002 seconds. Therefore, some consideration should be given when selecting the time step to be
used for numerical differentiation.
The average motion parameters (x, v, and a) for segment m+1 can not be ascertained from
measured data, but the displacement at GLm can be differentiated directly providing the velocity and
acceleration. Therefore, the toe segment is evaluated using the standard UP. These segments
typically are extremely short (1 - 2 m) producing little to no differential movement along its length.
CALCULATION OF SEGMENTAL STATNAMIC AND DERIVED STATIC FORCES
Each segment in the shaft is subjected to a forcing event which causes movement and
reaction forces. This segmental force is calculated by subtracting the force at the top of the segment
from the force at the bottom. The difference is due to side friction, inertia, and damping for all
segments except the bottom segment. This segment has only one forcing function from GLm and the
side friction is coupled with the tip bearing component. The force on segment n is defined as:
Equation 11
where Sn = the applied segment force from strain measurements
En = the composite elastic modulus at level n
An = the cross sectional area at level n
,n = the measured strain at level n
Once the motion and forces are defined along the length of the pile, an unloading point
analysis on each segment is conducted. The segment force defined above is now used in place of
the Statnamic force in Equation 2. Equation 12 redefines the fundamental equation of motion for
a segment analysis:
Equation 12
9
where, S n Static = the derive static response of segment n
mn = the calculated mass of segment n
cn = the damping constant of segment n
The damping constant ( in Equation 13) and the derived static response (Equation 14) of the segment
are computed consistent with standard UP analyses:
Equation 13
Equation 14
Finally the top-of-foundation derived static response can be calculated by summing the
derived static response of the individual segments as displayed in the following equation:
Equation 15
Software capable of performing SUP analyses (SUPERSAW™) is currently being developed
at the University of South Florida in cooperation with the Federal Highway Administration.
SITE CHARACTERISTICS AND SUP APPLICATION
Prepared in this section are examples of the motion parameters, segment forces, and load
displacement trends as analyzed by SUP. The foundation was instrumented with four strain gage
levels (m = 4) which produced five segments. Data was obtained at the 17th Street Causeway
Replacement Bridge project as part of an extensive load test program implemented by the Florida
Department of Transportation (FDOT), which included Statnamic load tests. Statnamic load testing
was performed using a 30MN Statnamic device equipped with a gravel catch structure. Shaft
instrumentation consisted of standard Statnamic equipment as well as, resistive type strain gages,
and a toe accelerometer. Instrumentation elevations are presented in Table 1.
The test shaft had a planned diameter of 1.22 m and was 22 m in length. It was constructed
using a temporary casing method and sea water as the drilling fluid. The 1.22 m O.D. steel casing
(1 cm wall thickness) was installed to elevation-18.96 using a vibratory hammer. The concrete was
placed using a tremie method, then the casing tip was pulled to elevation -0.9 m, using a vibratory
hammer.
A soil boring performed at the test shaft location indicated that the natural ground elevation
was approximately 1.5 m. The water table was reported to exist at elevation 0.3 m. SPT testing was
initiated at the ground surface (elevation +3 m) and extended to elevation -28.15 m. The upper two
meters of soil consisted of compacted limestone fill with SPT “N” values ranging from 27 to 16.
10
Table 1 Instrumentation Schedule
Instrumentation Elevation
(m) Number of Transducers Type of Transducer
Calibrated Load Cell, 2
3.0 4 Accelerometers, and
Laser Reference System
-1.8 3 Strain Gage
-4.2 3 Strain Gage
-17.0 3 Strain Gage
-18.3 3 Strain Gage
-19.0 1 Accelerometer
The following strata was reported as fine sand with fragments of limestone and shell. This strata
extended to elevation -14.7m, “N” values ranged from 9 to 57. From elevation -14.7m throughout,
the rock socket length averaged 34% RQD at 80% recovery. RQD values ranged from 18% to 73%
in the limestone below the shaft tip. Recovery values in this strata are generally greater than 70%.
Figure 5 shows the measured change in strain with respect to time t = 0 for each gage level.
Figures 6 through 8 illustrate the motion parameters determined for each of the five segments.
Figure 9 shows the forces calculated at each gage level with the true measured strain. Figure 10
shows the dynamic forces on each of the segments as calculated by Equation 11.
Figure 5 Strain versus Time
11
Figure 6 Segmental Average Displacements
Figure 7 Segmental Average Velocity
12
Figure 9 Segmental Average Acceleration
Figure 8 Force at Gage Levels
13
Figure 10 Segmental Statnamic Forces
Figure 11 Segment Load Displacement Curves
14
Using the segmental forces, the derived static soil resistance is determined for each of the
segments. Figure 11 shows a typical segment load versus segment displacement curve. By simply
dividing the segment force by circumferential surface area this curve can be converted in to a shear
stress versus displacement (T-Z) curve for that specific soil strata.
It can be seen in Figure 10 that the peak of each segment force may occur at different times
and therefore at different top-of-pile displacements. This can effect the pile capacity in that the
ultimate shear strengths of the strata should not be simply summed. This is most probably not a
significant concern with this pile due to the plunging nature of the failure. This is evidenced by the
similar top and toe movements shown in Figure 6. However, to be technically correct, SUP uses the
summation of segment forces as they were developed. This accounts for upper soil layers that may
fully mobilize and become residual in nature while lower soil layers begin to develop ultimate
strengths. Figure 12 shows the raw Statnamic load displacement curve as well as SUP, MUP, and
UP derived static capacity versus the top-of-pile displacement.
Figure 12 Top of Pile Load versus Displacement
15
SUMMARY
A new method of analysis called the Segmental Unloading Point Method (SUP) was
presented that evaluates Statnamically loaded foundations as segments whose lengths are defined
by embedded strain gage elevations. The recorded strain measurements are used to determine both
the segmental motion parameters as well as the segmental force traces. Each segment is then treated
as an individual foundation whose static response is derived using either the UP or MUP methods.
The summation of each segment contribution with respect to time provides a top of foundation
response that more closely incorporates the actual distribution of inertial and damping forces
throughout the foundation. This is most important in the analysis of relatively long or fixed-ended
piles. Although the UP and MUP methods of analysis are sufficient for most loading conditions,
SUP provides information for soil strata T-Z curves as well as cut-off elevation load-displacement
curves.
REFERENCES
Bermingham, P., and White, J., (1995), “Pyrotechnics and the Accurate of Prediction of Statnamic
Peak Loading and Fuel Charge Size”, First International Statnamic Seminar, 1995, Vancouver,
British Columbia Canada
Das, Braja M., (1993), “Principles of Soil Dynamics”, PWS-KENT Publishing Company, Boston,
Massachusetts.
Garbin, E. J., (1999), “Data Interpretation for Axial Statnamic Testing and the Development of the
Statnamic Analysis Workbook,” Master’s Thesis, University of South Florida, Tampa, FL.
Kusakabe, Kuwabara, and Matsumoto (eds), (2000), “Statnamic Load Test,” Draft of ‘method for
rapid load test of single piles (JGS 1815-2000),’ Proceedings of the Second International Statnamic
Seminar, Tokyo, October 1998 pp. 237-242.
Janes, M.C., Justason, M.D., Brown, D.A., (2000), “Long period dynamic load testing ASTM
standard draft,” Proceedings of the Second International Statnamic Seminar, Tokyo, October, 1998,
pp. 199-218.
Justason, M.D., (1997), “Report of Load Testing at the Taipei Municipal Incinerator Expansion
Project,” Taipei City, Taipei.
Lewis, C.L., (1999), “Analysis of Axial Statnamic Testing by the Segmental Unloading Point
Method,” Master’s Thesis, University of South Florida, Tampa, FL.
Middendorp, P., Bermingham, P., and Kuiper, B. , (1992). “Statnamic Load Testing Of Foundation
Pile.” Proceedings, 4th International Conference On Application Of Stress-Wave Theory To Piles,
The Hague, pp. 581-588.
Middendorp, P. and Bielefeld, M.W., (1995), “Statnamic Load Testing and the Infuluence of Stress
16
Wave Phenomena”, Proceedings of the First International Statnamic Seminar, Vancouver, Canada,
pp. 207-220.
Nishimura, S., Matsumoto, T., (1998), “Wave Propagation Analysis During Statnamic Loading of
a Steel Pipe Pile”, Second International Statnamic Seminar, 1998, Canadian Embassy of Japan,
Tokyo
Young, Hugh. D., (1992), “University Physics”, Eight Edition, Addison Wesley.
17
