GIN vp sine bar

Document Sample
GIN vp sine bar Powered By Docstoc
					                                                                                     GEOTECHNICAL INSTRUMENTATION NEWS

                                             Geotechnical Instrumentation News

                                             John Dunnicliff

Introduction                                     tilt sensors for the last two years. In-       I’d been hoping for case history in-
This is the fifty-fourth episode of GIN.         struments have included tiltmeters,        formation on MEMS-based inclinom-
A meaty one this time, all about MEMS.           and probe inclinometers for opera-         eters, in-place inclinometers and
                                                 tion vertically, horizontally, and in      tiltmeters, but nothing came out of the
MEMS                                             the sloping faces of concrete-faced        woodwork. Perhaps in the future—yes
In the June 2007 episode of GIN I                rock-fill dams. One interesting de-        please! I’d also welcome discussions of
wrote:                                           velopment has been the manufacture         these four articles. If anyone is inter-
   Micro-electro-mechanical systems              of addressable in-place inclinom-          ested in submitting a discussion, will
   (MEMS) appear to have potential               eters. At a construction site in           you please let me know as soon as
   applications in our business. Perhaps         Boston, twenty MEMS tiltmeters             possible, so that I can plan ahead?
   as tiltmeters. Perhaps as sensors in          have been installed in a borehole,         G u id elin es fo r articles are o n
   inclinometers. Perhaps as sensors             connected together by a single cable. Discussions should fol-
   in-place inclinometers, both for hor-         Each of the twenty MEMS tiltmeters         low the same guidelines, but should
   izontal and vertical deformation              is automatically addressed by a            have a maximum length of 1½ pages in
   monitoring.                                   datalogger.                                the magazine format. One particular is-
   When I wrote that, I knew nothing         • Since 2003, Soil Instruments Ltd.            sue for possible discussion—careful
about MEMS! But now, having twisted              ( have been manu-           readers will note some differences in
the arms of colleagues and done some             facturing probe inclinometers, por-        view about stability with time, and it
heavy interaction and editing, I know            tab le     tiltmeters,       in -p lace    would be good to have more views on
enough to be dangerous.                          inclinometers, track twist sensors         that.
   First, here’s what I found out about          and tiltmeters with MEMS sensors.
‘who is making what?’:                           A primary application for these            International Symposium on
• RS T       I n s tr u m e n ts ( w w w.        tiltmeters has been the monitoring of      Field Measurements in has been ac-                                                         Geomechanics (FMGM),
                                                 live railroad tracks for distortion
                                                                                            September 2007
   tively engaged in MEMS inclinome-             during adjacent work such as piling
   ter product development for three                                                        The 7th International Symposium on
                                                 or tunneling.
   years, with current MEMS products                                                        Field Measurements in Geomechanics
                                                 Perhaps other manufacturers are
   including digital inclinometer                                                           (FMGM) was held in Boston, MA dur-
                                             making MEMS-based instruments, and
   probes, in-place inclinometers, un-                                                      ing September 2007. The proceedings
                                             if they read this I’m sure they will tell
   derwater tiltmeters, and tilt beams.                                                     are available on CD from ASCE, and
                                             me, so that I can share with you next
   Upcoming MEMS products include                                                           c a n b e o r d e r e d o n - lin e at
   portable tiltmeters, horizontal digital                                                  h t t p s : / / w w w . a s c e . o r g / b o o k-
                                                 Here are four articles. The first one
   inclinometers, wireless devices, and                                                     store/book.cfm?book=7841. The list
                                             (Sellers and Taylor) gives some basics.
   24 bit digital bus in-place devices,                                                     price is $125 and member price is
                                             The second one (Sheahan et al) reports
   which support up to 128 points on a                                                      $93.75.
                                             on some performance testing. The third
   single 4-conductor cable.                                                                    There are some very worthwhile pa-
                                             (Abdoun and Bennett) and fourth
• Geokon ( has
                                                                                            pers in the proceedings. The one that
                                             (Barendse) tell us about one commer-
   been manufacturing MEMS-based                                                            impressed me most was “The Use of the
                                             cial product.
                                                                                            Fully-grouted Method for Piezometer

                                                                                                         Geotechnical News,   March 2008   31

 Installation” by Contreras, Grosser and       2007 Keynote Lecture and eight Theme         Here’s tae us, wha’s like us? Damn
 Ver Strate of Barr Engineering Com-           Lectures are posted on       few, and they’re a’ deid, mair’s the pity
 pany in Minneapolis. It includes com-            The next FMGM symposium will be        (Scotland).
 prehensive sections on the theoretical        held in Germany in 2011. Watch this          For those of you who have joined the
 background, materials, installation pro-      space!                                    GIN community recently, perhaps a few
 cedure, criteria for grout, lab testing of                                              words of explanation are needed here.
 grout, computer modeling to support           Closure                                   When I started writing these ‘columns’
 the applicability of the method, and          Please send contributions to this col-    in 1994, Birger Schmidt gave me a beer
 field verification. What more could we        umn, or an article for GIN, to me as an   mat with about 15 drinking toasts in dif-
 want?! I plan to re-publish this paper,       email attachment in MSWord, to            ferent languages, and I used them in
 and a few others selected from the  , or by      turn to close each episode. I decided to
 FMGM proceedings, in future episodes          fax or mail: Little Leat, Whisselwell,    keep the tradition going.
 of GIN.                                       Bovey Tracey, Devon TQ13 9LA, Eng-
     PowerPoint versions of the FMGM           land. Tel. and fax +44-1626-832919.

                                               MEMS Basics

                                               J. Barrie Sellers
                                               Robert Taylor

 What are They?                                   Figures 1 and 2 show commercial        ure-suspended proof mass which is de-
 An Internet search revealed: “Mi-             examples of tiltmeters. Note that the     flected as the component of gravity
 cro-Electro-Mechanical Systems are            small graduations on the scales are in    changes with tilt angle. A position sen-
 the integration of mechanical elements,       millimeters, the numbers in centi-        sor senses this movement by differential
 sensors, actuators and electronics on a       metres.                                   capacitance sensors of exceptional sen-
 common silicon substrate through                                                        sitivity.
 microfabrication technology”. There           How do They Work?                             The proof mass, flexible mounting,
 are numerous of types of MEMS, a              Like the accelerometer that has been      position sensor, and supporting elec-
 large and rapidly growing high technol-       used in inclinometers since about 1970,   tronics are all constructed from a single
 ogy area. The first to appear in the          a MEMS tilt sensor is based on a flex-    wafer of non-metallic material, usually
 geotechnical field are accelerometers                                                   silicon.
 (similar in principle to the sensors used
 in most inclinometers), which are being                                                 How are They Read?
 used as tilt sensors.                                                                   Analog measurement is typical: a DC
     The initial mass market for MEMS                                                    supply is applied and a signal propor-
 accelerometers was for automotive                                                       tional to acceleration is returned. Be-
 airbags, which typically have ranges ex-                                                cause of the long cable lengths that fre-
 ceeding +/-30 g. Sensors built for airbag                                               quently are required in geotechnical
 systems typically have noise and drift                                                  applications, some additional electron-
 much worse than older tilt technologies,                                                ics are typically included at the sensor:
 making them not very useful for                                                         such as a precision voltage regulator,
 geotechnical applications.                                                              and line-driving amplifier(s). They are
     Recently, MEMS sensors have be-                                                     read with the same type of readout that
 come available with full scale ranges of 1                                              is used for servo accelerometer type tilt
 g or less, and with drift and resolution                                                sensors. They may also be data-logged,
 specifications equal to or better than pre-                                             for example using a Campbell Scien-
 vious technologies. These have sparked                                                  tific CR-1000 Measurement and Con-
 considerable interest among geotechnical                                                trol System. To enhance resolution and
 instrument manufacturers for applica-                                                   noise rejection, the logger can make a
 tions including in-place inclinometers,       Figure 1. Two MEMS Sensors Mounted        number of readings and take the aver-
 tiltmeters, tilt beams, probe inclinometers   90 Degrees Apart to Create a Geokon       age. This allows resolutions of about 3
 and strong-motion accelerographs.             Biaxial Tiltmeter.                        to 5 arc-seconds to be attained.

32   Geotechnical News,   March 2008
                                                                                    GEOTECHNICAL INSTRUMENTATION NEWS

                                             g shock. They have low drift and ther-       require more conductors in the cable
                                             mal coefficients, about 1 arc-second per     and therefore more complicated
                                             degree centigrade. The intrinsic linear-     multiplexing.
                                             ity of MEMS accelerometers is very
                                             good, due to the minute deflections of       Summary
                                             the proof mass. They are stable, sensi-      At this time we believe that MEMS sen-
                                             tive and accurate, and can be used with      sors potentially have significant advan-
                                             cables up to about 500 meters. The sen-      tages over alternative tilt sensing tech-
Figure 2. RST Instruments’ Biaxial           sors themselves are small, about 13 mm       nologies, but time and experience will
MEMS Tiltmeter Internal Assembly,                                                         let us know more about performance. It
                                             in one case, permitting the packaging to
Including Two MEMS Sensors and                                                            is expected that as MEMS accelerome-
Supporting Electronics on a Single           be correspondingly small and light. The
                                             power consumption is low, typically 20       ters continue to improve, they will oc-
Circuit Board.
                                             mW, which is an advantage both in bat-       cupy an ever-increasing portion of the
                                             tery life and in warm-up time.               geotechnical tilt market.
How Many Conductors are
Required in the Cable?                       What is Bad about Them?                      Barrie Sellers, President, Geokon Inc.,
The uniaxial sensor requires four con-       As far as we can tell, not much. They are    48 Spencer Street, Lebanon, NH 03766,
ductors: positive power, ground, signal      slightly less sensitive than servo           USA, Tel. (603) 448-1562, email:
out and signal ground. The biaxial sen-      accelerometers and vibrating wire tilt
sor requires six conductors. If a therm-     sensors and the range is limited to +/-15
istor is added, two additional conduc-       degrees. Other ranges are available with     Robert Taylor, President, RST Instru-
tors are needed.                             a corresponding reduction in sensitivity     ments Inc., 200-2050 Hartley Avenue,
                                             and angular accuracy. They are voltage       Coquitlam, BC, Canada, V3K 6W5, Tel.
What is Good About Them?
                                             output devices so cable connections and      (604) 540-1100, email:rtaylor@
They are inexpensive. They have a very       waterproofing are a little more critical
high shock tolerance, very much more         than vibrating wire types. In compari-
so than servo accelerometers. For ex-        son with vibrating wire tilt sensors, they
ample, 0.2 g sensors can survive 20,000

                                             Performance Testing of MEMS-based Tilt

                                             Thomas C. Sheahan
                                             David Mazzei
                                             John McRae

Introduction                                 The need for this testing arose because      sensors were each mounted in a stain-
The introduction of geotechnical instru-     in-place inclinometer specifications for     less steel housing, and these housings
mentation based on Micro-Electri-            a construction project required that the     were fastened to a 15 in. long, stainless
cal-Mechanical Systems (MEMS)                tilt sensors should be tested by a testing   steel platform (so-called “sine bar”)
brings with it the need to prove the abil-   laboratory independent of the manufac-       with two-point support (Figure 1). The
ity of such devices to perform to project    turer and should have the performance        tilt sensor readings were recorded elec-
specifications. In particular, because       characteristics given in Table 1.            tronically using a Geokon datalogger
MEMS-based instrumentation is rela-              The equipment and testing proce-         a n d th e d a talo g g in g s o f twar e
tively new to geotechnical engineering,      dures used, the specifications to be met,    Multilogger Version by Ca-
and because of typically harsh applica-      and the results of the performance test-     nary Systems, Inc. run on a laptop dedi-
tion environments, project owners and        ing compared to those specifications are     cated to the test being performed.
other stakeholders want data that prove      presented. The work was carried out in           This set-up ensures that the six tilt
the performance of these devices over a      th e N o r th eastern U n iversity           sensors and housings are simulta-
variety of conditions and time ranges.       Geotechnical Research Laboratory.            neously inclined at the same angle in the
    This article describes the perfor-                                                    A-A axis (refer to Figure 1 for the axis
mance testing of MEMS-based tilt sen-        Basic Proof-Testing Equipment                definitions). Steel shims, previously
sors for use with in-place inclinometers.    Six identical MEMS-based Geokon tilt         calibrated by Geokon, were used to

                                                                                                     Geotechnical News,   March 2008   33

                Table 1. Outline of Specified Performance Criteria                            that over the range -10° to +10° to the
                                                                                              vertical in the A-A axis direction at a
                   Type of Test                        Performance Criteria                   constant temperature of 20°C (68°F),
                                                                                              the repeatability of readings had to be
     Short-term repeatability at constant       ± 40 arc-seconds or less.                     within 40 arc-seconds or less. This
     temperature                                                                              same tolerance also had to be met or ex-
     Temperature sensitivity                    With temperature increasing from 50°F         c e e d e d w h e n t h e B- B a x i s ( o r
                                                (10°C) to 68°F (20°C) and subse-              cross-axis) was set using the steel plat-
                                                quently decreasing from 68 to 50°F, a         form so that the sensors were 5° to the
                                                maximum indicated reading change              vertical, and the same set of A-A axis
                                                corresponding to 250 arc-seconds.             inclinations were cycled twice.
     90-day zero stability at constant tem-     A maximum deviation corresponding                 The specifications for the first part of
     perature                                   to 50 arc-seconds throughout the entire       this testing (varying A-A axis inclina-
                                                time period.                                  tion and no B-B axis inclination) were
                                                                                              met or exceeded with the exception of
                                                                                              one cycle difference reading for one tilt
 raise the sine bar to desired inclinations,     (0.5°F). Figure 2 shows the six tilt sen-    sensor (a difference of 43.3 arc-sec-
 resulting in the tilt sensors and their cor-    sors in housings mounted on the sine         onds). Because of the consistency of the
 responding housings being inclined              bar and placed in the temperature cham-      other data – all other cycle differences
 from 0° to +10° to the vertical. The tilt       ber. A side access port allows for the ca-   were less than 40 arc-seconds – this
 sensors can also be tested for angles 0°        bles to be connected externally to the       reading appears to be due to operator er-
 to -10° to the vertical by either using the     datalogger.                                  ror. It can reasonably be concluded that
 same shims on the other side of the sine           For all testing, the datalogger re-       the specification is met or exceeded. For
 bar or simply rotating the housings             corded tilt sensor readings in volts, and    the two-axis tests, in which the B-B axis
 180°. For B-B or cross-axis inclinations        then calibration factors (provided by        inclination was set at 5° to the vertical,
 (again, refer to Figure 1), the sine bar        Geokon) were used for converting volt-       there were no readings beyond the spec-
 was placed on a steel plate with                age readings to degrees and arc-seconds      ification threshold.
 three-point support and inclined using          (3600 arc-seconds = 1°).
 another set of calibrated shims pro-                                                         Temperature Sensitivity
 vided, with tilt sensor-housing inclina-        Short-term Repeatability over                Testing at Various Angle
                                                 Two Cycles at Constant                       Combinations
 tions varying from 0° to +5° to the verti-      Temperature                                  This set of performance tests examined
 cal. For temperature testing, the sine          The specification stated that the mini-      the repeatability of readings at positive
 bar-housing assembly and tilt sensors           mum angular range of the tilt sensors        and negative inclination angles and
 were placed in an environmental test
                                                 had to be ±10° to the vertical in the A-A    temperature variations. The sine bar
 chamber. This chamber has a tempera-
                                                 axis direction, and this was verified dur-   and six tilt sensors were placed in the
 ture range from -75 to 200°C (-103 to           ing the course of this repeatability spec-   environmental chamber (Figure 2) and
 392°F) with an accuracy of ±0.3°C               ification. This specification required       the sine bar inclined along the A-A axis

 Figure 1. Six tilt sensors and housings on the sine bar for axis      Figure 2. Six tilt sensors and housings on the sine bar for
 A-A inclination, axis definitions shown.                              axis A-A inclination in the environmental test chamber.

34    Geotechnical News,   March 2008
                                                                                  GEOTECHNICAL INSTRUMENTATION NEWS

 Table 2. Performance Results for 5° A-A Axis Inclination Stability,                    2, as an example, for sensor No. 1, the
                  B-B at 0°, Varying Temperature                                        maximum absolute difference was be-
                                                                                        tween the average of the first 0°C
                                   Average Tilt Sensor Readings voltsa                  (32°F) readings and the average of the
                                               Tilt sensor no.                          second 0°C (32°F) readings, which was
                                                                                        0.0003 volts or 3.88 arc-seconds. The
       Temperature             1          2        3       4         5        6         largest difference for all sensors was ap-
                                                                                        proximately 193 arc-seconds.
         0°C (32°F)        0.7148     0.6772   0.6804   -0.4997 -0.6044 -0.5488
                                                                                        90-day Zero Stability under
         20°C (68°F)       0.7147     0.6738   0.6792   -0.4930 -0.5997 -0.5575
                                                                                        Constant Inclination and
        40°C (104°F)       0.7147     0.6715   0.6768   -0.4887 -0.5966 -0.5636         Temperature
                                                                                        The final performance check was to de-
         20°C (68°F)       0.7148     0.6743   0.6787   -0.4944 -0.6022 -0.5583
                                                                                        termine how the tilt sensors would per-
         0°C (32°F)        0.7145     0.6776   0.6792   -0.4995 -0.6071 -0.5519         form over an extended time period un-
                                                                                        d er co n s tan t in clin atio n an d
                                                                                        temperature conditions. The six sensors
    Max.        volts 0.0003          0.0060   0.0037   0.0110   0.0106    0.0148       were placed on the sine bar, which was
     diff.      arc-    3.88           78.73    48.03   144.07    138.43    193.11      inclined along the A-A axis (Figure 1)
   between    seconds                                                                   such that the sensors and housings were
 reading sets                                                                           5° to the vertical. With the temperature
                                                                                        maintained at 20°C (68°F), readings
     Average of 5 readings for each temperature.                                        were taken at least twice per day for 90
                                                                                        days. The specification stated that the
such that the sensors were tilted 5° to    ature was held for two hours, and then       variation in readings for any sensor
the vertical. Three of the tilt sensors    temperature readings taken every five        could not exceed 50 arc-seconds, which
were rotated in the sine bar slots to give minutes for one-half hour to ensure sta-     Figure 3 shows was met or exceeded. It
negative readings (tilt sensors 3, 4 and   bility. After this, a set of five readings   is noted that the 90-day zero stability
5) for this sine bar inclination, while thefor each sensor was taken, with readings     data can be influenced by bending and
other three were set to give positive      taken about 15-20 seconds apart. These       tilting of the apparatus during the test.
readings. The chamber temperature was      readings were averaged for each sensor.      This is why the precaution was taken of
                                               For each sensor, the absolute differ-    orienting the positive axis of three of the
cycled through 0°C (32°F), 20°C
                                           ences were calculated between the aver-      sensors (1, 2, and 3) at 180 degrees to
(68°F), 40°C (104°F), 20°C (68°F), and
                                           ages of all temperature data sets, and the   the other three, resulting in 3 positive
0°C (32°F). Note that the temperature                                                   readings and 3 negative readings for the
range tested was significantly more rig-   maximum absolute difference identi-
                                           fied. According to the specification, this   same sine bar inclination. The higher
orous than that specified. Each temper-                                                 levels of drift shown by the two sensors
                                                                    maximum abso-
                                                                    lute difference     at the ends of the sine bar are mirror im-
                                                                    had to be less      ages of each other – as if the ends of the
                                                                    th a n       250    sine bar curled up very slightly during
                                                                    arc-seconds.        the test. A more accurate presentation
                                                                    This analysis is    of the long term drift results would in-
                                                                    shown in Table 2    volve elimination of these apparatus ef-
                                                                    for the six sen-    fects. This could have been done by re-
                                                                    sors. Raw volt-     peating the test with the sensors left in
                                                                    age readings are    the same relative positions on the sine
                                                                    given since these   bar but with each sensor rotated 180 de-
                                                                    are the basis for   grees.
                                                                    computing the           A less accurate method for eliminat-
                                                                    maximum dif-        ing the apparatus effect would be to
                                                                    ference between     subtract the average absolute drift after
                                                                    reading sets,       90 days of all 6 sensors (about 6 arc-sec-
                                                                    which are then      onds) from the long term drift of each
                                                                    converted into      sensor. When this is done, referring to
                                                                    arc-seconds. Re-    Figure 3, it can be seen that the maxi-
Figure 3. Performance testing for 90-day stability, A-A @
                                                                    ferring to Table    mum rate of drift (for tilt sensor no. 4)
5 , constant temperature.

                                                                                                   Geotechnical News,   March 2008   35

 was about 30 arc-seconds per 90 days.          the six tilt sensors tested met or ex-      MA 02115, tel. 617-373-3995, email:
 Another way of expressing this would           ceeded the performance criteria re-
 be approximately 0.2% of full scale per        quired by the project specifications.
 year. The rate of drift appeared to be di-     The results indicate that these devices     David Mazzei, Graduate Student, De-
 minishing towards the end of this              can provide reliable performance under      partment of Civil and Environmental
 90-day period. Tests have commenced            the specified conditions, and provide       Engineering, Massachusetts Institute of
 to investigate the rate of drift over          evidence that MEMS technology has           Technology, 77 Massachusetts Avenue,
 longer periods of time.                        tremendous potential for geotechnical       Ca m b r id ge, M A 0 2 1 3 8 , e m a il:
 A series of performance tests was per-         Thomas C. Sheahan, Professor and Act-       John McRae, Vice-President, Geokon,
 formed on MEMS-based tilt sensors to           ing Chair, Department of Civil and En-      Inc., 48 Spencer Street, Lebanon, NH,
 assess their performance in three pri-         vironmental Engineering, Northeastern       03766, tel. 603-448-1562, email:
 mary areas: short-term repeatability,          University, 400 Snell Engineering Cen-
 temperature sensitivity, and 90-day            ter, 360 Huntington Avenue, Boston,
 zero stability. In all three of these areas,

                                                A New Wireless MEMS-Based System for
                                                Real-Time Deformation Monitoring

                                                Tarek Abdoun
                                                Victoria Bennett

 Abstract                                       signals proportional to vibration during    degrees of freedom. These rigid seg-
 Geotechnical instrumentation using             earthquakes or construction activities.     ments and flexible joints are combined
 Micro-Electro-Mechanical Systems               Three accelerometers are contained in       t o f o r m a s e n s o r ar r a y c a l l e d
 (MEMS) are relative newcomers to this          each 30 cm (1 ft) long rigid segment for    ShapeAccelArray, or SAA, which is ca-
 conservative industry. This article aims       measuring x, y, and z components of tilt    pable of measuring three-dimensional
 to describe a new MEMS-based system            and vibration. The rigid segments are       (3D) ground deformations at 30 cm (1
 for in situ deformation and vibration          connected by composite joints that pre-     ft) intervals to a depth of 100 m (330 ft).
 monitoring. After an introduction to the       vent torsion but allow flexibility in two   These sensor arrays are manufactured
 design and methodology of this system,
 some of the questions surrounding the
 use of MEMS sensors will be ad-
 dressed, including temperature sensi-
 tivity, long-term stability and relative
 cost. Methods of installation and re-
 trieval will also be briefly described. Fi-
 nally, a description and validation of re-
 corded field data from an instrumented
 unstable slope in California will be pre-

 This system has been developed in an
 effort to combine recent advances in the
 miniaturization of sensors and electron-
 ics with an established wireless infra-
 structure to enhance geotechnical mon-
 itoring. The concept is based on triaxial
 MEMS accelerometer measurements of
 angles relative to gravity. These same
 MEMS accelerometers also provide               Figure 1. Schematic of SAA subarray assembly.

36   Geotechnical News,   March 2008
                                                                                     GEOTECHNICAL INSTRUMENTATION NEWS

b y Me a s u r a n d I n c . ( w w w.       valid installation.                             Temperature Sensitivity, with whom this arti-            The SAAs are factory-calibrated and         This is a topic of discussion for many
cle’s authors have worked for the past      completely sealed, requiring no field           geotechnical engineers in regard to all
five years.                                 assembly or calibration. Because each           instrumentation. A full temperature cal-
    Arrays are constructed by connect-      segment of the SAA contains three or-           ibration is done on each MEMS sensor
ing subarrays of eight segments             thogonal sensors, arrays can be in-             individually prior to its inclusion in an
end-to-end. Microprocessors, one per        stalled vertically or horizontally. The         array. Measurand has completed a study
subarray, collect data from groups of       intended installation orientation does          of the temperature coefficients of the
sensors and transmit this digital data to   not need to be specified when ordering.         MEMS accelerometers and found that
the surface using just two communica-       Orientation is selected in the software.        the change in output of the sensor is lin-
tion wires; see Figure 1 for a schematic    Each sensor has an output that is the           ear with temperature. Calibration files
of the SAA assembly. Because they re-       sine of the angle of tilt over a range of       associated with each SAA allow the au-
quire only two communication wires,         360 degrees. Calculations use data from         tomatic calibration for temperature ef-
arrays are thin enough to fit into 25 mm    the sensors having maximum sensitiv-            fects in each individual sensor. A digital
(1 in) casing for installation and are      ity for a given orientation. The sensor         temperature sensor is included within
flexible enough to be rolled up on a reel   arrays arrive at the jobsite on an 86 cm        the SAA near each microprocessor.
for shipping and storage. Traditional       (34 in) diameter reel, see Figure 2, and        Thus, each temperature sensor cali-
probe inclinometers require guide cas-      can be lowered into vertical, or pushed         brates the MEMS sensors in the eight
ing to measure ground deformations.         into horizontal, 25 mm (1 in) casing.           segments surrounding it. This configu-
The torsion-restrained joints of SAAs       The initial shape of the installation, or       ration is deemed sufficient for typical
and 3D calculation method associated        the absolute deviation of the installation      underground applications as the ground
with the torsion constraint make this       from a virtual vertical or horizontal line,     temperature is usually constant below
grooved casing unnecessary. Upon            can be immediately viewed on a laptop.          1.5 m (5 ft) from the surface. A denser
field installation, the manufacturer rec-   An SAA is modeled as a polyline in the          construction of temperature sensors
ommends lifting the SAA off the bot-        software, with x, y, and z data represent-      would be possible but only necessary if
tom of the 25 mm (1 in) casing and          ing the vertices of the polyline. In the        a large temperature gradient is expected
ensuring it can be rotated without resis-   case of near-vertical installations, the        across any consecutive eight segments,
tance. If it is purposely torqued with 10   vertices correspond to the joint-centers        or if the temperature gradient is in a lo-
N-m (7.4 ft-pounds), it is possible to      of the array in 3D. For near-horizontal         cation where ground deformations are
achieve approximately 0.3 degrees ro-       installations, the vertices show vertical       expected.
tation per segment, with elastic return     deformation only versus horizontal po-
if the torque is removed. Users should      sition.                                         Stability with Time
seek to minimize this resisting torsion.        Wireless SAA data transmission is           Data correction procedures have been
This can be realistically achieved in a     possible with the inclusion of an on-site       used for several decades to remove any
                                            data acquisition system, called a wire-         long-term drift of the gravity-sensing
                                            less earth station. For the fifteen field ar-   transducers from the calculated defor-
                                            rays installed to date, this wireless data      mation data of probe inclinometers. The
                                            transmission has been available within          use of MEMS accelerometers virtually
                                            24 hours of the instrument installation.        eliminates concerns of long-term drift
                                            Similar to traditional probe and in-place       in the SAA. MEMS accelerometers are
                                            inclinometers, data from the SAA rep-           manufactured from pure silicon, using
                                            resents deviations from a starting condi-       photolithographic methods developed
                                            tion or initial reading. These data are         by makers of computer chips. They
                                            sent wirelessly, over the cellular tele-        were originally developed for, and con-
                                            phone network, to Measurand’s auto-             tinue to be used in, the automotive in-
                                            mated server, where data are made               dustry for airbag deployment. Automo-
                                            available to users through Measurand’s          tive sensors undergo exhaustive testing
                                            viewing software (a download included           in extreme environments and must re-
                                            with purchase of the wireless earth sta-        main stable for at least 10-15 years. The
                                            tion) and an internet connection. Auto-         internal structure of MEMS accelerom-
                                            mated SAAs typically collect data once          eters is based on the bending of cantile-
                                            or a few times a day but this collection        vered beams of pure silicon, with di-
                                            frequency can be specified by the user          mensions less than 1 mm (0.04”), due to
                                            and changed at any time, through the            the force of gravity. Deflections of the
                                            same wireless interface used to receive         tiny beams are measured using elec-
Figure 2. 32 m (104 ft) SAA on
                                            the data.                                       tric-field technology, also built into the
shipping reel.

                                                                                                       Geotechnical News,   March 2008   37

 MEMS using photolithographic tech-              grees from vertical, the sine curve slope    accurate readings due to the effects of
 niques.                                         is degraded by only 1.5 percent, and at      shear across an abrupt boundary. The
    The lack of sliding parts and the            +/- 45 degrees is degraded by only 29        casing acts as a physical, spatial filter
 near-inertness of pure silicon make for a       percent. Similar numbers apply to devi-      for bends at shear boundaries. If one of
 stable sensor technology. Stability over        ation from a horizontal position, due to     the SAA’s rigid segments did bend, it
 time has been confirmed by data from            the use of three MEMS accelerometers         would cause an error in the tilt reading
 three SAAs, sampled several times per           per segment.                                 at that segment, which would propagate
 day, over a period of 1.5 years. The de-            Because the SAA is left in place per-    as an incorrect displacement in the rest
 formations from portions of the arrays          manently or semi-permanently, it does        of the data, to the extent that the tilt of
 known to be in stable soil (below ob-           not have the potential of errors due to      the circuit board inside no longer repre-
 served shear zones) were monitored              mechanical mismatch between a                sented the tilt between the nearby joint
 over the entire period of measurement.          wheeled instrument and a grooved cas-        centers. The ground deformation pro-
 After an initial settling-in period (see        ing, and there is less opportunity for op-   files above and below the damaged seg-
 “Installation and Retrievability” section       erator error. This advantage is most         ment would be correct but displaced
 below), data in the stable soil were            apparent when the casing becomes ex-         from each other. Further bending could
 found to deviate from their initial read-       tremely distorted.                           damage a circuit board; in this case, if
 ings by no more than +/-1.5 mm                                                               the circuits are not shorted, the data
 (+/-0.06 in), including in arrays over 32       Cost                                         above the affected segment could possi-
 m (104 ft) long.                                SAAs are priced to be competitive with       bly still be retrieved. In a recent installa-
                                                 inclinometers including the consider-        tion, a segment was bent about 15 de-
 Accuracy of Deformation                         ation that with SAAs the user has 3D         grees near its center by a large impact
 Measurement                                     deformation and 3D vibration data,           during shipping. Since the circuit
 The accuracy of the deformation mea-            wireless transfer of data, and an auto-      boards are near the ends of the seg-
 surement of the SAA is +/- 1.5 mm per           matic data collection at frequent time       ments, the segment was straightened
 30 m (+/- 0.06 inches per 100 ft). This         intervals and 30 cm (1 ft) spatial inter-    and the array checked out fine during
 figure can be directly compared to the          vals. A 29 m (96 ft) long SAA, includ-       on-site diagnostic testing.
 reported system accuracy of traditional         ing software, costs less than $14,000.
 probe inclinometers, +/- 7.6 mm per 30                                                       Installation and Retrievability
 m (+/- 0.3 inches per 100 ft), because          Ordering Lengths                             The SAA system was designed to be re-
 both of these specifications are refer-         SAAs are ordered in multiples of eight       trievable; a desirable option to offset the
 enced to a virtual straight line, or the ini-   segments (2.4 m or 8 ft) to a total length   capital cost. The method of SAA instal-
 tial reading of both instruments. The           of 100 m (328 ft) and can be sent back to    lation has evolved in response to main-
 SAA system accuracy specification was           Measurand for shortening, lengthening        taining retrievability and accurate de-
 derived empirically from thousands of           or repair. Arrays are shipped fac-           formation measurements. The first
 frames of wireless data over a period of        tory-calibrated and ready to install, see    three field SAAs were installed in tradi-
 1.5 years, from three different field lo-       Figure 2. There are no provisions for        tional probe inclinometer casing. The
 cations.                                        modifying the length of the array in the     casing was grouted into the borehole
     The MEMS accelerometers                     field.                                       and the annulus between the SAA and
 mounted in pipes on a mechanical                                                             the casing was backfilled with sand; this
 goniometer readout have an absolute             Bending of a Segment in the                  method is briefly described in the fol-
 accuracy similar to that of conventional        Array
                                                                                              lowing section. The sand backfill was
 inclinometers and excellent ‘linearity’         Conventional in-place inclinometers
                                                                                              chosen because it could be flushed out
 over a 45 degree range. ‘Linearity’is ac-       stop giving correct data if the ground
                                                                                              of the casing with pressurized water
 tually the match of the arcsine of the          deformation is large enough to cause
                                                                                              from the top of the borehole and the
 output to the tilt in degrees since MEMS        the connecting rod to make contact with
                                                                                              SAA could be extracted.
 accelerometers exhibit a sinusoidal re-         the inside of the casing. The 30 cm (1 ft)
                                                                                                  Concerns about incomplete backfill
 sponse to tilt (output voltage = sin (an-       length of the rigid segments in the SAA,
                                                                                              due to sand bridging are valid. SAA
 gle)). Accuracy, as for traditional probe       however, makes it less likely that a seg-
                                                                                              data from field installations using this
 inclinometers, is best near either pure         ment would bend due to this kind of lo-
                                                                                              method of installation showed +/- 1 mm
 vertical or pure horizontal (probe incli-       cal shear. This short segment length
                                                                                              (0.04 in) zig-zags a few days after in-
 nometers are usually specified within           tends to reduce bending of the seg-
                                                                                              stallation. This shape remained con-
 +/-3 degrees of vertical). In the case of       ments, forcing bends to be taken in the
                                                                                              stant over a year of monitoring and is
 MEMS accelerometers, the sinusoidal             flexible joints which have a 45 degree
                                                                                              attributed to the slight settling of array
 response causes a very gradual degrada-         range of motion and pull strength of 2.2
                                                                                              segments within the casing as sand
 tion away from the pure pose, due to a          kN (500 lbs). The recommended instal-
                                                                                              bridges are broken, usually within the
 decrease in slope of the sinusoid away          lation of SAAs in 25 mm (1 in) casing
                                                                                              first month after installation.
 from its “zero-crossing”. At +/-10 de-          also helps prevent sensor damage or in-
                                                                                                  Because this settling trend was ob-

38   Geotechnical News,   March 2008
                                                                                    GEOTECHNICAL INSTRUMENTATION NEWS

served for the first three field installa-   shown to respond to sub-mm (less than         area of California was nearly 30 cm (1
tions of SAAs, an installation method        0.04 in) local shear deformations in the      ft) below the average and the recorded
using a 25 mm (1 in) casing to support       surrounding soil. In sand especially, the     ground deformations are correspond-
the SAA within a borehole was devel-         25 mm (1 in) casing will tend to bend         ingly low, less than 10 mm (0.4 in).
oped. The method allows for the use of       rather than shear and the short segment       However, the trends are visible and
either sand or grout backfill, depending     lengths of the SAA conform to the             comparable.
on the anticipated deflection magnitude      bends. Testing with a 25 mm (1 in) cas-
and the desire to retrieve the array. In     ing held by vises has shown that an “S”       Conclusion
this method, 25 mm (1 in) casing is held     curve of the casing with total lateral dis-   The California field test was an excel-
with sand or grout in a borehole, an ex-     placement of 50 mm (2”) within 60 cm          lent trial opportunity for this new
isting inclinometer casing, or other         (24”) does not bend the segments of an        MEMS-based system and SAA data
large casing. If sand is used, the in-       SAA inside the casing, and the SAA            were comparable to traditional probe
creased stiffness of the 25 mm (1 in)        may be pulled past such a curve by hand       inclinometer data. The results of the
casing compared to the SAA alone             for retrieval. This “gentle” response to      field installation also addressed some of
tends to minimize zig-zagging should         external forces is thought to account for     the questions about the use of MEMS
the sand settle, assuming due care has       the ability of the SAA to keep reading        sensors in geotechnical instrumentation
been taken to avoid voids larger than 30     when other nearby inclinometer casings        and furthered confidence in this sys-
cm (1 ft).                                   have sheared. Many users of the SAA           tem’s temperature calibration. The first
    In the small casing method, the SAA      opt for the use of sand backfill because      field installations also raised other
is lowered into the 25 mm (1 in) casing      of the greater likelihood of retrieving       questions about the best installation
using flat webbing to fill the 3 mm (0.1     the array.                                    method for the SAA. Several subse-
in) annular space between the SAA and                                                      quent successful installations have been
the casing. The SAA and the webbing          Unstable Slope Installation                   completed using the 25 mm (1 in) cas-
are anchored together at the far end of      In June 2006, a 20 m (64 ft) array was        ing method. The successful monitoring
the borehole by an enlarged end of the       installed in an unstable slope in Califor-    at these sites is ongoing and the data
webbing. Pull may be exerted on the          nia, which has been documented as an          confirms the SAA system accuracy to
webbing and the SAA to facilitate re-        ancient landslide prone area, 225 m
moval. This new method of installation       (738 ft) long and 100 m (330 ft) wide,
has been utilized with ten field arrays to   by Caltrans’ geotechnical site reports.
date (nine of them using sand as back-       Four conventional probe inclinometer
fill) and has eliminated the zig-zag pat-    casings were installed at the highway
tern of deformation observed with sand       level, through this region, from 2002 to
b ack fill in th e first th ree field        2005, and ground deformations were
installations.                               large enough to shear some of these cas-
    The small casing method has been         ings. The SAA was installed approxi-
                                             mately 1 m (3 ft) away from a new incli-
                                             nometer casing installation. Figure 3
                                             shows the SAA installation. Threaded 3
                                             m (10 ft) long sections of 7.6 cm (3 in)
                                             diameter inclinometer casing were used
                                             in a 12.7 cm (5 in) diameter borehole.
                                             The annulus between the SAA and the
                                             casing was backfilled with coarse sand.
                                             This backfill was compacted, to the ex-
                                             tent possible, by striking the side of the
                                             casing with a mallet during the sand
                                             placement (Abdoun et al. 2007). As
                                             mentioned above, this method of instal-
                                             lation has been superseded by the 25
                                             mm (1 in) casing installation method
                                             due to concerns about sand bridging.
                                             Figure 4 shows a comparison of SAA
                                             data with conventional probe inclinom-
                                             eter data for a nine month monitoring
Figure 3. Installation of the SAA for        period. The National Weather Service
unstable slope monitoring in                 reported that the amount of rainfall dur-     Figure 4. Unstable slope ground
California                                   ing the 2006-2007 rain season in this         deformations at California test site.

                                                                                                      Geotechnical News,   March 2008   39

 be comparable to conventional incli-           consideration that with SAAs the             late z “indirectly”, so in your sense
 nometers, but with improved spatial            user has 3D deformation ... data”.           both are ”2D" instruments. How-
 resolution. Automated wireless data            For me, “3D” in a vertical installa-         ever, the traditional probe inclinom-
 collection, retrievability, improved spa-      tion means that the instrument can           eter or a traditional chain of
 tial resolution, and long-term accuracy        be used to measure the same thing            in-place-inclinometers would not be
 will earn this MEMS-based system a             that a conventional inclinometer             able to conform to any but the most
 place in the instrumentation inventory         does (A and B) plus vertical com-            gentle of “S” or “C” shapes, or 3D
 of the geotechnical community.                 pression, for which the hardware             shapes. As a simple example of a 3D
                                                must telescope in conformance with           measurement, consider the mea-
 Reference                                      the ground alongside, and for which          surement of the settlement of sand in
 Abdoun, T., V. Bennett, L. Danisch, T.         there must be sensors to measure ax-         a large casing, by virtue of the
   Shantz and D. Jang. (2007). “Field           ial length change. Hence it isn’t            zig-zagging of an SAA in the sand.
   installation details of a wireless           clear to me how the SAA can be               Or, as a more useful measurement,
   shape-acceleration array system for          used to measure the z component of           consider data we have collected
   geotechnical applications.” Proceed-         deformation. Can we agree that,              from SAAs directly placed in full
   ings of SPIE, San Diego, CA, March           even though “three accelerometers            scale soil models (6 m; 19.7 ft
   19-22, Volume 6529.                          are contained in each 30 cm (1 ft)           height) without casing, in large
                                                long rigid segment for measuring x,          shake-table installations in the USA
 Tarek Abdoun, Associate Professor, and         y, and z components of tilt ...” that,       and Japan. Initially straight arrays
 Victoria Bennett, PhD Student,                 in the application corresponding to a        followed major lateral deformation
 Rensselaer Polytechnic Institute, 110          typical probe inclinometer applica-          (0.8 m; 2.6 ft) of the soil model, con-
 8th Street, JEC 4049, Troy, NY 12180,          tion, the SAA is used to measure             forming to the 3D shape of the fail-
 Tel. 518-276-6544, Fax 518-276-4833,           only the x and y components?                 ure, including settlement in z
 emails:, bennev              A. You are correct that the SAA is not          direction (0.16 m; 0.5 ft), which                                       compressible in z in the sense of a          agreed very well with independent
                                                telescoping array. However, be-              measurements of the settlement us-
 Editor’s Note                                                                               i ng tr adi t i onal d i s p l a c e m e n t
                                                cause the SAA can be mounted in a
 When editing the above article by              flexible small-diameter (25 mm; 1            sensors.
 Abdoun and Bennett I questioned their          in) casing or used without a cas-         Q. You say, “Arrays can be installed
 use of the term “3D”. The following            ing, is flexible at frequent intervals       vertically or horizontally. The in-
 summarizes my questions and the au-            (30 cm; 1 ft), and can measure data          tended installation orientation does
 thors’ answers.                                rapidly and automatically, it can            not need to be specified when order-
 Q. In the Introduction you say, “… is          be placed usefully in significantly          ing. Orientation is selected in the
    capable of measuring three-dimen-           non-straight shapes or can undergo           software”. This seems to say that an
    sional (3D) ground deformations”.           very large changes in shape. Exam-           individual sensor that was used in a
    And in the “Cost” section you say,          ples of such shapes include an ini-          vertical installation can also be used
    “SAAs are priced to be competitive          tially straight line, vertical or at an      in a horizontal installation, when it’s
    with inclinometers including the            angle, that becomes deformed into            zero axis has been changed by 90
                                                an “S” curve or a “C” curve during           degrees? How can this be so, unless
                                                slope failure. Other examples are            the sensors have a 360 degree
                                                shapes that are initially “S” or “C”         range?
                                                shaped and become more deformed           A. Each sensor has an output that is the
                                                during failure. Shapes like those            sine of the angle of tilt over a range
                                                will respond to 3D changes in the            of 360 degrees. Sensitivity is maxi-
                                                soil including vertical compression,         mum near the zero-crossing of the
                                                and the SAA will be able to follow           sine wave, so vertical calculations
                                                the changes at high speed, at high           employ the x and y sensor outputs,
                                                spatial resolution, and with minimal         which are at their zero crossings
                                                fear of the casing shearing or falsely       when vertical, and horizontal calcu-
                                                staying straight during the move-            lations employ the z sensors, which
                                                ment. It is true that a traditional          will be at their zero crossings when
                                                manual inclinometer scanning such            horizontal. The arrays “know” when
                                                a shape, and the SAA, both measure           they are upside down and vertical.
                                                x and y directly and use knowledge           This knowledge requires the z sen-
                                                of measurement intervals or seg-             sor in addition to the x and y. These
                                                ment length, respectively, to calcu-         are true 3D devices.

40   Geotechnical News,   March 2008
                                                                                   GEOTECHNICAL INSTRUMENTATION NEWS

                                             Field Evaluation of a MEMS-Based,
                                             Real-Time Deformation Monitoring System
                                             Matthew B. Barendse

Introduction                                 Most of the Casagrande classifications      post-construction settlement of the em-
In 2006, New York State Department of        plot in the CH range, with liquid limits    bankment and to minimize dragload
Transportation (NYSDOT) partnered            from 40 to 71 and plasticity indices        and lateral pressure on the piles. The
with Rensselaer Polytechnic Institute to     from 22 to 46. Laboratory testing indi-     PVDs were pressed to a depth of 20 m
participate in the field evaluation of a     cated the compression index varied          (66 ft) in a square grid pattern with 1.2
new in-place inclinometer (IPI) system       f r o m 0 . 5 2 to 1 . 7 6 an d th e        m (4 ft) spacing. A surcharge fill 1.5 m
based on Micro-Electro-Mechanical            overconsolidation ratio from 1.0 to         (5 ft) higher than the proposed finished
Systems (MEMS) sensor technology.            approximately 3.5.                          grade was then constructed utilizing a
T h e in s tr u m e n t, k n ow n a s th e                                               temporary three-sided geosynthetic re-
ShapeAccelArray, or SAA, is manufac-         Embankment Foundation                       inforced earth wall to match the pro-
tured by Measurand, Inc. of Frederic-        Treatments                                  posed footprint of the bridge abutment.
ton, New Brunswick, Canada. To the           The new highway alignment is shifted        A fill waiting period of one year was
author’s knowledge, Measurand is one         south of the existing, thereby requiring    specified with the provision that the
of only two companies to make such a         new bridge approach embankments up          waiting period could be shortened on
device; the other is Geodaq, Inc. of Sac-    to 5 m (16 ft) high. The 62 m (203 ft)      the basis of field monitoring results.
ramento, California. The demonstration       long truss bridge will be supported on
project selected was the replacement of      H-piles bearing on bedrock. Treatments      East Embankent
a 96-year-old steel truss bridge over the    for the east and west embankments dif-      Instrumentation Program
Champlain Canal in eastern upstate           fer slightly due to a number of factors:    Six subsurface settlement platforms
New York.                                    this article will focus on the east side,   and five vibrating wire piezometers
   The goals of this evaluation were to:     where the SAAs were installed.              monitored settlements and pore pres-
measure vertical and horizontal ground          Pre fabr i cat ed ver t i cal dr ai ns   sures during the waiting period. Lateral
displacement, assess the reasonable-         (PVDs), commonly known as wick              deformations of the foundation soils
ness of the measured data, and devise        drains, were determined the most            were monitored by a traditional probe
methods to install and subsequently          cost-effective solution to reduce           inclinometer located between the sur-
extract the SAA.

Project Site Conditions
The Champlain Canal was first opened
in 1819. In the project area, the canal
was largely coincident with an existing
meandering stream named Wood
Creek. Circa 1908, the canal was up-
graded for larger traffic by widening
and straightening its alignment to its
current position (Figure 1). Record
plans for the project site show that
dredged spoils were used to fill the old
creek and canal, as well as to build
berms adjacent to the new canal.             Figure 1. Bridge replacement site plan.
    Geologically, the site lies near the
n o r th e r n ex te n t o f a larg e
glaciolacustrine deposit known from
previous experience to contain deep de-
posits of soft, compressible varved silts
and clays. The generalized subsurface
profile is shown in Figure 2. In the layer
of primary geotechnical interest, the
very soft silty clay layer, natural mois-
ture contents ranged from 37% to 82%.
                                             Figure2. Interpreted subsurface conditions.

                                                                                                    Geotechnical News,   March 2008   41

 charge and the canal side slope. Supple-       sand. To free the SAA, the sand is            Cellular Contact and Power
 menting this were two SAAs, both 31.7          flushed from the hole by pumping water        Supply
 m (104 ft) long, one installed horizon-        to the bottom through the open-ended          Cellular coverage was verified prior to
 tally and one vertically as shown in Fig-      tremie hose. Water pressure up to ap-         the installation. Wires from both SAAs
 ures 1 and 2. (For a thorough descrip-         proximately 2 MPa (150 psi) is used to        were run to a lockable metal cabinet
 tion of the SAA technology, refer to           fully saturate and eventually liquefy the     housing the data collectors, 100
 Abdoun and Bennett in this episode of          sand. The use of sand backfill with in-       amp-hour deep-cycle battery and mo-
 GIN, or Danish et. al., 2004.)                 clinometers is acknowledged as a              dem. A solar panel trickle charger and
                                                method that is prone to voids or incon-       cellular antenna were mounted over the
 SAA Installation                               sistent support (Green and Mikkelsen,         cabinet. During the first 10 months, bat-
 Due to the relatively high unit cost of        1988) and will be a topic of further dis-     tery power remained fairly constant
 the SAA, devising an installation proce-       cussion later in this article.                given the moderate direct sunlight and
 dure that could also reasonably assure             The vertical SAA was installed at the     several readings taken per day.
 later retrieval was considered a worth-        demonstration site in a drill hole
 while endeavor. This proved challeng-          roughly 2.4 m (8 ft) south of the conven-     The Data
 ing as the SAA must be in intimate con-        tional inclinometer. Threaded sections        Figure 3 shows the profile of inclinome-
 ta c t w ith th e g r o u n d to r e c o r d   of monitoring well casing (50 mm [2 in]       ter SI B compared with the vertical SAA
 deformations during operation, and yet         ID Schedule 40 PVC) were inserted into        over the same time period. The general
 also be extractable from a potentially         the hole and grouted in place using a         shapes of the upper part of the curves
 severely distorted shape. The difficulty       water/bentonite/cement mix, mixed in          are comparable, however near elevation
 is compounded by the fact that the max-        that sequence and proportioned                25.5 the SAA shows approximately 2
 imum tension recommended on the                1/0.06/0.06 by weight. To keep the cas-       mm (0.08 in) of “negative” displace-
 joints is only approximately 2.2 kN            ing dry prior to adding the sand, steel       ment. This represents movement to-
 (500 lbs). Prior to the field installation,    drill rods were used to counteract buoy-      wards the surcharge load and therefore
 NYSDOT bench tested a technique us-            ancy in lieu of water. Next, 3 mm (0.125      raises some questions. Possible expla-
 ing a 9.8 m (32 ft) long mockup SAA in         in) wire-wound steel cable and 10 mm          nations are complications due to settle-
 a test boring. The method involves in-         (0.375 in) plastic tremie tubing were         ment of the sand, counterflexure of the
 stalling the SAA and a tremie hose in-         taped approximately every 1.5 m (5 ft)        casing, or an incorrect roll calibration
 side 50 mm (2 in) plastic casing and           to the SAA. The SAA was lowered into          (i.e. twist) in the instrument itself. It is
 backfilling the annular space with clean       the well casing by hand, taking care not      also noted that the SAA exhibited very
                                                to damage any joint by bending it be-         small zigzag movements between ele-
                                                yond 45 degrees, and held suspended           vations 9 and 13. At this point, the sand
                                                off the bottom of the hole by the cable.      is considered the most likely culprit for
                                                Silica sand was slowly funneled into the      both anomalies.
                                                top of the hole while striking the top of         Figure 4 shows the settlement profile
                                                the casing intermittently with a rubber       of the horizontal SAA. Interestingly, the
                                                mallet. No additional compactive effort       peaks of the curves are offset towards
                                                or tamping was used.                          the south side of the surcharge, reflect-
                                                    Installation of the SAA in the hori-      ing the influence of the existing em-
                                                zontal orientation was simpler than the       bankment on the preconsolidation of
                                                vertical in that intimate contact with the    the clay layer. The settlement magni-
                                                ground was less critical (gravity would       tude is about 30% less than measured
                                                keep the device on the bottom of the          with the settlement platforms, owing to
                                                casing), and both ends of the device          the SAA’s location nearer the end of the
                                                would be accessible during removal.           embankment.
                                                The SAA was slid directly into 25 mm
                                                (1 in) ID PVC electrical conduit pipe         Retrieval and Landslide
                                                and laid in a small trench within the         Monitoring
                                                gravel drainage blanket atop the PVDs.        In August 2007, an ancient landslide,
                                                As the SAA fits “snugly” within the           approximately 600 m (2000 ft) long by
                                                conduit, cable-pulling lubricant was          35 m (115 ft) deep, was reactivated on a
                                                used to reduce friction. To verify read-      NYSDOT construction project in west-
                                                ings later on and serve as a check for set-   ern New York. The decision was made
 Figure 3. Vertical SAA and                     tlement of the reference end, stickup         to pull the vertical SAA from the dem-
 inclinometer (SI B) comparison on              rods with survey targets were attached        onstration site slightly ahead of sched-
 July 13, 2007. Note:Corrected for              over both ends of the SAA.                    ule and install it on the slide.
 groove orientation

42   Geotechnical News,   March 2008
                                                                                   GEOTECHNICAL INSTRUMENTATION NEWS

    Field retrieval at the bridge site       previously installed 70 mm (2.75 in)        directly into the ground and simply in-
looked promising at first, as sand began     ABS inclinometer casing. To reach a         sert the SAA. While eliminating the
to flow from the hole approximately 15       stable layer beneath the assumed shear      need for sand backfill entirely is un-
minutes after the pump was turned on.        plane, extra sections of conduit were at-   doubtedly preferable, it is unknown
However, following a (perhaps over-          tached such that the top sensor of the      what radius of bending would render
zealous) water pressure spike, the           SAA was 10 m (33 ft) down from the          the device irretrievable in this scenario.
method became ineffective at removing        top of the hole. The annular space was      Furthermore, this method may be inap-
the sand. A new top-down approach            again filled with clean sand, although      propriate if vibration monitoring is con-
was adopted by gluing together 3 m (10       this time the space was filled with water   sidered a critical aim. In any case, mak-
ft) sections of 12 mm (0.5 in) PVC pipe      so that the sand would be allowed to        ing the device sacrificial is perhaps
and working them down the hole while         slowly settle out of suspension. Inside     perfectly justifiable on larger projects.
jetting water. Although having to glue       the 25 mm (1 in) conduit, a flat nylon      Smaller scale jobs stand to benefit from
sections of pipe was somewhat inconve-       strap was used as space filler to provide   further thinking and experimentation
nient and ran the risk of losing a section   a tighter fit with the SAA. The retrieval   with installation and retrieval methods.
down the hole, this method did have the      system was also modified in the hopes       Clearly there is opportunity for innova-
advantage of requiring very low pres-        that lower pressures can be used to ex-     tion in this area.
sure to remove the sand in 3 m (10 ft) in-   pel the sand. Three tremie hoses were           A potential drawback is that once it
crements. Approximately 5 m (16 ft)          installed with their ends at different      is installed, the SAA cannot be removed
from the bottom of the hole the wash         depths within the hole. During removal      to inspect a faulty sensor and reinstalled
water suddenly became turbid, likely         they will be attached to the pump one at    as easily as traditional IPI probes might
indicating that the pressure spike           a time from highest to lowest.              be. At both test sites occasional minor
caused a break in the well casing. This                                                  “hiccups” were recorded; that is, read-
presumably interfered with maintaining       Discussion                                  ings which were considered question-
adequate pressure using the tremie hose      The practice of using sand backfill with    able or erroneous based on engineering
method.                                      inclinometers has a somewhat speckled       judgment. These unexpected hiccups
    Installation at the landslide was        past and this project proved no excep-      may very well be attributed to other fac-
modified based on lessons learned at the     tion. Data stability at the landslide was   tors, such as the sand backfill, however
demonstration site and other experience      improved by the addition of the smaller     one cannot completely rule out instru-
by the manufacturer. The SAA was in-         casing to house the SAA. The manufac-       ment error. This issue might eventually
stalled inside 25 mm (1 in) PVC electri-     turer’s current installation recommen-      be resolved by an improved understand-
cal conduit which was placed into a          dation is to grout 25 mm (1 in) casing      ing of the signal processing of the
                                                                                         MEMS microchips themselves.

                                                                                         The SAAs provided continuous ground
                                                                                         deformation profiles in both vertical
                                                                                         and horizontal applications, utilizing
                                                                                         autonomous remote data acquisition.
                                                                                         Given the small overall displacement
                                                                                         magnitudes at the bridge site, the verti-
                                                                                         cal SAA results correlate fairly well
                                                                                         with the traditional probe inclinometer.
                                                                                         The horizontal SAA results are consis-
                                                                                         tent with the predicted foundation re-
                                                                                         sponse and, as one would expect based
                                                                                         on elastic theory, show proportionally
                                                                                         less subsidence than the subsurface set-
                                                                                         tlement platforms.
                                                                                            The sand backfill in the vertical in-
                                                                                         stallation provided reasonable support,
                                                                                         as demonstrated by the correlation of
                                                                                         results with the probe inclinometer,
                                                                                         however it may also have been the cause
                                                                                         of some of the apparent discrepancies.
Figure 4. Cross Section through Horizontal SAA. Notes: Initialized on April 24,          The sand backfill was successful in al-
2007. Fill completed on May 1, 2007. The slight upward movement at the south             lowing subsequent extraction of the
end of the array is due to the stickup rod being moved by construction operations.       SAA.

                                                                                                    Geotechnical News,   March 2008   43

    This article demonstrates that al-        References                                   tion Measurement Device and
 though some practical details remain in      Green, G.E. and Mikkelsen, P. E., 1988,      Method,” Patent Application.
 need of further field trials, this             “Deformation Measurements with
 MEMS-based IPI system can be used              Inclinometers,” Transportation Re-      Matthew B. Barendse, Civil Engineer
 for real-world applications in its current     search Record 1169, Transportation      II, New York State Department of Trans-
 state of development. In summary,              Research Board, National Research       portation, Geotechnical Engineering
 NYSDOT is encouraged by these initial          Council, Washington D.C., pp. 1-15.     Bureau, 50 Wolf Rd., Albany, NY 12232,
 results and will continue applying and       Danish, L.A., Lowery-Simpson, M.S.,       Tel. 518-457-4796.
 testing this new technology.                   Abdoun, T., 2004, “Shape-Accelera-      email:

44   Geotechnical News,   March 2008

Shared By: