Friday, March 13, 2015

Benefits of Automated Monitoring

From a paper published by the author that describes the use of tiltmeters for advanced warning for natural disasters (landslides) and construction induced movements.  Both of these examples show fine correlation of instrument readings and observed phenomenon.  For those that wonder about the application of trigonometry in the real world - see Case History 2.

Because of recent advances in monitoring systems, their benefits are within reach of large and small civil engineering firms, and government agencies.  When included as part of the project design, monitoring systems can reduce costs that would otherwise be incurred as a result of overly conservative design assumptions.  Monitoring new or existing construction helps maintain safe working conditions by providing early warning of instability and therefore time to correct the problem.  Also, monitoring provides valuable quality assurance for verifying that as-built construction conforms to design.


This project shows that a relatively simple, easy to install automated system can be used to provide early warning of catastrophic slope failure. 

Project Description
The winter of 1998 brought devastating rain and landslides to California.  Several automated remote systems were designed and installed to monitor the stability of slopes during repair and cleanup.  One of these systems involved monitoring at the top of the slope in the vicinity of the Laguna Niguel landslide in southern California.  Slope movements long before the slide revealed that the homes had been built too close to the edge of a steep slope.  Personnel safety of current residents was of primary concern.  Several homes had to be abandoned, and a program of automated monitoring was implemented to provide advanced warning of imminent slope failure. 
Automated Instrumentation
The instruments installed to monitor the landslide consisted of an array of in-place inclinometers (IPIs) buried in shallow holes along the top of the slope, between the residences and the break in slope (Figure 1). 
Figure 1 - Plan view of instrument locations at Laguna Niguel
An inclinometer measures its own rotation and, therefore, the rotation of the structural element or portion of ground to which it is connected.  These instruments are particularly adept at measuring movement of landslides occurring on circular slip surfaces.  However, even landslides that are predominantly translational will produce tilts that are easily detected with conventional tiltmeters or inclinometers.  If a slope is moving, tiltmeter surveying can determine the direction of movement, delimit the areas of deformation and, in many cases, reveal the mechanism of movement (slumping, slope creep, settlement, etc.) (Figure 2).
Figure 2 - Tilt vectors show different slope failure mechanisms
In the case of the Laguna Niguel slide, the shallow inclinometers measure near-surface movements in the critical area between the free face of the slope and the house foundations.  With sensitivities on the order of 0.1mm/m, the buried inclinometers can measure small movements that are unobservable with other techniques.
The inclinometers were sanded into holes excavated with a hand-auger, and the cables were routed along the ground surface to an automated data acquisition system equipped with a phone modem and autodialer.  Threshold limits were used to trigger the autodialer and notify the project engineers via pager.  Measurements for comparison with the thresholds were taken once per minute, and recorded into the datalogger every fifteen minutes.

Upon installation and activation, the inclinometers showed a pattern of ground movement consistent with a “slump” type failure.  Inclinometers outside of the headscarp area showed no movement.  Those in the crown area rotated toward the slope.  And those in the headscarp rotated upslope, indicating backwards rotation of the slump block.  The movement was at a constant velocity for over 10 days.  Then several inclinometers showed a very gradual increase in the rate of movement over the next 7 to 10 days.  On March 15th, the 20th day after monitoring commenced, several inclinometers showed a distinct increase in the rate of movement, indicating a change in state of stability of the landslide mass (Figure 3).
Figure 3 - Recorded relative angular rotation versus time shows change in rate
The failure occurred in the early morning hours of March 19, 1998 in rather dramatic fashion (Figure 4).  At 2AM part of the fill slope failed.  Two homes were located in the headward part of the slump and were destroyed as they were carried downward on the upper part of the slump. Three other homes were left partly cantilevered over the crown and main scarp of the slump. On the 20th one of the partly cantilevered homes was destroyed as it fell over the main scarp of the slump.  As the displaced material comprising the slump moved downslope, the toe ( the most distal part) of the landslide moved against and destroyed five condominiums at the base of the fill slope.
Figure 4 - False color infrared view of Laguna Niguel landslide (from Geo-Tech Imagery Intl.)
Advance warning
The autodialer was activated at approximately 4AM on March 17th .  The increase in rate of movement was significant enough to cause the evacuation of several more houses in the vicinity of the headscarp (two houses were already evacuated when the automated instrumentation was installed).  Four days after the significant change in rate was observed, and two days after the alarm was triggered, the slope failed catastrophically.
The use of the buried inclinometers to monitor ground movement gave a minimum of 4 days advance notice of the impending catastrophic failure.  Sensitive instruments such as these can measure movements much smaller than can be observed with the naked eye.  A continuous record of instrument readings provides important information about the nature of subsurface movements, and aids engineers and earth scientists in establishing thresholds for failure monitoring.

This project shows that sensitive automated instrumentation can be used to distinguish between normal and construction induced movement of a bridge.  Real-time monitoring can incorporate modeling of normal movements to establish “baseline” behavior.  Alarming the difference between modeled behavior and measured behavior results in almost instantaneous notification of excessive construction induced movements, which streamlines the construction process.

Project Description
In June of 1999, Hayward Baker performed compaction grouting of the foundation soils beneath Laurel Street Bridge in Santa Cruz, California.  This work was performed as part of an extensive program of seismic upgrades to many of California’s bridges after the 1989 Loma Prieta earthquake.  The Laurel Street Bridge is a cast-in-place reinforced concrete structure that spans approximately 350 feet (106.7 m) across the San Lorenzo River near downtown Santa Cruz.  It is supported on a battered pile foundation.  Each of the two side spans is approximately 100 feet (30.48 m) long.  The length of the center span is 150 feet (45.72 m).

Automated Instrumentation
The project specifications limited vertical bridge deck movement to 0.1 inch during any grouting episode.  A good rule of thumb is to use an instrument with at least 20 times higher resolution than the minimum specified movement.  High-resolution tiltmeters are one of the few instruments that can reliably measure angles smaller than 1 arc second.  The tiltmeters used were Model 800 and Model 711, manufactured by Applied Geomechanics.  The tiltmeters have a published resolution of between 0.25 and 0.5 arc seconds – or 50 to 100 times smaller than the maximum allowable movement.
To convert rotation measured by the tiltmeters to displacement requires integration of the angular measurements over some finite length.  The rigidity of the structure allows for a fairly simple model for calculating displacements.  The tiltmeters measuring rotation parallel to the bridge axis were used to measure vertical movement of the bridge deck between the abutment and support piers.  For this purpose the abutment is assumed to be a fixed point, and the bridge deck is assumed to be rigid (Figure 5).  Vertical displacement (heave) is then calculated by assuming the angle of rotation, theta, measured by the tiltmeter is occurring over the entire span.  Heave (h) is therefore calculated as h=(70ft)*sin(theta), where theta is the angle measured with the mid-span tiltmeter. 
Figure 5 - Math in action.  The simplified model for calculating bridge deck displacement.

The tiltmeters were monitored continuously using a Campbell Scientific CR10X datalogger.  Alarm thresholds were used to activate a strobe light in the event of excessive movements.  Four of the six tiltmeters were installed near the joining of the support columns and bridge deck to provide a first indication of movement transferred through the footing to the deck.  Two of the tiltmeters were installed along the span midway between the footing and abutment to measure changes in deck elevation (Figure 6).
Figure 6 - Location of tiltmeters used to measure bridge deck movement

Figure 7 shows the results obtained from tiltmeter 21, mounted on the eastern span during grouting beneath the east footing.  The tiltmeter shows excellent correlation to the average of four vertical survey points on the bridge deck throughout the 60+ day period of monitoring.  However, the tiltmeter is able to accurately measure displacements less than 0.02 inch (0.5mm).  This is approximately 10 times better precision than that available using conventional surveying.  However, the real benefit to this approach is the ability to measure and respond to bridge movement in real time.
Figure 7 - Calculated displacement from Titlmeter 21 compared to average of 4 surveying points
Real Time Modeling
The sinusoidal nature of the data obtained from the tiltmeters is the thermoelastic expansion and contraction of the bridge due to diurnal temperature changes.  After the onset of baseline monitoring, it was discovered that the normal daily movement of the bridge was about the same as the specified maximum allowable movement.  Distinguishing the normal daily movements of the bridge from those caused by the compaction grouting turned out to be the most challenging aspect of the job.
In this instance it was decided to model the diurnal bridge motion with the simple sine wave function that includes parameters to adjust the amplitude, phase, and symmetry (skewness) of the waveform.  This is relatively easy to program within the datalogger and results in alarms that are responsive to grout-induced movement (Figure 8).
Figure 8 - Modeled diurnal bridge movement used to establish baseline for alarming
  Periodic adjustment of these parameters was necessary to account for variations in the diurnal behavior – caused for instance by the increased firmness of the foundation as the grouting proceeded.  The program was written to activate a flashing light when the difference between the model and the measured values exceeded 0.1 inch.  The flashing light was a signal to the grouting operators to cease pumping within the current stage and move up to the next stage.  After five minutes the program turned off the light and “re-zeroed” the alarm threshold by bringing it in conformance with the current tiltmeter reading.

Automated data acquisition systems allow for instruments to be sampled at any rate - typically multiple samples per minute, per hour, or per day.  Measurement accuracy is improved, and data can be remotely processed to provide useful information to the project team.  Other benefits of automated data acquisition include:
  • Human errors associated with manual reading and data transfer are virtually eliminated.
  • Data collection can be performed easily at remote site locations and in bad weather.
  • Data are available 24 hours per day.
  • Precise timing of structural and Geotechnical events makes it possible to correlate them with external factors such as rainfall, earthquakes, grading and repairs.
  • Continuous monitoring means that critical changes can be detected quickly, so that action can be taken before adverse conditions become worse.
  • Automated data acquisition systems can be programmed to monitor threshold values and rates of change and, therefore, can issue automatic warnings when predetermined limits are exceeded.

Wednesday, March 11, 2015

Cold Circuits versus the Naked Eye

Musings on Automated Monitoring in the Digital Age
I was introduced to the practice of "automated monitoring" in the Civil/Structural profession where we connected sensors that measured displacement, tilt, strain or water level to computers to monitor physical processes for primarily advanced warning purposes.  For example we monitored slopes and excavations for failure, bridges and buildings during construction and dams for post earthquake displacement.  Back when we started we were just learning about collecting digital measurements and finding timely ways to turn them into usable information.  The Internet was in it's infancy before broadband and mobile appliances.  At that time you could hear an interesting argument going on in publications and at conferences between those of us in the new school who were excited by what we saw in the digital signature of an analog instrument, and the old school who were trained to 'observe' with the physical world with the naked eye. The Master of the school of observation was a man by the name of Karl Terzaghi.  I call him a Master because he was a great, great scientist and engineer who made a huge mark on the practice of civil engineering and geology (see for example Karl Terzaghi: The Engineer as Artist
by . He was a renaissance man and natural observer - one who tried to see the physical world without prejudice and used his technical expertise and human understanding - intuition perhaps - to solve many engineering problems.  His teachings and books informed an entire generation of geotechnical engineers, and I came along at the beginning of the digital age and was taught by his pupils.  One foot in flesh-and-blood engineering and one foot in the digital age.
Many engineers back then were concerned that automated monitoring results in a “black box” approach where visual examination and experience are replaced by cold circuits and relays to provide advance warning of potential problems.  I suppose this could happen.  At a minimum we would have to acknowledge that communication technologies are not infallible and waiting for an alarm transmitted through some wires and not going out and using active observational techniques would be foolish.  But Terzaghi and his students advocated one to go out and look for problems.  He thought that was the best way to find issues before they became big problems.  
Another point of view argued at that time was that data obtained from continuous monitoring can provide the engineer with new information about the behavior of the structure being monitored that is not apparent to the naked eye.  This added perspective can actually increase the engineer’s knowledge base, giving them a deeper understanding of structural response.  Furthermore, the utilization of real-time monitoring systems in combination with advanced analysis tools enables designers to take economic advantage of these new insights without sacrificing construction safety.
Are cold circuits better than flesh-and-blood?  Ironically in a few short years we've probably travelled to the other end of the spectrum.  Walking across a busy street with our eyes glued to our progress on a digital map?  Hopefully not that bad - although I think I've seen it.  Perhaps it wouldn't be a bad idea to study the teachings of people like Terzaghi and Ralph Peck even in non-engineering fields to remind ourselves of what the naked senses are and how they have been applied to accomplish some incredible things.  Automated monitoring has definite benefits to the welfare of our society and environment.  But we should get outside and feel our environment at the same time.