Figure 1. London Canal Levee BreakCourtesy of http://911review.org/Hurricane_Katrina
Hurricane Katrina produced a massive surge of water on the U.S. Gulf Coast that overtopped and eroded away more than 50 levees and floodwalls that compose the New Orleans and Southeast Louisiana Hurricane Protection System (Figure 1). In the aftermath of this catastrophic disaster, several investigations were conducted into the performance of the levees and floodwalls and the causes of the damage and failures.
For one study, the U.S. Army Corps of Engineers assembled an Interagency Performance Evaluation Task Force (IPET) that included Rensselaer Polytechnic Institute.1 The task force used centrifuge models of the levees on 17th Street, London Avenue, and the Orleans canals to provide detailed insights into the mechanisms that led to the breaches.
A geotechnical centrifuge consists of an arm that rotates a basket containing a payload around a central vertical axis. The payload is a geotechnical specimen or model constructed within a strong box that is subject to high acceleration forces during rotation.
The behavior, including the failure modes, of any structure or material dependent on its self-weight can be simulated on a reduced size and time scale using centrifuge testing.2 Geotechnical centrifuges are commonly used for modeling the response of geotechnical material or structures such as soil, dams, and foundations to natural and man-made hazards like earthquakes, floods, and explosions.
As the behavior of soil is stress dependant, it is crucial in studying the performance of a geotechnical system to ensure that the correct stresses are applied to each element in the levee and foundation. This is difficult to achieve in a scale model under earth’s gravity alone because the weight of the model is not sufficient to bring the soil to the correct state.
In a centrifuge, however, a model at a scale of 1/N can be subjected to a steady acceleration field equivalent to N times earth’s gravity. In this state, the same stress conditions that exist in the field can be effectively reproduced at all points in the model.
By applying appropriate data acquisition techniques, these stresses can be verified throughout the model during centrifuge testing. Moreover, detailed information can be collected on the response of a geotechnical system including measurements of water pressure in the ground at different locations, the movement of a structure such as the flood wall and ground surface, and video imagery of a sequence of events leading to a breach.
Data Acquisition Hardware
Rensselaer Polytechnic Institute’s Center for Earthquake Engineering Simulation maintains a 150g-ton centrifuge with a 3.0 meter arm radius and maximum payload of 1.5 ton spinning at 100g and maximum acceleration 150g.
To capture the necessary data, Rensselaer uses a highly configurable, high-performance DAQ system developed by Bloomy Controls. (www.bloomy.com/centrifuge) As shown in Figure 2, the system includes National Instruments’ PXI and SCXI hardware to condition signals from various sensor types. The controller runs Bloomy centrifuge DAQ application software and communicates to the control room PC via Ethernet.
Figure 2. Block Diagram of the Centrifuge DAQ System Hardware
Each DAQ module is cabled to a SCXI signal-conditioning chassis. The DAQ modules perform high-speed analog-to-digital conversion of the conditioned analog signals and analog output control of an external shaker table and camera. The SCXI chassis contain modules for conditioning strain gages, pore pressure transducers, accelerometers, LVDTs, thermocouples, and analog voltages. A single model can be instrumented with more than 128 sensors.
The SCXI chassis are mounted in the centrifuge basket adjacent to the steel box containing the model. This helps to minimize wire length and signal attenuation between sensors inside the model and the SCXI chassis. Consequently, the SCXI chassis are subject to similar accelerative forces as the model.
The PXI chassis mounts in a cabinet at the center of the centrifuge and connects to the control-room computer and the LAN via an optical fiber running through a rotary optical coupling. The centrifuge DAQ software is operated from the control-room computer via a Windows remote desktop.
Signals from multiple SCXI chassis are synchronized using a common sample clock shared by all DAQ modules via the PXI timing and synchronization bus. Synchronization is very important in centrifuge testing because acceleration compresses the time scale of geological responses and events.
Data Acquisition Software
The centrifuge DAQ GUI allows you to select the type of sensors conditioned by the SCXI module using the sensor selection control. Supported sensors include accelerometers; LVDTs, quarter-, half-, and full bridge strain gages; pore pressure sensors; thermocouples; and analog voltages. Each sensor selection has a corresponding form that allows you to configure the channels of a module based on the applicable parameters.
For strain gages, the bridge completion, excitation voltage, and filter setting may be individually set for each channel, and the lead resistance, nominal resistance, and gage factor may be selected for each module. Additionally, the channel name, location within the model, sensor serial number, scaling factor, and engineering units are common settings applicable to all sensor types.
A waveform chart can be displayed that represents live data from the active sensors connected to any SCXI module. It assists with preparation of the model and identification of any problems with the sensors during centrifuge spin-up.
Data acquisition is initiated via the start button and ended either manually via the stop button or automatically based on completion of the acquisition time interval. The sample rate is adjustable on the fly using the sample rate control. Also, you can control pumps and valves simulating geological events such as flooding.
After testing, you can open any data file, view selected channels on one graph or on separate graphs by sensor type, and convert the data from binary to text files or XML. Figure 3 shows the response of two pore pressures to increasing water levels at the levee.
Water Level at the Levee
Levee Tests
Rensselaer researchers built small-scale models of typical levee sections from several locations in New Orleans including the 17th Street Canal and the London Avenue Canal.3,4Figure 4. The models were 50 times smaller than the actual levees and tested in the centrifuge spinning at 50g acceleration, accurately simulating the field conditions during Katrina. The models of the levees and soil profiles were constructed of clay, metal, and peat as shown in
Figure 4. Centrifuge Model Container and Setup
During the test, water was pumped from the reservoir through a pipe into the model at the canal side of the levee to simulate flood conditions. Pore pressure sensors were installed in each soil layer to measure soil pressure, and laser displacement sensors were installed behind the sheetpile to measure the rotation of the sheetpile.
The centrifuge DAQ software was configured to acquire approximately 25 pore pressure sensors, three laser displacement gages, and two water-level sensors, and a relay was configured to control the water pump. Data was acquired at 100 samples/s while the water level in the canal side of the levee was raised at 100-s intervals for 30 minutes.
The IPET study shows that, in the 17th Street model, the wall in the middle of the earthen structure started to move before the water reached the top. The weak clay directly underneath the peat layer sheared first, causing the whole levee to slide.
Summary
According to officials with the IPET Geotechnical Structure Performance Analysis Team, the independent centrifuge modeling experiments conducted at Rensselaer using Bloomy Controls’ centrifuge DAQ system greatly assisted with the repairs and improvements of the New Orleans hurricane protection system following Katrina. The Rensselaer centrifuge experiments, coupled with those conducted by the U.S. Army Corps of Engineers, discovered and validated floodwall failure mechanisms. These lessons learned were factored into the system improvements to provide much better protection for the citizens of New Orleans.
References
- Performance Evaluation of the New Orleans and Southeast Louisiana Hurricane Protection System, U.S. Army Corps of Engineers, June 1, 2006.
- Turner, P., Geotechnical Centrifuges, University of Cambridge Department of Engineering.
- Sasanakul, I., Sharp, M., Abdoun, T., Ubilla, J., Steedman, S., and Stone, K., “New Orleans Levee System Performance During Hurricane Katrina: 17th Street Canal and Orleans Canal North,” Journal of Geotechnical and Geoenvironmental Engineering ASCE, May 2008.
- Sasanakul, I., Sharp, M., Abdoun, T., Ubilla, J., Steedman, S., and Stone, K., “New Orleans Levee System Performance During Hurricane Katrina: London Avenue and Orleans Canal South,” Journal of Geotechnical and Geoenvironmental Engineering ASCE, May 2008.
About the Authors
Peter Blume is founder and president of Bloomy Controls. He is the author of The LabVIEW Style Book and has published articles for various trade magazines. e-mail: [email protected]
Greg Burroughs is senior project engineer with Bloomy Controls. He has 16 years experience developing systems for automated test, data acquisition, and control and is a National Instruments certified LabVIEW architect and professional instructor. Mr. Burroughs graduated from Rochester Institute of Technology with a B.S. in electrical engineering. e-mail: [email protected]
Bloomy Controls, 839 Marshall Phelps Rd., Windsor, CT 06095, 860-298-9925.
Inthuorn Sasanakul, Ph.D., is a research assistant professor and technical manager for the Center for Earthquake Engineering Simulation in the Civil and Environmental Engineering Department at Rensselaer Polytechnic Institute. Previously, she was a graduate research assistant at Utah State University and the University of Texas and a project assistant in the Asian Center of Soil Improvement and Geosythetics at the Asian Institute of Technology in Bangkok. Dr. Sasanakul is the author of many publications and a member of several professional societies. Rensselaer Polytechnic Institute, Center for Earthquake Engineering Simulation, Troy, NY 12180, 518-276-6944, e-mail: [email protected]
January 2007