Equipment Site Description

Equipment Site Description

The NEES Equipment Site at the University of California, Berkeley, nees@berkeley ES, is one of the 15 equipment site comprising the George E. Brown, Jr. Network for Earthquake Engineering Simulation (NEES). nees@berkeley is designed to support the development and active research use of a new generation of geographically distributed hybrid testing methods. Given a fast and reliable NEESgrid network, the limits of laboratory size, computer location and geography virtually disappear, enabling hybrid and distributed simulations of structures at scales and levels of complexity not possible today. The structural testing methodology and hardware, information technology and operation components of the nees@berkeley ES are described in the following sections.

Hybrid Simulation Method

Response of a structure to an earthquake is the result of complex interaction of all elements of that structure. Today, our understanding of earthquake response of individual sub-structures is quite good. However, we do not fully understand how interactions among sub-structures define the response of the entire structure. The nees@berkeley ES is designed to enable modeling and simulation of the interaction among different sub-structures. A model in such hybrid simulation consists of sub-structures, many instantiated in powerful computers as finite element sub-models, and some instantiated as physical specimens in laboratories representing sub-assemblages of the prototype that are too complex to model numerically. Hybrid simulation is conducted using combined numerical analysis and test control procedures that smoothly integrate physical and computational sub-structures into a single model. The interaction among the sub-structures in a hybrid test occurs in the network that binds together the computers and laboratory facilities where the sub-structures are instantiated. Such simulation is rooted in the pseudo-dynamic testing method (PDTM) developed since the 1970’s at UC Berkeley and elsewhere. Even though the hybrid simulation test is happening on the Ethernet, PDTM integration procedures and dynamic testing similitude rules still apply. Our challenge was to build the technology to enable such hybrid testing and test result interpretation on the NEESgrid network.

The technology for geographically distributed hybrid testing is based on a three-loop hybrid test control architecture. The innermost loop is the conventional PID actuator control loop updated at a 1kHz rate. The outermost loop is the conventional time-step integrator used in PDTM updated at 0.1kHz rate. The intermediate loop, comprising a predictor and a corrector, acts as a buffer between the inner and outer loops. When a hybrid test is run in a local mode (with all sub-structures connected using a high-speed shared memory network at nees@berkeley ES), the three-loop control architecture enables fast testing at real-time rates. However, the principal advantage of the three-loop architecture is that it lends itself to implementation on the NEESgrid network. The fast-rate communication between the inner and intermediate loops must be done locally using a shared-memory network, but the slower-rate communication between the outer and the intermediate loop may be shifted to NEESgrid relying on 1Gb/sec Ethernet spanning the NEES Equipment Sites.

However, such control architecture is not sufficient to enable geographically distributed testing. The time-delays or unexpected outages on the Ethernet introduce a random element into the control loop. An event-based control strategy, event-based on a finite state machine, was develop to encapsulate the logic required to make the geographically distributed hybrid test robust enough to be completed on the Ethernet. A proof-of-concept test using the sub-structure configuration, network deployment and event logic shown in Figure D.6-1 was successfully accomplished during the May 2003 NEES Awardees Meeting at Park City, UT.

m2=0.1 kips/g m1=0.1 kips/g d2 d1 m2=0.1 kips/g m1=0.1 kips/g d2 d1

Figure D.6-1: Proof-of-concept for geographically distributed hybrid simulation.

Structural Testing Equipment

The nees@berkeley equipment site leverages the capabilities of existing testing facilities located at the University of California, Berkeley Richmond Field Station. It builds on an existing 6.1x18.3m strong floor and an existing 4-million-pound axial compression-tension machine (Figure D.6-2a). A new reconfigurable strong wall, built using 24 0.5m-tall post-tensioned blocks, can be configured in three principal ways to enable: 1) multiply sub-structured testing (Figure D.6-2b); 2) high-axial-load testing (Figure D.6-2c), and

N E T W O R K PC Integrator Algorithm Analysis DSP Signal Generation j i d+1 m r m d PID Servo-hydraulic Actuator control Load cell Remote DSP Signal Generation j i d+1 m r m d PID Servo-hydraulic Actuator control Load cell Remote

D target reached/updateForce()

free_vibration

hold

extrapolate

D update

interpolate

e

D update

TimeOut

TimeOut

augment an existing set of static actuators. These actuators are powered by a hydraulic pump system with a 2000 liter accumulator system and controlled by a hybrid simulation controller with 8 independent control channels enabling quasi-static, near-real-time and real-time hybrid simulation, or by a 4-channel MTS Flex-test system enabling conventional static and quasi-static Flex-testing. Matlab/Simulink models of the reaction wall and actuators are available for computer-only pre-test simulation of hybrid simulation tests. Conventional deformation and force measuring instruments is available to support data collection using a new 128-channel data acquisition system that features a hardware-based data ring buffer and a Matlab and LabView user interface. A video-as-data collection system is also available. This system comprises high-resolution cameras (6 video and 4 still cameras), image storage and editing facility, as well as time-stamping using a network time protocol server to synchronize images and test controller commands.

(a)

(b)

(c)

(d)

Figure D.6-2: nees@berkeley reaction floor and wall configurations.

Information Technology Equipment

The nees@berkeley ES is connected to the Ethernet using a 1Gb/sec fiber-optic network. A dedicated router located at the Richmond Field Station is at the outer edge of the UC Berkeley network and thus close to the Ethernet backbone, a feature crucial for minimizing the communication delays on NEESgrid. Behind the router, the nees@berkeley ES traffic is split into three sub-nets. One sub-net is hosting a wireless network for NEES research use during test preparation and local observation. The second subnet is dedicated to the video and telepresence equipment, while the third sub-net is dedicated to the hybrid test control and data collection (Figure D.6-3). By dividing the network in such way, the mission critical hybrid test control sub-net can be insulated from the bandwidth-hungry video sub-net and casual user traffic.

Figure D.6-3: nees@berkeley local area network.

The hybrid simulation sub-net comprises the hybrid simulation controller and interfaces the controller local shared-memory network with the high-speed Ethernet and NEESgrid. The equipment on the telepresence subnet enables remote researchers to interact with the staff at nees@berkeley before, during and after the test and remote users to observe a NEES test. The NEES-wide teleobservation system based on Axis/Broadware hardware and Chef workgroup software is provided for use by remote observers. NEES researcher can use a Polycom video conferencing system for interaction with the staff. Finally, a robot avatar provides the most direct method for remote interaction. This robot, a mobile teleconferencing station with a laser pointer, can move under control of a remote researcher and thus provide the remote user with a personal roaming presence at nees@berkeley.

The nees@berkeley local area network provides a 1TB mirrored RAID disk array for local data and video storage facilities. Both remote and local users can utilize the facilities provided by the NEES System Integrator and nees@berkeley staff to generate the meta-data appropriate for their test, describe the collected data, and upload it to the NEES data repository. Finally, secure and redundant connection to NEESgrid makes it possible to browse and download NEES data from the NEES data repository and store it locally.

Operation of the nees@berkeley ES

The nees@berkeley equipment site operation is lead by the PI, Professor Nicholas Sitar, and the co-PI, Professor Bozidar Stojadinovic. An organization chart showing the reporting hierarchy of the nees@berkeley staff is shown in Figure D.6-4. A potential NEES researcher will contact the NEES Operations Manager at nees@berkeley to prepare and submit a NEES proposal. When the proposals are funded, nees@berkeley facility manager will interact with the Consortium to schedule and organize the tests, while the remote NEES researchers and their students attend nees@berkeley training. Such training comprises a course on hybrid simulation methods, offered by the co-PIs, and lectures on safety and equipment usage at nees@berkeley provided by the staff.

Router

NEESpop

Telepresence

Video

NEESpop

Hybrid

Simulation

Telepresence/Video Subnet

UC Berkeley Backbone

NSF NEES Integrator NEES @ berkeley PI UC Berkeley Nicholas Sitar EERC Co-PI Bozidar Stojadinovic

Admin. Asst. – New Position

System Administrator System Administrator Operation Manager Operations Manager

Networking and Communications

Hybrid Simulation System Administrator

Facilities and MTS Systems Manager

NEES Facility Project Coordination and

Management

Debra Bartling New Position Don Clyde

MTS Systems Operations Hybrid System Operation Sr. Development Engr. Sr. Development Engr.

Wes Neighbour New Position

Exp. Setup and Execution System Maintenance Asst. Research Eng.

Sr. Lab. Tech _{New Position}

David MacLam Specimen Construction and Maintenance Sr. Lab. Tech. New Position NEES Researcher

Figure D.6-4: nees@berkeley organization chart.

The main component of the outreach and information effort at the nees@berkeley equipment site is our web page: http://nees.berkeley.edu. The web page contains the documentation and examples of use of the nees@berkeley ES. It is operational now: visit to start using the nees@berkeley equipment site.