The NEESgrid Experience: 2000 - 2003

Post on 23-Jan-2016

34 views 0 download

description

The NEESgrid Experience: 2000 - 2003. Tom Finholt School of Information University of Michigan - PowerPoint PPT Presentation

Transcript of The NEESgrid Experience: 2000 - 2003

SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN

The NEESgrid Experience:2000 - 2003

Tom Finholt

School of InformationUniversity of Michigan

This work was supported primarily by the George E. Brown, Jr. Network for Earthquake Engineering Simulation (NEES) program of the National Science Foundation under award number CMS-0117853. Support was also provided by the National Science Foundation through the ITR program under award number IIS-0085951.

SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN

Outline

The earthquake engineering community The George E. Brown, Jr. Network for

Earthquake Engineering Simulation NEESgrid – the collaboratory element of

NEES Challenges & Successes The field of dreams

SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN

The earthquake engineering community

Research– University-based– Funded by NSF and industry– Focus on simulation

• Physical models (e.g., reduced scale specimens)• Numerical models (e.g., finite element analysis)

Practice– Professional firms– Structural engineering (e.g., earthquake remediation)– Formulation of uniform building codes– Lifelines (e.g., ensure survival of roads, gaslines,

power distribution)

SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN

Bhuj, India. One of the towers of this apartment complex totally collapsed,and the central stairway leaned on another building of the complex.

Photo courtesy of Dr. J.P. Bardet, University of Southern Californiahttp://geoinfo.usc.edu/gees/RecentEQ/India_Gujarat/Report/Damage/Bhuj/Bardet_Feb18.html

SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN

Instruments Structural

– reaction walls– shake tables– field test

Geotechnical– centrifuges– field test

Tsunami– wave basins

SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN

Shake table: Nevada, Reno

SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN

Reaction wall: Minnesota

SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN

Centrifuge: UC Davis

SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN

Wave basin: Oregon State

SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN

Survey methods

Technical surveys– Response from fourteen of fifteen sites

Practice survey (2002)– Administered to 444 engineers– 187 responses (42%)

• 11% non-NEES equipment sites• 9% women• 56% students• 39% from Year 1 (26% repeat response rate)

SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN

Practice survey: Data use

0% 20% 40% 60% 80% 100%

Do you use data you collect? (2001)

Do you use data you collect? (2002)

Do you use data collected by othersworking with you? (2001)

Do you use data collected by othersworking with you? (2002)

Do you use data collected by othersindependent of you? (2001)

Do you use data collected by othersindependent of you? (2002)

Not usually or never

sometimes

usually or always

SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN

Practice survey: Collaboration2001 2002

Item Mean SD Mean SD

Number of collaborations you are currently involved with

2.5 6.1 2.3 4.7

Number of collaborations with remote participants

1.4 3.2 1.3 3.6

Number of collaborators on your primary collaboration

5.7 7.1 6.1 7.3

Number of collaborators from prior collaborations in primary collaboration

1.6 3.1 1.7 3.0

SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN

George E. Brown, Jr. Network for Earthquake Engineering Simulation NSF Major Research Equipment and Facility

Construction award (MRE) $82 million, 2001-04

– $10 million for system integration (NCSA, ANL, USC-ISI, Michigan, Oklahoma)

– $2 million for consortium development (CUREE)– $60 million for new equipment sites

• 3 shake tables (Buffalo, Nevada-Reno, UCSD)• 2 centrifuges (RPI, UC Davis)• 5 reaction walls (Berkeley, Buffalo, Colorado, Illinois, Lehigh,

Minnesota)• 3 field test (Texas, UCSB/USC/BYU, UCLA)• 1 lifeline (Cornell)• 1 tsunami (Oregon State)

SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN

                  

                                   

               

                                                   

SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN

NEES expectations Community survey (2002)

• Stratified on region (six regions – CA, NW, SW, MW, SE, NE) and sub-field (e.g., structural, geotechnical etc.)

• EERI membership roll and key tsunami mailing lists as sampling frame

• n = 361 (99 responses; 27%)

Workshop survey (2002)• Registered participants in nineteen of the twenty regional

workshops– The survey administration was not ready for the first

workshop held in Charleston, SC

• n = 287 (260 responses; 91%)

SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN

Summary of survey results Workshop and community respondents share

consensus about NEES expectations– …but level of agreement is much higher among

workshop respondents Access to experimental data is paramount

– …and is more important than access to experimental facilities for both workshop and community respondents

Producing standard data formats is a critical goal Workshop respondents skewed toward

academic, structural earthquake engineers

SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN

Why is NEES exciting? Funding

– scientists recruited through the NEES equipment site program and through the NEES grand challenge program

Recognition– innovative system– highly visible to earthquake engineering and computer

science communities (e.g., SC 2002 demo) Novel capabilities

– first operational use of Globus/OGSA technology– “hybrid” operations -- combining numerical and physical

simulations

SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN

NEESgrid – the collaboratory element of NEES

Primary– Shared instrument– Community data system

Secondary– Distributed research center– Virtual community of practice

SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN

SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN

Technology involved Globus/OGSA (USC ISI and ANL)

– GSI and Gridftp CHEF (Michigan) Telepresence systems (ANL) Data repository (NCSA) Deployment, operations, and support (NCSA)

SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN

What’s A Grid?

http://http://

http://http://

Web: Uniform access to documents

Grid: Flexible, high-performance access to resources for distributed communities

Sensors andinstruments

Data archives

Computers

Softwarecatalogs

Colleagues

Source: Harvey Newman, Caltech

              

SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN

NEESgrid High-Level Architecture

Capabilities– data, simulation, collaboration, visualization, telepresence

validated and improved via EA demonstration scenarios and delivered via– APIs and tools for users– services and interfaces at equipment and resource sites– management services for operation

System Resources: Compute, Network, Data Storage, Testing Sites

Grid Resource Management Middleware

APIs Supporting Higher Level Information Services

Collaboration Services

Data & InformationServices

TelepresenceServices

Simulation & Analysis Services

Portal, Web and 3rd Party End User Interfaces

APIs, Tools and Libraries Supporting End User Interfaces

SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN

NEESgrid interface

SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN

System evolution Scoping study

– NCSA, ISI, ANL, UM Alpha 1.0

– demonstrated at UNR, November 2002– released February 2003

Alpha 1.1– released June 2003

MOST experiment– real-time control of reaction wall from numerical

simulation– UIUC and Colorado, July 2003

SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN

gx

Multi-Site, On-Line Simulation Test (MOST)July 2003

UIUC

Experimental Model

gx

f1

m1

NCSA Computational Model

m1

f1f2 f2

Colorado

Experimental Model

gx

SAC Consortium Benchmark Structure

SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN

Multi-Site, On-Line Simulation Test (MOST)

ColoradoColoradoExperimental Model

gx

f2m1, 1

F2

F1

e

gx

=

gx

f1, x1

UIUCUIUC Experimental Model

gx

m1

f1 f2

NCSANCSA

Computational Model

SIMULATIONSIMULATION

COORDINATORCOORDINATOR

NEESpop NEESpop

NEESpop

UIUC MUST-SIM•Dan Abrams•Amr Elnashai•Dan Kuchma•Bill Spencer• and othersColorado FHT•Benson Shing•and others

SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN

MOST Column Test Specimens

Illinois Test Specimen

Colorado Test Specimen

SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN

Challenges

Confused lines of authority

Cultural differences– Different jargon– Different world views

SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN

Confused lines of authority

NEES program (2000-2004)– NEESgrid, system integration– NEES Consortium Development– 15 equipment sites

NEES Consortium (2003-2014)– Board and Executive Committee– Standing Committees

O rganization of the NEES Program , 2000-03

S ystem Arch itec tsC arl K esse lm an & Ian Fo s ter

C o m m un ity R eq u irem en tsT o m Fin ho lt

S im ula tionK im M ish

D eplo ym ent an d Tra in ingR an dy Bu tler

P ro jec t D irec torD an R eed

N EE S g rid

IT and N etw ork ingC h err i P ancake

B u s in ess P lanR o bert N igb or

C o llab o ra tory Vis ionS teve M ah in

E d uca tion an d O u treachS h aro n W o od

P ro jec t D irec torR o bert R e ith e rm an

N EE S C o n so rtium D eve lo pm ent

U C S an D iegoA n dré F ilia trau lt

B u ffa loM iche l Bru neau

N evad a , R enoIan Bu ck le

S h ake T ab les

B u ffa loM iche l Bru neau

U C B erke leyJack M oeh le

L ehighJ im R ic les

M inn eso taC ath y F ren ch

Illin o isA m r E ln ash ai

C olo radoB en so n Sh ing

R eaction W alls

U C L AJo h n W a llace

F ie ld T est

S tru ctu ra l S ites

R P IR icard o D ob ry

U C D av isB ru ce Ku tter

C entr ifug es

T exasK en Sto koe

B YU /U C SB /U S CL es Yo u nt

F ie ld T est

C o rn e llH arry S tew art

L ifeline

G eo tech n ical S ites

O rego n S ta teS o lo m o n Y im

W ave B as in

T su nam i S ite

N EE S Eq u ipm en t S ites

N SF NEE S pro g ramJo y P au sch ke , Pro g ram M anag er

BYU/UCSB/USCLes Youd

O rganization of NEES Consortium , 2003-2014

T B DExecu tive D irec tor

Ian Bu ck lePresid ent

B ru ce Ku tterV ice Pres id ent

R icard o D ob ry

A n dre Re inh o rn

K enn eth S tokoe

So lo m o n Y im

Eq u ipm ent S ite D irec to rs

Jam es W ig ht

H e lm u t Kraw in k ler

T h alia An agn os

In s titu tio n al Directo rs

R o bert N igb orSecre ta ry

G rego ry D e ier le in

Sh aro n W o od

In d iv id ua l Directo rs

C raig Co m artinT reasurer

K im M ish

C arl S tepp

B o ard -elec ted Directo rs

N EES Co n so rtiu m , In c.B o ard of Directo rs

O rganization of N EES C onsortium , 2003-2014

T B DE xecu tive D irec tor

M arc Eb erh ard

R o berto Leon

S h ir ley D yke

Jo An n Bro w n ing

E lec ted b y a ll ca tego ries

R o ss B ou lan g er

S tep hen M ah in

A n dré F ilia trau lt

D av id S and ers

Jo h n W allace

T B D

E lec ted by equ ip m en t s ites

T B D

T B D

B o ard ap po inted

S h ared Use Co m m ittee

Jacob o Bie lak

F in ley Ch arn ey

P h ilip L .-F L iu

E lec ted b y a ll ca tego ries

T B D

T B D

T B D

B o ard ap po inted

D ata C o m m ittee

B o ard ap po inted

IT C om m ittee

B o ard ap po inted

E d u cation C om m ittee

T B D

T B D

T B D

B o ard A p po inted

T B D

T B D

T B D

E lec ted b y a ll ca tego ries

N o m in atio n Co m m ittee

N E E S C o n so rtiumB o ard of Directo rs

SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN

Earthquake engineers – in Hofstede’s scheme Power distance

– Hierarchical– Bias toward seniority

Individualist– “My lab is my empire”– Solo PI model

Masculine– Adversarial– Competitive

Uncertainty avoidance– Highly skeptical of new technologies– Extremely risk adverse

SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN

Grid specialists – in Hofstede’s scheme Power distance

– Egalitarian at development level– Bias toward talent

Collectivist– Use the Internet to create worldwide communities– Project model

Masculine– Adversarial– Competitive

Uncertainty avoidance– Extremely open to new technologies– Extremely risk seeking

SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN

Agreeing on termsTerm What grid specialists

heard What earthquake engineers heard

“user” HPC users

earthquake engineers

“community” NEES awardees broad array of earthquake engineers, including researchers and practitioners, in the diverse settings where earthquake engineering occurs (centers, under-represented institutions, under-resourced institutions)

“requirements” Description of high level system architecture

Description of detailed user requirements and their relationship to functional specifications

SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN

How earthquake engineers think

Customer Need

Customer Requirements

Requirements Analysis

Structure Design

Structure Construction

Structure Acceptance

Structure Operations

Customer Needs Assessment

Design, Engineering, and

Development

Deployment and Operations

Progress

Feedback

SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN

Prelim Analysis

START

Prelim Design

Prototyping

Proto Evaluation

System AnalysisSystem Design

EvaluationImplementation

More Iterations

How grid specialists think

SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN

Successes Scientific

– First teleobservation of shake table (November 2002)

– First data saved to repository (November 2002) Community

– NEES Consortium incorporated (January 2003)– First NEES Consortium meeting (May 2003)– Use of prototype tools

Future– MOST experiment, July 2003– Operational collaboratory October 1, 2004

SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN

NEESgrid November 2002 Demonstration Earthquake simulation at

UNR early adopter site– biaxial shake table with

cameras and instrumentation– 40% scale model of a two

span bridge• concrete slab over steel

girders

Bridge model instrumented with sensors– strain gauges, load cells,

displacement, acceleration

SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN

NEESgrid November 2002 Demonstration

CHEF-based collaborative framework

Electronic notebook for data recording

Experiment management tools– data and metadata

Streaming data and video– teleobservation of experiment– data channels from sensors

Data analysis and visualization

SAP2000 model

Tele-observation

SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN

Prototype tool use

H.323 videoconferencing

Worktools

Placeware

SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN

NEES MCU Usage

0

100

200

300

400

Jan

Feb

Ma

rch

Ap

ril

Ma

y

Jun

e

July

Au

gu

st

Se

pte

mb

er

Oct

ob

er

No

vem

be

r

De

cem

be

r

Jan

ua

ry

Feb

rua

ry

Ma

rch

Ap

ril

Ma

y

Jun

e (

est

)

Month

Po

rt H

ou

rs Commons

Michigan / USC

Total

Use of H.323 videoconferencing

UNR Demob c d

a = initial ES-TF meeting; b = ES-TF meeting time changed; c = succession to new ES-TF chair; d = change to biweekly ES-TF meetings

a NSF LAN meetings

SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN

Use of H.323 videoconferencingGeneral vs. specific meeting topics, ES-TF sessions,

1/24/02 to 6/12/03

0

5

10

15

20

25

30

35

40

45

0

90

18

0

27

0

36

0

45

0

Time (in days since first ES-TF session)

Nu

mb

er

of

ES

-TF

se

ss

ion

s

Specific topic

General topic

Total

SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN

Use of H.323 videoconferencingSession leadership, ES-TF sessions, 1/24/02 to 6/12/03

0

5

10

15

20

25

30

35

40

45

50

0 90 180

270

360

450

Time (in days since first ES-TF session)

Num

ber

of E

S-T

F se

ssio

ns

Total

ES-led

SI-led

Cancelled

SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN

Use of Worktools

Cumulative Frequency of Resources Uploaded to ES-TF Space

0

5

10

15

20

25

30

35

40

Time

Nu

mb

er

of

Re

so

urc

es

SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN

Use of WorktoolsCumulative Frequency of Messages sent to ES-TF

Mailing List

0

50

100

150

200

250

300

350

400

450

Time

Nu

mb

er

of

Me

ss

ag

es

SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN

Use of WorktoolsCumulative Frequency of Resources Uploaded to UR Workshop

0

10

20

30

40

Tim e

Nu

mb

er o

f R

eso

urc

es

SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN

Use of WorktoolsCumulative Frequency of Message sent to UR

Workshop Mailing List

0

10

20

30

40

50

Time

Nu

mb

er o

f m

essa

ges

SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN

The field of dreams

I was sitting on the verandah of my farm house in eastern Iowa when a voice clearly said to me, “If you build it, he will come.” – Ray Kinsella in Shoeless Joe, by W.P. Kinsella

Image source: http://www.fieldofdreamsmoviesite.com/

SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN

Building it so they will come…

Balance contributions (pp. 50-51 in the Atkins report)– Weight domain science too heavily?

• Overemphasize procurement of existing technologies• Computer scientists become viewed as “merely”

consultants and implementers

– Weight computer science too heavily?• End user needs insufficiently addressed• Emphasis on novelty at the expense of usability and

stability

Source: Atkins report --http://www.communitytechnology.org/nsf_ci_report/

SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN

Innovation

Extrapolation

Social &Technological

Forces•digital libraries

•computational grids

•video conferencing•electronic journals

•distance learning

•electronic commerce

•collaboratories

•community networks

•e-science

Innovation vs. extrapolation

SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN

Building it so they will come…

Give users objects to think with (scenarios, mock-ups, prototypes)

Be patient…let users convince themselves

Know where you’ve been (collect baseline data) and what’s changed (collect data as you go along)

SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN