CONGRATULATIONS TO THE WINNERS OF THE

NIST–ATC BLIND PREDICTION CONTEST 

The winning team in the simple category (using simple nonlinear models intending to capture the main response parameters) includes JJ Tobolski (Project Engineer) and Zachary Treece (Senior Engineer) of Thornton Tomasetti, Chicago, Illinois. The entry utilized Abaqus software and earned the highest number of points of the 15 entries when judged against the test results.

The winning team in the comprehensive category (demonstrating the full nonlinear cyclic response of the test specimens, and intending to predict overall and local response parameters) includes Alexander Hartloper (Doctoral Assistant), Ahmed Elkady (Postdoctoral Research Scientist), and Dimitrios G. Lignos (Associate Professor) of Ecole Polytechnique Federale de Lausanne (EPFL) Switzerland. The entry utilized Abaqus software and earned the highest number of points of the 11 entries when judged against the test results.

In addition to the two winners, the contest judges would like to commend Mariyam Amir (PhD Student), K.G. Papakonstantinou (Assistant Professor), and G.P. Warn (Associate Professor) of the Department of Civil Engineering at the Pennsylvania State University. This team’s results in the comprehensive category were developed by their own code and model, developed in Matlab, and earned the second highest number of points when judged against the test results.

One representative of each of the three teams will be invited to present their methods and findings at the upcoming AISC Steel Conference (NASCC) taking place in Baltimore, Maryland on Friday, April 13, 2018.  The presentations will take place between 8:00am and 9:30am, in Session V1, Blind Prediction of Cyclic Response of Deep Wide-Flange Columns for Special Moment Frame Applications.  Also in this session, Professor Chia-Ming Uang of University of California San Diego will provide a summary of the overarching testing program. The session will be moderated by James O. Malley of Degenkolb Engineers, San Francisco.

 



NIST–ATC BLIND PREDICTION CONTEST ON DEEP,

WIDE-FLANGE STRUCTURAL STEEL BEAM-COLUMNS

SUBMISSIONS CLOSED

The National Institute of Standards and Technology (NIST) and the Applied Technology Council (ATC) blind prediction contest was conducted based on the seismic response of three deep, wide-flange structural steel beam-columns.  These members were tested quasi-statically as part of a comprehensive research program funded by NIST at the Seismic Response Modification Device (SRMD) facility of the University of California, San Diego.

CONTENTS

Purpose and Background
Contest Rules
Timeline
Awards
Provided Information, Assumptions, References
Submittal Rules
Questions and Answers

PURPOSE AND BACKGROUND

The purpose of the contest was to assess the uncertainty in the predicted response of wide-flange structural steel beam-columns by predicting various response parameters of beam-column test specimens of known dimensions, material properties, and loading protocols.

The contest was a component of the ongoing NIST-funded ATC‑106-1 project.  Since the project’s initiation in September 2015, 23 deep, slender wide-flange beam-column specimens have been tested at the Seismic Response Modification Device facility of the University of California, San Diego.  The overarching goal of the ATC-106-1 project was to determine the fundamental behavior of these structural members under constant or variable axial force loads.  The project was a continuation of the ATC-106 project that took place between 2013 and 2015 and tested 25 specimens at UC San Diego.

 

CONTEST RULES

  • Contestants may consist of individuals or teams.
  • An individual could only be involved in a single team.
  • If an individual was part of a team, the individual could not enter the competition separately as an individual.
  • A company or a research institution may have submitted several predictions.
  • Contestants submitted a single entry of one of the following two categories:
    • Simple: This category was recommended for structural engineers and researchers using simple nonlinear models intending to capture the main response parameters.
    • Comprehensive: This category was recommended for structural engineers and researchers intending to model the test specimens to capture the full nonlinear cyclic response of the test specimens, and intending to predict overall and local response parameters.
  • There was one winner for each of the two categories.
  • Although names and affiliations of participating teams were recognized publicly, results were be presented anonymously except for those of the “winning” entries.
  • Questions about the blind prediction contest or details of the column specimens could be submitted via email until December 15, 2017. Questions and answers were be posted on the web site and were updated weekly. For updates on questions and answers, please see below.

TIMELINE

  Event   Date
  Submittals Due1   January 12, 2018
  Category Winners Notified2   January 26, 2018
  Category Winners Awarded3      April 10-13, 2018

 

 

 


1
  Contestants submitted their final submittal spreadsheet on or before 11:59pm Pacific on Friday, January 12, 2018, via email here. An email acknowledging the submittal was sent to each team.
2   Category winners were notified by January 26, 2018 via email used for submittal.
3   Category winners were awarded at the AISC Steel Conference (NASCC) taking place in Baltimore, Maryland on April 10-13, 2018.  One representative of each category winner was invited to make a presentation on the techniques used (model and analysis) and challenges arising in the prediction, which resulted in a winning team.


AWARDS

Full registration to NASCC 2018 for one representative
from each category winner, as well as other prizes, sponsored by the AISC!
 
The winners will be recognized at the NASCC session
and highlighted in the ATC Newsletter.



PROVIDED INFORMATION, ASSUMPTIONS, REFERENCES

The blind prediction contest directory contains the following information:

  • Summary of contest rules
  • General information on specimens and loading protocol.
  • Material test results
  • Corrected loading protocol
  • Submission spreadsheets (see below for a detailed description)
  • The following references that summarize previous tests conducted under this program.
    • Ozkula, G., Harris, J. and Uang, C.M., 2017. Observations from Cyclic Tests on Deep, Wide-Flange Beam-Columns. Engineering Journal, 1, pp.45-59.
    • Ozkula, G., Harris, J. and Uang, C.M., Classifying Cyclic Buckling Modes of Steel Wide-Flange Columns under Cyclic Loading. In Structures Congress 2017, pp. 155-167.

Computational models can assume the column ends were fastened to rigid elements as data has been post-processed to remove the flexibility occurring at the column ends due to the deformation of tie-down rods and end plates.

SUBMITTAL RULES

The individual or team were required to use the contest submittal spreadsheet and input values as follows:

Category A. Simple–  Submit results using spreadsheet Simple_Entry.xls:

  • Questionnaire tab: Provide brief description of the method of analysis.
  • For Specimens A and B, tested under constant axial load:
    • Ratio Mmax / Mpc to two decimal places, where Mpc is the reduced plastic moment.
    • Lateral force in units of kip to one (1) decimal place versus drift ratio envelope, for various drift ratios specified.
    • Mode of primary nonlinear response.
  • For Specimen C, tested under variable axial load:
    • Ratio Mmax / Mpc to two (2) decimal places for two axial loads.
    • Positive lateral force in units of kip to one (1) decimal place versus drift ratio envelope, for various drift ratios and corresponding axial loads.
    • Negative lateral force in units of kip to one (1) decimal place versus drift ratio, for various drift ratios and corresponding axial loads.
    • Mode of primary nonlinear response.

Category B. Comprehensive – Submit results using spreadsheet Comprehensive_Entry.xls:

  • Questionnaire tab: Describe the method of analysis
  • For Specimens A and B, tested under constant axial load:
    • Ratio Mmax / Mpc to two decimal places.
    • Lateral force in units of kip to one (1) decimal place versus drift ratio envelope for the drift ratios specified.
    • Hysteretic energy in units of kip-in to zero (0) decimal places calculated between various drift ratios.
    • Column shortening in units of inches to two (2) decimal places calculated for various drift ratios.
    • Mode of primary nonlinear response.
  • For Specimen C, tested under variable axial load:
    • Ratio Mmax / Mpc to two (2) decimal places for two axial loads.
    • Positive lateral force in units of kip to one (1) decimal place versus drift ratio envelope for drift ratios and corresponding axial loads.
    • Negative lateral force in units of kip to one (1) decimal place versus drift ratio for the drift ratios and corresponding axial loads.
    • Hysteretic energy in units of kip-in to zero (0) decimal places calculated between the various drift ratios and corresponding axial loads.
    • Column shortening in units of inches to two (2) decimal places calculated for various drift ratios and corresponding axial loads.
    • Mode of primary nonlinear response.

QUESTIONS AND ANSWERS

For answers to questions regarding the blind prediction contest, please see below. Answers to questions are posted below. 

Question: I´m trying to feel the comprehensive entry sheet for specimen A. I´m a bit confused with the rows indicated in the last part on the hysteretic loops. I thought these rows will correspond to those in the loading protocol excel for the drifts indicated in the table, but when I check rows 601-667 for example, I have a drift ratio of 0.036% and 0.1747%. I calculate the drift ratio as the imposed lateral displacement divided by L=18feet.
Answer: Please refer to Corrected Load Protocol r1.0, which has been uploaded in the "04 Load Protocol" folder of the blind prediction contest directory. Your question prompted a revision of this spreadsheet.  During testing, data is acquired at discrete intervals.  This means that the points are collected near the zero drift points but not necessarily at the zero drift.  The original spreadsheet presented the data as collected, whereas the Comprehensive spreadsheet made reference to a revised spreadsheet where the zero drifts had been inserted at the corresponding locations.  The revised spreadsheet includes the added zero drift rows not only for Specimen A, but also for Specimens B and C.  The revised spreadsheet now includes color coded cells, and makes explicit reference to the key points in the hysteretic responses.  Since the applied axial force at the zero drift points was not measured, we suggest that participants linearly interpolate the axial load using the measured drift ratio and axial load points above and below where the zero drift rows were inserted.
While reviewing the spreadsheet we found that the drift ratios listed in cells D53-E57 of worksheet “Specimen C” had been mistakenly calculated using a column length of 216 in. The correct length is 212 in.  The correct drift ratios are listed in r 1.0 of the Corrected load protocol.

Question: It is not clear in the Simple Entry spreadsheet at what cycles the predictions are to be made. Please confirm that the predictions are to be made at the points indicated in the image, below. 

Answer: The interpretation is correct.

Question: I wonder if the measured dimensions of the specimens are available.
Answer: A file containing this information “Measured section dimensions for Specimens A-B-C.pdf” is now posted in the “02 Test Specimen Information” folder of the blind prediction contest directory

Question: Would you please let me know the geometry of the specimens used to obtain the material properties listed in Table 2 of the reference:  "2017a Ozkula et al. (Observations from Cyclic Tests on Deep Wide Flange Beam Columns)".
Answer: A file containing this information “tensile coupon dimensions for 2017a.pdf” is now posted in the “06 References” folder of the blind prediction contest directory

Question: Regarding the units on the material test coupon C -- want to confirm that the units should be as listed in the data [in/in] and not as plotted [%], i.e. failure strains are on the order of 40%/25%, and not less than 1%.
Answer: The test data for all three specimens are correct; however, the horizontal axis for Specimen C tensile coupon test results was not labelled correctly.  A corrected file, Engineering Stress-Stress Data_rev1.xls, is now posted under Material Test Results in the blind prediction contest directory

Question: Would it be possible to provide the chemical decomposition of the specimens’ steel materials?
Answer: The chemical properties of Specimens A, B, and C are provided in the table shown below and also uploaded as a pdf under the Test Specimen Information folder in the blind prediction contest directory.

Mill Certificate: Chemical Composition

Specimen
Designation
Column
Size
Chemical Composition (%)
C Si Mn P S Cu Ni Mo Cr Al V Nb Sn CE
A W30×148 0.17 0.15 0.104 0.013 0.011 0.23 0.08 0.02 0.12 0.002 0.056 0.001 0.013 0.4
B W14×82 0.08 0.2 0.2 0.02 0.032 0.34 0.11 0.022 0.18 0.003 0.001 0.021 0.01 0.29
C W18×130 0.07 0.23 1.3 0.015 0.025 0.32 0.11 0.03 0.15 - 0.04 0.001 0.01 0.36


Question: The submittal date on the announcement email listed January 22, 2017. Has the deadline passed for submittals?

Answer: We apologize for the mistake. The submittal deadline is January 12, 2018. Good luck!

Print