Farzad S. Dizaji

M.S. thesis · University of Virginia · 2019

High Performance Superelastic Materials for Seismic Applications

Buildings that survive an earthquake can still be lost if they are left leaning. My thesis asks whether an ultra-strong shape memory alloy, NiTiHfPd, can pull steel buildings back to plumb after shaking, and turns the answer into a new device: the SMA re-centering damper (SMARD).

  • Author Farzad Shafiei Dizaji
  • Advisor Prof. Osman E. Ozbulut
  • Committee Prof. Devin K. Harris, Prof. Jose Gomez
  • Degree M.S., May 2019
−55 to +40 °Ceight test temperatures on two NiTiHfPd specimens
4.31% → 0.69%median residual drift, 4-story frame, 2%-in-50-years shaking
4.21% → 1.93%median peak drift, 6-story frame at MCE, without vs with SMARDs
44far-field ground motions (FEMA P695) per analysis
Chapter 1

An alloy that remembers its shape

Superelastic shape memory alloys can be squeezed several percent and spring back with almost no permanent deformation. Their stress–strain curve is a flag: a plateau on loading, a lower plateau on unloading, and a return to zero.

Ordinary NiTi is limited in strength and changes behavior with temperature. NiTiHfPd alloys from NASA reach about 2000 MPa with large damping. With Prof. Haluk Karaca's team at the University of Kentucky, two specimens were tested in compression at three loading frequencies (0.05, 0.5 and 1 Hz) and four strain amplitudes, from −55 °C to +40 °C. I then calculated their energy dissipation and equivalent viscous damping and built an OpenSees model of their response.

  • Sample 2single crystal [111] Ni45.3Ti29.7Hf20Pd5, 7.71 mm tall, 3.84 × 3.85 mm; tested at −55, −35, −15, +5 and +25 °C
  • Sample 1polycrystal Ni45.7Ti29.3Hf20Pd5, 7.76 mm tall, 3.68 × 3.78 mm; tested at 0, +20 and +40 °C
  • Testingaged at 400 °C for 3 h; 100 kN MTS Landmark frame; 12 mm high-temperature extensometer; heating by band heaters, cooling by liquid nitrogen
  • Modeltwo self-centering materials in parallel in OpenSees, fitted to the 1 Hz tests (Table 4-1)

Quick check: when the alloy gets colder (25 °C → −35 °C), what happens to its energy dissipation at 1 Hz?

Load the specimen (model)Interactive

Loop: OpenSees model with the Table 4-1 parameters (fitted to 1 Hz tests at −35, +5 and +25 °C).

Measured damping, Tables 3-1 to 3-8Data

Chapter 2

SMA braces bring a 4-story frame back to plumb

A steel moment frame can survive a strong earthquake but stay permanently tilted, and a building left leaning more than about 0.5% is often demolished.

The case study is a 4-story steel office building in Los Angeles (soil type D) with special moment frames and reduced beam sections. I modeled the perimeter frame (two moment-frame bays and two gravity bays) in OpenSees. For the SMA-braced version, the frame was made lighter, to about 75% of the original base-shear capacity, and NiTiHfPd braces were added in a chevron layout in every bay of every story. Both frames were analyzed under 20 SAC ground motions for Los Angeles at three hazard levels: 50%, 10% and 2% probability of exceedance in 50 years (return periods of 72, 475 and 2475 years).

The braced frame matched the original in peak drift and floor acceleration but cut the drift left behind dramatically. Changing the brace temperature between −35 °C and +25 °C barely mattered: the largest change was the median residual drift at 2%-in-50-years shaking, from 0.69% at +25 °C to 0.95% at −35 °C.

Quick check: under the rarest shaking (2% in 50 years), what was the median residual drift of the frame without SMA braces?

Shake both framesInteractive
Median peak drift5.88 / 5.23%moment frame / SMA-braced
Median floor accel.14.03 / 13.14m/s², moment frame / SMA-braced
Median residual drift4.31 / 0.69%moment frame / SMA-braced

Medians of 20 records: moment frame from Table 4-3, SMA-braced frame at the chosen brace temperature from Table 4-5. The animation is illustrative: each frame's final lean is scaled from its median residual drift.

Chapter 3

The SMA re-centering damper (SMARD)

The SMARD combines two jobs in one compact device: NiTiHfPd bars that re-center the structure, and high-force friction springs that dissipate energy and give it a large stroke.

An SMA–spring assembly sits inside an enclosure with rigidly attached baffle plates, and a piston passes freely through holes in the baffles. Whether the piston is pulled out or pushed in, the assembly is compressed, which suits NiTiHfPd because it behaves best in compression. The bars can have up to four different diameters (SMA1 and SMA2 here): in a weak event the softer bars work, in a strong event both groups engage, and in an extreme event the bars stiffen again and limit displacement.

  • Design3600 kN force capacity, 150 mm stroke, loss factor over 0.30
  • SMA barstwo groups of four 25 mm bars, 200 mm long, design stress 1800 MPa; buckling load 1473 kN > 900 kN
  • Friction springsfour 900 kN springs in parallel, each 12 elements (type 24301), 400 mm long; about 2/3 of the energy is absorbed by friction between the rings
  • Buildings6-story (FEMA P-751) and 9-story (SAC) steel moment frames in Seattle, two SMARDs per story in the 2nd and 4th bays, chevron braces; periods 2.04 → 1.29 s and 2.37 → 2.19 s
  • AnalysisOpenSees, 22 far-field record pairs (44 records, FEMA P695) at the design (DBE) and maximum (MCE) earthquakes; collapse at 10% drift

Quick check: in the 6-story frame at MCE, what was the median peak drift with SMARDs?

Drive the pistonDrag or play
or drag the piston rod left and right
Results over 44 records

Conclusions

What the thesis shows

A strong, damping alloyNiTiHfPd kept its superelastic flag from −55 to +40 °C. Colder and slower loading means more damping; from +5 to +25 °C damping and energy change by less than 10%.
Braces that re-centerIn a 4-story frame, SMA braces cut the median residual drift at the rarest shaking from 4.31% to 0.69% (0.95% at −35 °C), with similar peak drifts and accelerations.
A practical deviceThe SMARD cut the 6-story median peak drift from 4.21% to 1.93% at MCE, nearly removed residual drift, and raised floor acceleration only moderately. Next steps: test a prototype and study temperature effects on the full device.

Read the full thesis   See the SMARD research on the main site