Oscillating Steel Rod Bench Tests
Before the container left Australia I conducted a series of tests with oscillating lengths of 12mm, 16mm and 18mm mild steel rods at Newington Armoury, Sydney.
These tests were conducted between the 31st March - 11th April.
I built a small hydraulic test bench to mimic the horizontal motion of the shake table. Using the same hydraulics off the main table I was able to replicate the actual frequency and stroke the EQ machine being installed in Parkfield operates at.
See below video of Hydraulic Bench Test with various lengths (1.5m to 6.0m )and 12, 16 and 18mm steel rods.
http://youtube.com/watch?v=NRKn8d6DA4A
David Gessel ran an FEA Model of 16mm steel rod.
http://youtube.com/watch?v=-uAsI6DysXA
Bench Test Photographs
Decided not to pack any metric materials in the container to travel to Parkfield. Will use American Stock of 1/2 inch and 5/8 rods to 20ft in length. During the three months of installation in Parkfield various configurations will be installed on the shake table.
I will be based at The Shipyard, Berkeley from around the 12th May. I need to fabricate mount connections for the rods to be attached to the shake table. Ideally I would like to have up to 110 connection points. Realistically I will aim for 21 connection rods to be secured to the shake table for the 28th June when the intervention begins.
Attached is sketch of mount connection.
Videos
FEA Modelling by David Gessel
Posted: April 24th, 2008 under blog.allshookup.org.
Comments: 2
Comments
Comment from dvr
Time: April 24, 2008, 6:23 am
Correspondance Between D.V.Rogers and David Gessel towards FEA modelling and results of Bench Tests that took place
Posted by David Gessel Sat, Mar 29, 2008
Ultimately, I hate to say, that as cool as simulation and FEA is, this sort of application is the kind of thing that is very hard to predict. It’s not that it is particularly tricky in design, but rather that the damping factors and fatigue factors are not deterministic (as stiffness or yield tends to be).
For the latter one just wants to be at least 50% below yield and hope for the best. From the sound of it the tubes won’t be wiggling at resonance all the time, so that should be fine.
For the former… much depends on final construction and material condition and even atmospheric conditions and peak velocity and etc. Small variations would make a big difference in the peak sustained movement range at resonance and hence the peak stress at the root. I am doing a worst case calculation - sustained resonance and minimal damping, peak stress across the frequency sweep. Even so I’d want to do some sustained testing with worst-case synthetic motions before I’d let the boy scout troop stand around it (Oh no, it killed Kenny!).
I think for the long tubes either a solid root or reinforced wall root (heavy wall tube inserted instead of solid bar) and a rigid top plate makes for a strong, simple attachment that won’t result in any surprises and should be very easy to fabricate. One could start with the simplest mechanism and give it a test. If it doesn’t perform as expected, then move on to enhancements.
————————————
Posted By D.V.Rogers Mon, Apr 7, 2008
I ran out of time to run tests with hollow wall tube, I made an order
for steel and cancelled at the last minute. In a nutshell I realised I
have potentially a major engineering issue with attaching 12m (40ft)
lengths to the table in that it worries me a little that it might rip
the floor pieces off the alignment couplings attached to the
actuators.
————————————
Posted by David Gessel Wed, Apr 9, 2008
attached graph shows 12mm rod first and second modes varying as length
varies from 6000-1500mm
at 6M first 12 modes (x and y, so two of each, really just first 6 modes)
are summarized below.
12mm x 6M (hertz)
1 2.357561e-01
2 2.362168e-01
3 1.478321e+00
4 1.480304e+00
5 4.139427e+00
6 4.144540e+00
7 8.110290e+00
8 8.117729e+00
9 1.340968e+01
10 1.342741e+01
11 2.003592e+01
12 2.005786e+01
16mm x 6m (hertz)
Mode Frequency (Hz) Convergence
—- ————– ———–
1 3.131754e-01 0.1%
2 3.133190e-01 0.1%
3 1.962842e+00 0.1%
4 1.963462e+00 0.1%
5 5.495826e+00 0.1%
6 5.497376e+00 0.1%
7 1.076893e+01 0.1%
8 1.077171e+01 0.1%
9 1.779970e+01 0.0%
10 1.780542e+01 0.1%
11 2.658673e+01 0.1%
12 2.659356e+01 0.1%
18mm x 6m (hertz)
Number of Modes: 12
Mode Frequency (Hz) Convergence
—- ————– ———–
1 3.523737e-01 0.1%
2 3.525208e-01 0.1%
3 2.208470e+00 0.1%
4 2.209098e+00 0.1%
5 6.183519e+00 0.0%
6 6.185075e+00 0.1%
7 1.211623e+01 0.0%
8 1.211908e+01 0.1%
9 2.002627e+01 0.0%
10 2.003207e+01 0.0%
11 2.991178e+01 0.0%
12 2.991872e+01 0.0%
attachments:
modal_graph.jpg
16mm_modes.jpg
————————————-
Except Posted By D.V.Rogers Fri, Apr 11, 2008
… the16mm rod model at 6m you produced almost matches visually the actual
test that took place.. (though the length of 16mm rod on the .mp4 is
only 4m in length)
When viewing the .mpg4 file the tallest rod is 18mm at 6m and the next
size down is 16mm at 4m. The rest is a mix of 12mm rods. My guess is
the table is only running at about 2hz max, but no way to tell at this
point. At least the bench tests gave me some idea of what I can play
with steel rod up to 6m.
I did not pack any metric material so will have to source US stock and
ship it to Parkfield. Will use 1/2 inch and 5/8 mild steel rods to
various lengths up to 20ft (6m) attached to 1 inch 1020 tight fitting
machined collars at 4 inches by length welded to 5/8 plate (centre of
plate drilled out to 1 inch with collar dropped right through) mounted
with four 7/16 or 3/8 bolts to table floor pieces. I am looking at attaching no more than 50-60 rods at any one time. Various configurations will be tried out over the 3
month installation.
http://youtube.com/watch?v=NRKn8d6DA4A
—————————————
Posted by David Gessel Sat, Apr 12, 2008
some analysis including the collar
1/2 x 20′ Number of Modes: 6
Number of Modes: 8
Mode Frequency (Hz) Convergence
—- ————– ———–
1 2.475023e-01 0.4%
2 2.483675e-01 0.4%
3 1.550116e+00 0.3%
4 1.556321e+00 0.2%
5 4.338577e+00 0.3%
6 4.357011e+00 0.2%
7 8.498200e+00 0.3%
8 8.535927e+00 0.1%
the second analysis is the 1/2 rod, 20′ long, driven 2″ at the resonant frequencies found in the first analysis (I think… honestly I’m not entirely sure if that’s exactly what I’m calculating, but I think so). Peak stress is the top of the graph. the numerical results are below. 20′ rods driven at 1.55 hz 2″ horizontally will probably yield the rods. Might also happen with 20′ 5/8 rods (probably would). Interestingly, it seems like the first mode results in lower peak stress, possibly because the lower frequency results in lower velocity means lower peak acceleration. Higher modes have so much less displacement that peak stress is lower. 2nd mode seems to be the killer.
that said, it looks like the collars will be fine - it’ll be the rods that bend. So keep an eye on them.
attachments:
first_four_modes.jpg
driven_2_inches_at_resonance.jpg
Comment from dvr
Time: April 24, 2008, 6:23 am
Steel Rod Materials, Contacts and Costs
60 Lengths of 1/2 inch rods @ 20ft $721.00
60 Lengths of 5/8 inch rod @ 20ft $1127.00
+ $60 Transport from Paso Robles
P.R.W. Steel Supply, INC. Paso Robles
Rajon Begin
General Manager
Ph: (805) 239-1688 ext. 203
Fax: (805) 239-5753
e-mail: rajon@prwsteel.com
———————————–
1 length of 1 inch 1018 Bright Bar
1 length of 1 1/4 inch 1018 Bright Bar
2 lengths of 3′ x 5/8 Flat Bar Mild Steel
Total $290.00
Quote No: 239810
Ray Perdue RPerdue@pdmsteel.com
Stockton, California
(209) 943-0513
(209) 466-8420 (Fax)
—————————————-
Machining
smpmachine.com
Kevin Binkert
kevin@smpmachine.com
558 Bryant St.
San Francisco, CA 94107
Tel. 415-546-6784
Fax. 415-543-3472
———————————
440 3/8 by 3 1/2 inch High Tensile Bolts
(10mm x 90mm)
———————————
Write a comment