New DelCompliance

 

2.157

The old version, .88 Fgrad

New Version

ORS shortcut

jul10

p125mm

 

.125mm edge length: 1.4815GPa, .091337 GPa Std Dev, 499 runs.

Add another 100 runs

If we add these 100 runs to the p125mm dataset, the average goes:

1.4848 GPa

.0937 GPa StdDev

599 Runs.

1mm:

1.5Gpa

.095 GPA stdev...

Jul9

Trying 500 runs of p125 again:
2620005[].hhpc-hn1

July 8th

2619843.hhpc-hn1
p125mm/Atest.uda.103
B4C.xml
dk2:6e-1  Will try lowering this to e-12 next.
eta:25.4
poisson, 1-5^2.6

 

This didnt run the proper file!! EDIT BTM.xml, NOT B4C.xml!!!

2619845.hhpc-hn1

p125mm/Atest.uda.104

B4C.xml
dk2:6e-1  Will try lowering this to e-12 next if it doesnt do anything.
eta:25.4
poisson, 1-5^2.6

Increased strength, but thats about it.
Trying the dkappa2.
2619856.hhpc-hn1

p125mm/Atest.uda.105

B4C.xml
dk2:6e-12 
eta:25.4
poisson, 1-25^2.6

NIIICE
2619858[].hhpc-hn1
p125mm/Atest.uda.106-205
B4C.xml
dk2:6e-12 
eta:25.4
poisson, 1-25^2.6
1.502GPa Average
.1011GPa stdev...

 Lets do a 500 runs.  There is probably something wrong with how the random variable sets eta for each bin.

Start with .125mm
 500 runs:
p125mm/Atest.uda.000-499
B4C.xml
dk2:6e-12 
eta:25.4
poisson, 1-25^2.6
2619981[].hhpc-hn1

This didnt pan out, starting throwing errors at 30.

Recent Results

This is a plot of 30 runs for the 4mm cube, poisson dist from 1 to 25um. The variation occurs in the later stages once failure has already set in. 1.528 GPa

2619035[].hhpc-hn1
Bypass inclusion on
4mm/Atest.uda.0032-->62

...

2619260.hhpc-hn1
4mm/Atest.uda.063
4mm with 3 grid
Inclusion calculation off
B4C.xml
dk2:13e12
eta:6e-12
poisson, 1-25^2.6

2619695.hhpc-hn1
4mm/Atest.uda.065
4mm with 3 grid
B4C.xml
dk2:6e-5
eta:832000
poisson, 1-25^2.6
1.45GPa
takes about 15 minutes irl


 Turned out that $RANDOM doesn't work in tcsh, use bash instead.
cat >
cat >>

2619704.hhpc-hn1
4mm/Atest.uda.067
testing the built B4C.xml
1.48GPa!!

Testing the built addric.pbs
2619706.hhpc-hn1
4mm/Atest.uda.068
 1.48GPa- this one isnt right.

2619765.hhpc-hn1

Lets try a new setup that allows us to retain input decks so we know the random seed. We will use jobno as the .xml input into the .ups file.  Lets try it out.

2619765.hhpc-hn1- should make a new $PBS_JOBID.ups and .xml files and run off them...
4mm/Atest.uda.069
1.58GPa

2619766[].hhpc-hn1
4mm/Atest.uda.070-30?
eta=13e12*4e-3^3=832000
-1.5137e+09 GPa
stdev:1.1212e+08
30 runs
lets get more runs on this after we get data for .125mm



2619769[].hhpc-hn1
1mm/Atest.uda.000-99?
Running into a problem with simultaneous calls to sus I think.
Only got to 72 runs.

eta=(13e12*1e-3^3)=13000
-1.4877e+09
9.2777e+07


2619770[].hhpc-hn1
p125mm/Atest.uda.003-102?
did not work, trying again.


eta=13e12*.125e-3^3=25.39

2619843.hhpc-hn1
p125mm/Atest.uda.103
Dropped 25um max to 5um...is it possible the flaw size is too big for the cell size??

This is behavior we get at .p125 mm with 1um to 25um flaw size:

Why does stress critically increase??

Post June 17th Update

2617527.hhpc-hn1
HuRamCode2/Atest.uda.003
flawdensity: 13e12
velocity defined
 distribution: delta6um
interval:1e-8
dkappa2:6e-12
chi:0

As an aside: the diluteness can be characterized as eta*s^3 / volume. In a delta distribution of 6um, that comes out to (13e12* 6e-6^3)/(4e-3^3)=43875.  As discussed, we need to drop eta appropriately.

If eta is 13e12 #/m^3, then for a 4mm edge length, #=832000.

832000*6e-6^3/4e-3^3=.0028, which is properly dilute.  We will need to do this going forward.


poisson 1um to 25um exp2.6
 possible sus or pdTdt error.
2617783.hhpc-hn1

recopied sus, same error
fixed pdTdt, recompiling.
2617784.hhpc-hn1

 visit

2.7.3 doesnt work
2.6.0 seems to work
/data1/hhpc-hn8/Programs/visit/visit_install
hhpc-hn8.idies.jhu.edu

Looks like particle id 281483566776320 is the central particle.

Questions for Ravi- why do we need ghost cells?
Why does ghost cells generate 27 particles? I guess 26 are ghost particles....check this with ravi.

Looks like a non-uniaxial stress state. will modify to drop other stresses.

2618383.hhpc-hn1
def grad: .88 1 1

looks like it went the wrong direction. lets try 1.1
2618389.hhpc-hn1
def grad: .88 1.1 1.1
Atest.uda.009
around 100th timestep- looks like too much tension on the non central axis:

Atest.uda.010
def grad .88 1.05 1.05
2618401.hhpc-hn1
B4C.xml
dk2:6e-6

2618413.hhpc-hn1
 Atest.uda.012
 def grad .88 1.05 1.05
B4C.xml
dk2:6e-12

328th timestep, doesnt make curve better.  1.57 GPa. I thnk 1.04 is best.

Atest.uda.013
2618479.hhpc-hn1
def grad .88 1.04 1.04
B4C.xml
dk2:6e-12
eta:10e12

comes to 2.38e9 GPa

3e12 = .81 GPa... 2.38-3.75=1.37/.81--> subtract 5e12.

Atest.uda.014
def grad .88 1.04 1.04
B4C.xml
dk2:6e-12
eta:5e12
2618559.hhpc-hn1
2.42 GPa

Atest.uda.016
def grad .88 1.04 1.04
B4C.xml
dk2:6e-10
eta:5e10
2618563.hhpc-hn1

2.43 seems to be the saturated strength.

Lets try a run with the expected flaw density for a 4mm cube.
2618569.hhpc-hn1
Atest.uda.017
def grad .88 1.04 1.04
B4C.xml
dk2:6e-5
eta:832000

2.52e9, beautiful curve.

 

Atest.uda.018

def grad .88 1.04 1.04
B4C.xml
dk2:6e-5
eta:832000
poisson, 1-25^2.6
2618581.hhpc-hn1

1.85 GPa!!!

Go back to 6um delta.  try different volume sizes.

Atest.uda.019

def grad .88 1.04 1.04
.4mm edge length
B4C.xml
dk2:6e-5
eta:832000
delta,6um
2618647.hhpc-hn1
2.52...... wtf

2618652.hhpc-hn1
Atest.uda.020
def grad .88 1.04 1.04
.4mm edge length
B4C.xml
dk2:6e-2
eta:832
delta,6um

3.54


2618678.hhpc-hn1
2618679.hhpc-hn1
Atest.uda.021
def grad .88 1.04 1.04
1mm edge length
B4C.xml
dk2:6e-3
eta:832
delta,6um

3.54GPa again....makes sense, didnt adapt eta.

 2618681.hhpc-hn1
Atest.uda.022
def grad .88 1.04 1.04
1mm edge length
B4C.xml
dk2:6e-3
eta:13000
delta,6um


Array with poisson:
2618682[].hhpc-h - didnt run far enough, i think im under an order of magnitude
2618687[].hhpc-h  -1.52, -1.52, -1.52....

2618695.hhpc-hn1
Atest.uda.000
1mm, new grid.
B4C.xml
dk2:6e-3
eta:13000
poisson, 1-25^2.6

1.53


2618696.hhpc-hn1
4mm/Atest.uda.000
4mm, new grid.
B4C.xml
dk2:6e-5
eta:832000
poisson, 1-25^2.6
1.53

2618697.hhpc-hn1
p125mm/Atest.uda.000
.125mm with 3 grid
B4C.xml
dk2: 25.39
eta:6e0
poisson, 1-25^2.6

1.53??
curve wasn't as pretty, no dropoff.

2618698.hhpc-hn1
p125mm/Atest.uda.001
.125mm with 3 grid
B4C.xml
dk2: 25.39
eta:6e-1
poisson, 1-25^2.6

Using the Matlab version of Ravi's Code

F-1)/(tmax-tmin) = Sr

change time to be appropriate with dt and Sr
1e3 rate
tmax=100e-6
dt=1e-7
time=4e-6us

1e-3 rate
tmax=100
dt=1e-2
time=120

Paliwal/Ramesh 2008

The purpose of using this deprecated model, that only incorporates material parameters and a size distribution, is to see if this simpler model can accurately capture rate dependent strength when using model inputs from characterization experiments.

The most critical governing equations in this model are as follows:


 

Shortcuts nix

Alegra: /usr/cta/restricted/MHD

From JHU:
Have to use ossh
Have to use wired ethernet for ./pkinit dmallic1@HPCMP.HPC.MIL


SSH:
ssh -X -Y dmallic1@hhpc-hn8.idies.jhu.edu
ssh -X -Y dmallic1@hhpc-hn1.pha.jhu.edu
Moving files
scp works great for hn8

Emacs:
ctrl x ctrl s - save
ctrl x ctrl c- quit
ctrl v- page down

visit
/data1/hhpc-hn8/Programs/visit 2.6.0

find . -name "foo*"

Aliases/Path:
emacs -nw ~/.bash_profile

alias xyz='cd /xyz'
PATH=$xyz/

source ~/.bash_profile

Uintah-
To install:
config log
make cleanreally
 make -j 6
After making changes:
/data1/hhpc-hn8/dmallic1/JHUUintah_SVN_ravi/src/CCA/Components/MPM/ConstitutiveModel
move HuRam.cc and .h, then make cleanreally then make -j 6IN THE UINTAH SVN DIR

and constitutive model.xml into standalone/inputs/UPS_SPEC
 HuRamesh.cc
remove pdTdt "_preReloc"


Test

4mm.uda.010-My input deck with Ravi's dimensions gives same answer as Ravi's input deck with my dimensions. This indicates that its not the input deck thats causing the problem.

Ravi suggests increasing flaw density and checking dkappa2 parameter on Jun3

2613095.hhpc-hn1 4mm.uda.008?

2613096.hhpc-hn1? 4mm.uda.009

 

This run isn't complete yet but i suspect the same outcome.
2613164.hhpc-hn1  4mm.uda.010? i think this was dkappa e-6

New calculation:
for 4mm^3 volume, at 22e12 #/m^3, 4mm^3 = 4e-9 m^3, so #=88000  This makes eta way too low to make sense.

2613218.hhpc-hn1
4mm.uda.011
flawdensity:13e14
distribution:delta6um
interval:1e-8
dkappa2:6e-14
chi:2

What is that sub peak? I thought moving dkappa2 would fix it?
Max peak occurs at about 40um.... this is 15 um past what experiments say it should be. Perhaps increase flaw density?
3.75e9 strength!

2613226.hhpc-hn1
4mm.uda.012
flawdensity:13e14
distribution:delta6um
interval:1e-8
dkappa2:6e-6
chi:2

Increasing dkappa clearly makes the curve cleaner.
3.83GPa
SR: 1242


2613228.hhpc-hn1
4mmHigherEta/4mm.uda.000
flawdensity:13e16
distribution:delta6um
interval:1e-8
dkappa2:6e-16
chi:2

Increasing the flaw density to 13e16 puts us far past a normal strength for the material. It looks like the inflection point at about 40us was missed, so the actual time to failure may not have changed.

2613296.hhpc-hn1

4mmHigherEta/4mm.uda.001
flawdensity:13e15
distribution:delta6um
interval:1e-8
dkappa2:6e-15
chi:2

2613297.hhpc-hn1
HuRamCode2/p125mm.uda.001
flawdensity:13e14
distribution:delta6um
interval:1e-8
dkappa2:6e-6
chi:2

5.5Gpa at smallest size (.125mm^3) at 1101 1e3 SR

2613314.hhpc-hn1
HuRamCode2/p125mm.uda.002
flawdensity:13e14
velocity .18
distribution:delta6um
interval:1e-8
dkappa2:6e-6
chi:2
check this at correct SR- came out to 1029 from 0 to 4300 step.

3.51GPa

Try .19 next

2613324.hhpc-hn1
HuRamCode2/p125mm.uda.003
flawdensity:13e14
velocity .19
distribution:delta6um
interval:1e-8
dkappa2:6e-6
chi:2
CANCELLED THIS, ABOVE SR IS FINE AT 1000



2613316.hhpc-hn1
HuRamCode2/p25mm.uda.001
flawdensity:13e14
velocity .4
distribution:delta6um
interval:1e-8
dkappa2:6e-6
chi:2

SR Came to 1138
3.69GPa

2613334.hhpc-hn1
HuRamCode2/p25mm.uda.002
flawdensity:13e14
velocity .35
distribution:delta6um
interval:1e-8
dkappa2:6e-6
chi:2

STR; 3.69GPa
SR 994.62

2613317.hhpc-hn1
HuRamCode2/p5mm.uda.001
flawdensity:13e14
velocity .8
distribution:delta6um
interval:1e-8
dkappa2:6e-6
chi:2

SR:1141-drop to .7
3.51 GPa

2613332.hhpc-hn1
HuRamCode2/p5mm.uda.002
flawdensity:13e14
velocity .7
distribution:delta6um
interval:1e-8
dkappa2:6e-6
chi:2

This result is not converging!!!!

I looked at dStress:

Looks like the two peaks are there but simply hidden.  They occur at ~6050 and ~6700.

SR is 964 through 6000
4.4GPa

2613318.hhpc-hn1
HuRamCode2/1mm.uda.001
flawdensity:13e14
velocity 1.6
distribution:delta6um
interval:1e-8
dkappa2:6e-6
chi:2

SR:1141- lets drop to 1.4
3.39GPa
Probably need to drop dkappa2 or change chi

2613330.hhpc-hn1
HuRamCode2/1mm.uda.002
flawdensity:13e14
velocity 1.4
distribution:delta6um
interval:1e-8
dkappa2:6e-6
chi:2

4.54GPa
984 SR

2613319.hhpc-hn1
HuRamCode2/2mm.uda.001
flawdensity:13e14
velocity 3
distribution:delta6um
interval:1e-8
dkappa2:6e-6
chi:2

SR:1073- a little high.  Should decrease velocity to 2.85
3.54 GPa?? Higher SR explains this

2613325.hhpc-hn1
HuRamCode2/2mm.uda.002
flawdensity:13e14
velocity 2.825
distribution:delta6um
interval:1e-8
dkappa2:6e-6
chi:2
SR: 1005
3.68GPa

2613321.hhpc-hn1
4mmHigherEta/4mm.uda.002
flawdensity:13e14
velocity 7
distribution:poisson 4um to 8um exp3
interval:1e-8
dkappa2:6e-6
chi:2

what the hell is going on with the poisson distribution??? gonna zero out variability and hit it with 6um to 6um.

2613350.hhpc-hn1
4mmHigherEta/4mm.uda.003
flawdensity:13e14
velocity 5.6
distribution:poisson 6um to 7um exp3
interval:1e-8
dkappa2:6e-6
chi:2


2613327.hhpc-hn1
HuRamCode2/4mm.uda.013
flawdensity:13e14
velocity 5.75
distribution:delta6um
interval:1e-8
dkappa2:6e-6
chi:2
3.61GPa
SR: 1025

Final Results for delta 6 distribution
4mm:3.61
2mm:3.68
1mm:4.54
.5mm: 4.4
.25mm: 3.69
.125mm: 3.51

2613498.hhpc-hn1
4mmHigherEta/4mm.uda.004
flawdensity:13e14
velocity 5.6
distribution:poisson 6um to 7um exp3
interval:1e-8
dkappa2:6e-14
chi:2

2613499.hhpc-hn1
HuRamCode2/4mm.uda.014
flawdensity:13e13
velocity 5.6
distribution:poisson 6um to 7um exp3
interval:1e-8
dkappa2:6e-6
chi:2

2613504.hhpc-hn1
4mmHigherEta/4mm.uda.005
flawdensity:13e12
velocity 5.6
distribution:poisson 6um to 7um exp3
interval:1e-8
dkappa2:6e-6
chi:2

We should be able to follow one of ravi's poisson runs to get results. Lets do that.
The run with zeros changed nothing.

2613912.hhpc-hn1
HuRamCode2/4mm.uda.015
flawdensity:13e12
velocity 5.75
distribution:poisson 1um to 7um exp3
interval:1e-8
dkappa2:6e-12
chi:2
Bottom of first peak: 1.2e-1

2613957.hhpc-hn1
HuRamCode2/4mm.uda.016
flawdensity:13e12
velocity 5.75
distribution:poisson 1um to 7um exp3
interval:1e-8
dkappa2:6e-12
chi:0
Looks like it went earlier.... not what u want

2613958.hhpc-hn1

4mmHigherEta/4mm.uda.006
flawdensity:13e11
velocity 5.75
distribution:poisson 1um to 7um exp3
interval:1e-8
dkappa2:6e-11
chi:2
looks a little higher....

2613960.hhpc-hn1
HuRamCode2/4mm.uda.017
flawdensity:13e10
velocity 5.75
distribution:poisson 1um to 7um exp3
interval:1e-8
dkappa2:6e-10
chi:0
e4.

2614091.hhpc-hn1
 HuRamCode2/4mm.uda.018
flawdensity:13e9
velocity 5.75
distribution:poisson 1um to 7um exp3
interval:1e-8
dkappa2:6e-9
chi:0

Trying addric.ups
2614756.hhpc-hn1
flawdensity:22e12
velocity time def grad
distribution:poisson 1um to 25um exp2.6
interval:1e-8
dkappa2:6e-12
chi:0
will need a try with chi if this doesnt work.
2614842.hhpc-hn1 again
HuRamCode2/Atest.uda.001
2615163.hhpc-hn1 again
HuRamCode2/Atest.uda.001
Is it possible that the boundary conditions are causing non uniaxial stress??
Make the box on both sides.


2615370.hhpc-hn1
HuRamCode2/4mm.uda.019 possibly 20 its 19
flawdensity:22e12
velocity 5.75
distribution:poisson 1um to 25um exp2.6
interval:1e-8
dkappa2:6e-12
chi:0

2615371.hhpc-hn1
HuRamCode2/Atest.uda.002
flawdensity:22e12
velocity defined
distribution:poisson 1um to 25um exp2.6
interval:1e-8
dkappa2:6e-12
chi:0

=.5*(Y2^2+AB2^2+AE2^2-1)

Code updated on June 17th