OnScale Cloud Failures Answered

5 comments

• Official comment

  

  Chloe Allison

  August 27, 2020 12:48

  Hi Tiffany,

  Your model is reporting 'Data manager contains bad values' which means it is going unstable. One trick to try when this happens is to lower the time stability factor of your model to 0.8 (it is currently at 0.95).

  

  In this case it stabilises the model.

  Best Regards,

  Chloe

  Comment actions Permalink
• 

  Tiffany

  August 31, 2020 07:04

  Hello Chloe, I tried lowering the stability factor, increasing my meshing, and replacing the absr to impd. But it still fails especially at higher frequencies. The error report is saying there is a memory issue. Please let me know what I can do to solve this problem. Additionally, I am not sure why I am getting a reflection from the top boundary. I included my updated code below. Please note I changed the boundary conditions back once impd did not work. I am attempting the sweep from 500kHz to 2 MHz (32 simulations).

   

  c mem 800 200 /* Allocate 800 megawords of memory - 3 GigaBytes (Not necesarry for Windows Operating Systems)
  c NOTE: MEM Command must be first command in file, if used. (Line 1)
  rest no

  c *************************************************************************************************************
  c
  c Generated Flex Input File
  c
  c *************************************************************************************************************
  c
  c DESIGNER :SonicMEMS CORNELL
  c MODEL DESCRIPTION :
  c DATE CREATED :28 Aug 2020
  c VERSION :2.0
  c *************************************************************************************************************

  mp
  omp * * /* Number of CPUs to be used in the execution.
  end
  
  titl Focusing_Neuralproject

  c *************************************************************************************************************
  c
  c Define User Variables
  c
  c *************************************************************************************************************
  c
  c These variables have been set by the user through the interface.
  c
  c *************************************************************************************************************

  symb coordFactor = 1.0 /* Coordinate conversion factor
  symb timeFactor = 1.0 /* Time conversion factor
  symb dMassFactor = 1.0 /* Mass conversion factor
  symbx freqint = 1e+6 /* Determined Drive Frequency (Hz)
  c symb freqint = 500e+3 /* Determined Drive Frequency (Hz)
  c *************************************************************************************************************
  c
  c Define Meshing
  c
  c *************************************************************************************************************
  c
  c Set the variable for the approximate element size for the model. Must be
  c sufficient to represent the wavelengths of interest. Recommended that at least
  c 15 elements per wavelength are used.

  c
  c *************************************************************************************************************

  symb freqdamp = $freqint
  symb wavevel = 1496 /* Longitudinal wave velocity in material (m/s)
  symb wavelgth = $wavevel / $freqint /* Wavelength of sound in material (m)
  symb nmesh = 40 /* Elements per wavelength
  symb box = $wavelgth / $nmesh

  c *************************************************************************************************************
  c
  c Geometry Locations (XYZ)
  c
  c *************************************************************************************************************

  
  c Scale Parameters
  symb xmin = 0.0 * $coordFactor
  symb x_particle = 0.0007 * $coordFactor
  symb x_pzt = 0.01 * $coordFactor
  symb xmax = 0.034 * $coordFactor

  symb ymin = 0.0 * $coordFactor
  symb y_pzt_b = 0.004 * $coordFactor
  symb y_pzt_t = 0.006 * $coordFactor
  symb y_particle_b = 0.016 * $coordFactor
  symb y_particle_t = 0.0165 * $coordFactor
  c symb ymax = 0.046 * $coordFactor
  symb ymax = 0.05 * $coordFactor

  
  c Determine lengths of the model
  symb xlen = ( $xmax - $xmin )
  symb ylen = ( $ymax - $ymin )

  c ***************************************************
  c
  c Keypoints in the X-Direction
  c
  c ***************************************************

  symb #keycord x 1 $xmin $x_particle $x_pzt $xlen
  symb #get { idx } rootmax x

  c ***************************************************
  c
  c Keypoints in the Y-Direction
  c
  c ***************************************************

  symb #keycord y 1 $ymin $y_pzt_b $y_pzt_t $y_particle_b $y_particle_t $ylen
  symb #get { jdx } rootmax y

  c *************************************************************************************************************
  c
  c Indices Locations (IJK)
  c
  c *************************************************************************************************************

  c Grid in I direction, using approximately element size of 'box' and at least 1 element
  symb #keyindx i 1 $idx 1 $box 1
  symb indgrd = $i$idx

  c Grid in J direction, using approximately element size of 'box' and at least 1 element
  symb #keyindx j 1 $jdx 1 $box 1
  symb jndgrd = $j$jdx

  
  c *************************************************************************************************************
  c
  c GCON Grid & Geometry Allocation
  c
  c *************************************************************************************************************

  grid $i$idx $j$jdx axiy
  geom
  keypnt $idx $jdx
  end

  c *************************************************************************************************************
  c
  c Driving Conditions
  c
  c *************************************************************************************************************
  c
  c A number of predefined waveform functions can be accessed in OnScale. The DATA
  c HIST option is used below, other examples include wavelets, gaussians and step
  c functions. The manual details the function entries

  c
  c *************************************************************************************************************

  symb freqtimefunc_1 = $freqint
  symb timeperiod = 1 / $freqint
  symb numbercyc = 2 * ( ( $ylen - $y_pzt_t ) / $wavelgth ) /* change the number to multiply the transiant time

  c func
  c name timefunc_1
  c sine $freqtimefunc_1 1.0 0. $numbercyc 0. 0.0 /*frequency amplitude phaseshift nperiod valuadd rampcycle tdelay
  c end
  func sine $freqint 50.0 0. $numbercyc 0. 0.0 /*drfine (freq amp phaseshift #cycles)

  c --------------------------------------------------------------
  c Project Material List
  c --------------------------------------------------------------

  
  c -------------------------------------------------------------------------
  c Global variables used in all the material definitions
  c -------------------------------------------------------------------------
  c
  symb epvacm = 8.854e-12 /* dielectric constant for vacumn
  symb freqdamp = 1.e6 if noexist /* specified frequency for damping model
  symb rmu0 = 1.2566e-6

  symb #msg 5
  ********************************************************
  Damping matched at $freqdamp Hz
  Redefine variable 'freqdamp' if device centre frequency
  varies significantly from this value
  ********************************************************

  
  c
  c -------------------------------------------------------------------------
  c Now define the axis transformation - only posx used in this file
  c -------------------------------------------------------------------------

  axis
  form vctr
  defn posx car2 0. 0. 0. 1. 0. 0. 0. 1. 0.
  defn negx car2 0. 0. 0. -1. 0. 0. 0. 1. 0.
  defn posy car2 0. 0. 0. 0. 1. 0. 0. 0. 1.
  defn negy car2 0. 0. 0. 0. -1. 0. 0. 0. 1.
  defn posz car2 0. 0. 0. 0. 0. 1. 1. 0. 0.
  defn negz car2 0. 0. 0. 0. 0. -1. 1. 0. 0.
  end

  c -------------------------------------------------------------------------
  c Input material properties to program
  c -------------------------------------------------------------------------

  matr
  c --------------------------------------------------------------
  c type : BIOLOGICAL :
  c name : watr :
  c desc : Water at 25C :
  c --------------------------------------------------------------

  
  wvsp on
  type elas
  prop watr 1054.00 1540.00 0.00000 0.010000
  vdmp watr $freqdamp db 0.500000 0.00000 1e+6 1.00000 0.010000 1.00000
  symb eps = 1
  symb aeps = $epvacm * $eps
  elec watr $aeps

  
  
  c --------------------------------------------------------------
  c type : GAS :
  c name : air :
  c desc : Air, room temp :
  c --------------------------------------------------------------

  
  wvsp on
  type elas
  prop air 1.24000 343.000 0.00000 0.
  symb eps = 1
  symb aeps = $epvacm * $eps
  elec air $aeps

  c --------------------------------------------------------------
  c type : PIEZO :
  c name : pmt3 :
  c desc : CTS 3203HD :
  c --------------------------------------------------------------

  
  symb rho = 7820.00 /* density
  symb qdmp = 90.0000 /* Mechanical Q at 1e6
  symb qsdmp = 90.0000 /* Mechanical Q at 1e6
  symb freqdamp = 1.e6 if noexist
  symb freqloss = 1.00000e+06 if noexist /* Frequency that loss was measured at

  wvsp off
  type lean

  c define baseline stiffness coefficients (Constant electric field)

  symb c11 = 1.3745e+11
  symb c12 = 8.79e+10
  symb c13 = 9.23e+10
  symb c14 = 0
  symb c15 = 0
  symb c16 = 0
  symb c22 = 1.3745e+11
  symb c23 = 9.23e+10
  symb c24 = 0
  symb c25 = 0
  symb c26 = 0
  symb c33 = 1.2574e+11
  symb c34 = 0
  symb c35 = 0
  symb c36 = 0
  symb c44 = 2.2279e+10
  symb c45 = 0
  symb c46 = 0
  symb c55 = 2.2279e+10
  symb c56 = 0
  symb c66 = 2.4779e+10

  prop pmt3 $rho
  $c11 $c12 $c13 $c14 $c15 $c16 $c22
  $c23 $c24 $c25 $c26 $c33 $c34 $c35
  $c36 $c44 $c45 $c46 $c55 $c56 $c66

  c define baseline dielectric coefficients

  symb epxx = 1305.8
  symb epyy = 1305.8
  symb epzz = 1200.2

  c scale material properties as specified above
  symb aepxx = $epvacm * $epxx
  symb aepyy = $epvacm * $epyy
  symb aepzz = $epvacm * $epzz

  elec pmt3 $aepxx $aepyy $aepzz

  c define baseline piezoelectric coupling coefficients

  symb ex1 = 0 /* coupling constant
  symb ex2 = 0 /* coupling constant
  symb ex3 = 0 /* coupling constant
  symb ex4 = 0 /* coupling constant
  symb ex5 = 16.054 /* coupling constant
  symb ex6 = 0 /* coupling constant
  symb ey1 = 0 /* coupling constant
  symb ey2 = 0 /* coupling constant
  symb ey3 = 0 /* coupling constant
  symb ey4 = 16.054 /* coupling constant
  symb ey5 = 0 /* coupling constant
  symb ey6 = 0 /* coupling constant
  symb ez1 = -9.44 /* coupling constant
  symb ez2 = -9.44 /* coupling constant
  symb ez3 = 22.495 /* coupling constant
  symb ez4 = 0 /* coupling constant
  symb ez5 = 0 /* coupling constant
  symb ez6 = 0 /* coupling constant

  piez pmt3 1 1 $ex1 1 2 $ex2 1 3 $ex3 1 4 $ex4 1 5 $ex5 1 6 $ex6 &
  2 1 $ey1 2 2 $ey2 2 3 $ey3 2 4 $ey4 2 5 $ey5 2 6 $ey6 &
  3 1 $ez1 3 2 $ez2 3 3 $ez3 3 4 $ez4 3 5 $ez5 3 6 $ez6

  rdmp pmt3 $freqdamp q $qdmp $qsdmp 1e+6 1.0

  axis pmt3 posy /* relate materials local system to global system

  
  c --------------------------------------------------------------
  c type : MISC :
  c name : si :
  c desc : Silicon, generic :
  c --------------------------------------------------------------

  
  wvsp on
  type elas
  prop si 2330 7526 4346 0.01
  vdmp si $freqdamp db 0.1 0.3 1e+6 1 0.01 1
  thrm si 702 124 124 124 1.0 0 0 27
  symb eps = 11.9
  symb aeps = $epvacm * $eps
  elec si $aeps
  
  elec void $epvacm
  
  
  c --------------------------------------------------------------
  c type : EPOXY :
  c name : arald1 :
  c desc : Araldite MY750/HY956EN :
  c --------------------------------------------------------------

  
  wvsp on
  type elas
  prop arald1 1146 2658 1237 0.01
  vdmp arald1 $freqdamp db 4 12.59 1e+6 1 0.01 1
  symb eps = 3.5
  symb aeps = $epvacm * $eps
  elec arald1 $aeps

  elec void $epvacm
  
  end

   

   

   

  c *************************************************************************************************************
  c
  c Primitive Definitions
  c
  c *************************************************************************************************************

  site
  regn void
  blok watr $xmin $xmax $y_pzt_t $ymax
  
  c blok si 0.0 0.001 0.026 0.027 0.0 0.0 /* rectangle chip location
  c blok watr 0.0 0.0005 0.026 0.027 0.0 0.0 /* square chip location
  c blok watr $xmin $x_particle $y_particle_b $y_particle_t /* water rectangle chip location (X: 0-0.00052, Y: 0.016-0.0165)
  
  c ***Chip or Bead Properties***
  c blok si $xmin $x_particle $y_particle_b $y_particle_t /* Water rectangle chip location (X: 0-0.00052, Y: 0.016-0.0165)
  c blok si $xmin $x_particle $y_particle_b $y_particle_t /* Si rectangle chip location (X: 0-0.00052, Y: 0.016-0.0165)
  c blok arald1 $xmin $x_particle $y_particle_b $y_particle_t /* Epoxy rectangle chip location (X: 0-0.00052, Y: 0.016-0.0165)
  circle2d arald1 $xmin $y_particle_b $x_particle /* circle bead defined as arald1 to define an interface to apply a load to

  c *****************************

  blok air $xmin $xmax $ymin $y_pzt_t
  c blok pmt3 $xmin $x_pzt $y_pzt_b $y_pzt_t /*PZT block
  end
  
  
  c Plot Modl
  grph
  nvew 2 1 /* Set up 2 views
  plot matr /*i $i1 $i4 j $j1 $j6 /* Plot Model
  mirr x on /* Apply visual symmetry
  plot matr /* Replot Model
  end
  term

  c *************************************************************************************************************
  c
  c Boundary Definitions
  c
  c *************************************************************************************************************

  boun
  side xmin symm
  side xmax absr
  side ymin absr
  side ymax impd
  end

  c *************************************************************************************************************
  c
  c Calculated Properties
  c
  c *************************************************************************************************************
  c
  c By default, Flex only calculates the minimum required data set, typically this
  c means only velocities. This is done for memory efficiency. Should other
  c properites be required (e.g. displacements, stresses, strains, pressure), then
  c these must be requested by the CALC command. The manual lists all these options

  c
  c *************************************************************************************************************

  calc
  pres acoustic /* calculate acoustic pressure
  max aprs none apmx /* calculate min/max acoustic pressure
  c disp y /* calculate displacements
  end

  c *************************************************************************************************************
  c
  c Piezoelectric Load Definitions
  c
  c *************************************************************************************************************

  c piez
  c wndo auto piez
  c defn load_1 6.28
  c symb bxmin = 0.0 * $coordFactor - $box
  c symb bymin = 0.006 * $coordFactor - $box
  c symb #get { is js * } clsnode $bxmin $bymin *
  c symb bxmax = 0.01 * $coordFactor + $box
  c symb bymax = 0.006 * $coordFactor + $box
  c symb #get { ie je * } clsnode $bxmax $bymax *
  c nod2 pmt3 watr $is $ie $js $je * *
  c bc load_1 volt functimefunc_1 10.0 0.0
  c
  c defn load_2 6.28
  c nod2 pmt3 air
  c bc load_2 grnd
  c slvr drct
  c end

   

  c *************************************************************************************************************
  c
  c Pressure Load Definitions
  c
  c *************************************************************************************************************

  plod
  pdef pld1 func /*define a load
  c matr vctr1 air out
  vctr vct1 0. 1. 0. /*define the vector
  c sdef load_1 vector_1 $i1 $i2 $j2 $j2 /*apply a load in the vector direction
  sdef pld1 vct1 $i1 $i3 $j3 $j3 /*apply a load in the vector direction

  end

  
  time * * 0.8 /*time stability factor

  c *************************************************************************************************************
  c
  c Process Model
  c
  c *************************************************************************************************************
  c
  c Issue process (PRCS) command. Checks model integrity, and calculates stable
  c time step. NOTE: Process command must always be issued

  c
  c *************************************************************************************************************

  c symb #msg c Checking Model Integrity......
  prcs

  grph
  colr user size 5 /* how many materials are defined in this model
  colr user 1 0.0 0.0 1.0
  colr user 2 0.0 1.0 1.0
  colr user 3 0.0 1.0 0.0
  colr user 4 1.0 0.0 0.0 /* Si chip (red)
  colr user 5 1.0 1.0 0.0 /* arald1 chip (yellow)

  colr tabl matr 5
  map watr 1
  map air 2
  map pmt3 3
  map si 4
  map arald1 5

  end
  
  c grph
  c line off
  c nvew 2 1
  c arrow pole
  c plot piez
  c plot matr piez
  c arrow off
  c end
  
  term

  c *************************************************************************************************************
  c
  c Choose Time Histories To Store
  c
  c *************************************************************************************************************
  c
  c Save field (such as displacement or pressure) from a node or element for all
  c time steps with POUT command. Histories are referenced by order of
  c specification. Histories will be saved in the Flex History file (flxhst).

  c
  c *************************************************************************************************************

  pout
  rate 1
  symb xloc = 0.0 * $coordFactor
  symb yloc = $y_pzt_t * $coordFactor
  hist xyz aprs $xloc * * $yloc * *
  c hist func_1
  c histname electrode vq all
  end

  c *************************************************************************************************************
  c
  c Run the Model
  c
  c *************************************************************************************************************
  c
  c Specify the number of time steps to be run. Can be set to auto by using
  c 'Ringdown'.

  c
  c *************************************************************************************************************

  c User defined runtime for the model
  symb #get { step } timestep
  c symb ttime = 4e-5
  c symb ttime = 8e-5
  c symb ttime = 12e-5
  c symb ttime = 1e-5
  symb ttime = 1 * $numbercyc * $timeperiod

  symb nexec = $ttime / $step
  symb nloops = 30
  symb nexec2 = $nexec / $nloops

  
  c Set up Plotting
  grph
  nvew 2 1
  set imag tiff
  end

  c Set up Snapshot file

  
  c Create run plot procedure
  proc plot save

  c Run some timesteps
  exec $nexec2

  c Plot model
  grph
  plot aprs
  plot apmx
  c plot ydsp
  c plot yvel
  imag
  end
  
  c data
  c out yvel
  c out ydsp
  c out xvel
  c out aprs
  c end

  
  data
  file out 'water_output_data.flxdato'
  out modl
  out aprs
  out apmx
  c out yvmx
  out yvel
  c out xvel
  end
  c
  c data
  c file out 'snapshot.mat'
  c form out matlab output_data
  c out yvel
  c c out ydsp
  c out xvel
  c c out aprs
  c end
  c
  end$ proc

  c Run model then wait
  proc plot $nloops

  term

  
  c *************************************************************************************************************
  c
  c Save symbol variables to file for later use
  c
  c *************************************************************************************************************

  symb #get { labl } jobname
  symb #save '$labl.symb'
  stop /* return to command prompt

  Comment actions Permalink
• 

  Chloe Allison

  September 01, 2020 10:32

  Hi Tiffany,

   

  If there is a memory issue, here is a blog that will hopefully help you overcome this.

   

  Also if you are sweeping from 500 kHz to 2 MHz ensure that you sweep from the largest simulation to the smallest so enough memory is assigned (i.e. use 2 MHz as your start value and 500 kHz as your smallest). This should stop the failures.

   

  Best Regards,

  Chloe

  Comment actions Permalink
• 

  Tiffany

  September 01, 2020 10:41

  I will read the article you sent. However, I switched the starting value and I still get crashing. Additionally, it crashes sometimes when I run one at a time. 

  Comment actions Permalink
• 

  Chloe Allison

  September 01, 2020 10:43

  Tiffany,

   

  In this case you will likely need to override the estimate as mentioned in the blog post.

   

  Best Regards,

  Chloe

  Comment actions Permalink

Please sign in to leave a comment.