Issues with my input file

[Kenneth Obrien M]

I have the following issues with my input file:

• The code takes to long to simulate how can I speed simulation.
• The stone is buried 0.0762 in the soil and the code does not generate the geometry view.
• Computing the number of scans to get a B-scan file, its not clear. I did it by taking the size of the x-domain dividing by spatial step but it has not worked, I get errors.
• how can one create a B-scan.

below is my input file:

#title: B-scan from a stone buried in a heterogenous soil
#domain: 0.5 0.40 0.2
#dx_dy_dz: 0.001 0.001 0.001
#time_window: 6e-9

#material: 3.5 0 1 0 my_stone

#waveform: ricker 1 1.5e9 my_ricker
#hertzian_dipole: z 0.040 0.260 0 my_ricker
#rx: 0.080 0.260 0
#src_steps: 0.001 0 0
#rx_steps: 0.001 0 0

##sphere: 0.25 0.3238 0.0 0.01 my_stone
#soil_peplinski: 0.5 0.5 2.0 2.66 0.001 0.1 my_soil
#fractal_box: 0 0 0 0.5 0.40 0.2 1.5 1 1 1 20 my_soil my_soil_box

#geometry_view: 0 0 0 0.5 0.40 0.2 0.001 0.001 0.001 soil_stone n

Warning Errors I'm getting:

WARNING: '#hertzian_dipole: z 0.040 0.260 0 my_ricker' sources and receivers should not normally be positioned within the PML.

WARNING: '#rx: 0.080 0.260 0' sources and receivers should not normally be positioned within the PML.

[Craig Warren]

Hi Kenneth,

Firstly, the warning messages are designed to be informative and help you understand what the problem is with your model so you can solve it. Both warning messages are to inform you that you have placed your source (Hertzian dipole) and receiver in the Perfectly Matched Layer (PML) boundaries of the model. You need to have a basic understanding of the FDTD method and the example models in the User Guide. The PMLs are there to truncate the model domain and the electromagnetic fields inside the PML are usually not of interest.

To answer your questions:

1. The code takes to long to simulate how can I speed simulation. FDTD is a volumetric method, thus computational demands are largely governed by the size of the model. This is true for any FDTD code. Running big 3D models (your model is 500x400x200 = 40 million cells) on laptops or modest desktop machines is demanding. However, gprMax has been parallelised to take advantage of multi-core CPUs, and more recently can run on NVIDIA GPUs.
2. The stone is buried 0.0762 in the soil and the code does not generate the geometry view. Your 'my_stone' sphere is placed before the soil in your input file and therefore will be overwritten by the soil, which is why it doesn't appear in the geometry. Remember gprMax builds the geometry based on a layered canvas approach, so the order of the objects matters.
3. Computing the number of scans to get a B-scan file, its not clear. I did it by taking the size of the x-domain dividing by spatial step but it has not worked, I get errors. Decide what length of B-scan you need. This will help guide the size of your model in that direction. Decide on the inline spacing. This will set the source and receiver steps. Then run gprMax using the '-n' option. The number of A-scans ('-n') will be the length of the B-scan divided by the inline spacing. See the example in the User Guide - http://docs.gprmax.com/en/latest/examples_simple_2D.html#b-scan-from-a-metal-cylinder

Kind regards,

Craig

[Kenneth Obrien M]

Hi Craig,

I kindly appreciate you support.

Due to the high demand of computational space need to simulate the heterogeneous model (sent earlier), my computer cannot handle the task. So decided to use a less demanding lossy soil model. The model below was adopted and modified according to specification I want to evaluate antenna frequency and target size. hope its fine to use it.

#title: B-scan from a metal cylinder buried in a dielectric half-space
#domain: 0.240 0.210 0.002
#dx_dy_dz: 0.002 0.002 0.002
#time_window: 3.5e-9

#material: 10 0.01 1 0 half_space
#material: 3 0 1 0 MyStone

#waveform: gaussian 1 1.5e9 my_gaussian
#hertzian_dipole: z 0.100 0.170 0 my_gaussian
#rx: 0.140 0.170 0
#src_steps: 0.002 0 0
#rx_steps: 0.002 0 0

#box: 0 0 0 0.240 0.170 0.002 half_space
#sphere: 0.120 0.08 0.002 0.010 MyStone

#geometry_view: 0 0 0 0.240 0.210 0.002 0.002 0.002 0.002 buried_stone n

once again thank you.

[Craig Warren]

Hi Kenneth,

Using the Peplinski soil model and the fractal distribution to create a heterogeneous soil, should not really increase the computational requirements - this is mainly the number of cells in the model, i.e. domain size, and spatial discretisation.

The model you are now using is 2D (a single cell slice of 3D), so your Hertzian dipole is effectively a line source. Your source and receiver are position over your sphere (which is now a cylinder in 2D), so you'll have to be careful if you are planning on stepping the source and receiver to create a B-scan. Otherwise, the model looks reasonable.

Kind regards,

Craig

[Kenneth Obrien M]

Hi Craig,

Peplinski soil model and the fractal distribution I'm still working on them.

Question: For a 2D line source (Hertzian dipole), can it be used effectively to evaluate target size or do I have to use a different source.

Kenneth,

[Kenneth Obrien M]

Hi Craig,

Peplinski soil model and the fractal distribution I'm still working on them.

Question: For a 2D line source (Hertzian dipole), can it be used effectively to evaluate target size or do I have to use a different source. Does it mean everytime I need to produce a B-scan to see the size effect.

Kenneth,

[Craig Warren]

Kenneth,

It depends what you mean by size effect. You can easily create a series of 2D models with different size targets and use the Hertzian dipole (line source in 2D) above the target. You could run a single A-scan for each of the different targets and compare the results.

Kind regards,

Craig