Jump to content
Publi

Incorrect simulation results in WinProp ProMan with low antenna heights.

Recommended Posts

Hello

 

For a real implementation, I am running simulations of a certain area (parking garage) where the

antenna is only 15 cm above the floor plate. I noticed that the results of my measurements differ

very strongly from the values I find in the simulation.

 

To check if I had made a mistake I tried to recreate the 2-ray path model using a floor

plate with normal material properties (for example concrete). This was successful because the

resulting power (and field strength) values decreased with 40 dB per decade as we expect from the

theoretical explanation. This of course after the oscillations stop (> critical distance).

 

However, when I rerun this simulation with lower antenna heights (coming from 5 m to 0.5 0 or even

0.15 m) the results get worse and worse. The lower I place my antenna (but still above the floor

obviously) the worse my results get. At 0.15 m antenna (and resulting prediction height) I get just

over 20 dB/decade. I have graph where I simulated the exact same plate, exact same antenna

location, but with adjusted antenna and prediction heights where you can see this trend.

 

I feel as though there is something wrong with either the interaction between the antenna and the

floor (incorrect resulting radiation pattern) or that the separate rays don't add up correctly at the

resulting point when they are this close together. I have tried this with several material properties but I still get this result.

 

Any help would be much appreciated.

 

Kind regards

Publi

 

PS: to recreate the results:

 

Step 1: create database in WallMan, long plate (>2km) of a certain with with 5.5 relative permittivity and

0.01 S/m conductivity.

Step 2: Make SRT project and load this into ProMan.

Step 3: Place an antenna with normal gain and isotropic radiation pattern and arbitrary transmitted

power at the start of the plate. Choose for example 10 m height (also for prediction plane).

Step 4: Simulate it and plot the resulting power from the transmitter to the end of the plate.

Step 5: Confirm that you have 40 dB/decade power loss.

Step 6: Repeat starting from step 3 but now choose a lower antenna height and prediction plane.

Confirm that you power loss is now lower than the first case.

2Ray-AntennaHeights.eps

Share this post


Link to post
Share on other sites

You mention SRT (Standard Ray Tracing) and you say that the transmitter is at the start of the plate. I wonder if edge diffraction plays a stronger-than-expected role. It might be worth a try to place your antenna farther from any plate edges, and to disable refracted rays in the SRT settings. Also, make sure the superposition of rays is coherent (taking phase into account) in the SRT settings.

Another thing: the critical distance may change when you move the antenna down. Make sure you calculate it correctly.

Share this post


Link to post
Share on other sites

In addition, I wonder if you are maybe looking at a very large dynamic range (hundreds of dB). That might be a little much. Note that under the parameters of Standard Ray Tracing, there is a setting to exclude automatically the secondary rays below a certain level, e.g. below -200 dB, while keeping always the direct ray, no matter how low. That setting might explain why you see a transition to -20 dB per decade.

Share this post


Link to post
Share on other sites

Hello Publi,

 

independent from the observed behaviour when using the two-path model for low tx antenna heights (there might be some limitation due to the grazing incidence causing lower attenuation rates than 40 dB per decade in distance) I am wondering why you are using the two-path model for the mentioned scenario of a parking garage.

Because in such a scenario you will always have more than two paths (i.e. besides the direct ray, ground reflection also ceiling reflection, reflections at the vertical walls, ...). Therefore in such an indoor scenario the propagation exponents are typically not above 20 dB per decade and in some cases (e.g. in tunnels) even lower.

On the other hand I am wondering why you measured 40 dB attenuation in such an environment.

Share this post


Link to post
Share on other sites

Hi everyone, terribly sorry for the late response; I forgot to switch on "notify me of replies".

 

First af all I want to thank both of you for checking my issue and for your input!

 

@mvogel 1) I excluded diffracton from the results for the reason you mention and also to verify the theoretical result with the simulation, so that shouldn't be a problem. I have indeed selected coherent superposition. 

Yes, that is true, however I find that the critical distance is lower for low antenna's, so if I'm correct I should be well over this distance.

@mvogel 2) At this time the exclusion is at -200 dB, but I will try to play with this number a bit. I'm not sure this will help though, because when I view the .str file, with all the point and ray data, I find that direct en reflected ray are closely related. 

 

@reinerh I felt as though this was the reason as well, I wonder if there is something I can do to increase the accuracy of this angle?

I am not actually using this model to describe the parking garage. I did measurements in this garage and got more or less 33 dB attenuation. I then tried to simulate this same environment and compare the results. Because there was such a big difference I assumed I was doing something wrong. So after many new simulations where I changed numerous parameters (material properties, frequency, radiation pattern, ...), I tried taking one step back and starting from the beginning, just to see where I made a mistake.I was going to simulate theoretically known situations and conforming the data with my calculations.

I started with free space simulation, which was perfectly correct and corresponding with the theoretical expected values.

Then I moved on to the 2-ray model and this was almost completely correct as well, however I noticed that as I lower the antenna, while keeping critical distance in mind, the attenuation started to deviate more and more fro the theoretical values. So because I could confirm both situations I knew that my databases and other settings were more or less correct, but when only changing antenna heights, I got worse and worse results. 

Share this post


Link to post
Share on other sites

Update on the project:

I figured it might be a problem with the thickness of the plate, considering that internal reflections could be added in WinProp and counted as 1 ray. However, simulating with different thicknesses (1cm, 50 cm, 5m, ...) I consistently get the same results. 

 

I am starting to suspect that there might be an issue with the resulting phase and thus the size of the resulting field.

Share this post


Link to post
Share on other sites

×
×
  • Create New...