Scribd
Upload
32 views

L15 Syllabus

The document discusses the constraints on cell size (dx) and time step (dt) for stability in finite-difference time-domain (FDTD) simulations. It states that dt must be less than or equal to dx/cmax to ensure stability, where cmax is the maximum wave propagation speed in the model. It also notes that numerical dispersion, where wave propagation speed varies with frequency and direction, can occur due to the finite dx and errors from numerical differentiation in the FDTD method. The document recommends choosing a dx no larger than λ/10 to limit numerical dispersion errors to less than 10%.

Uploaded by

Wasyhun Asefa
Copyright
© Attribution Non-Commercial (BY-NC)
Available Formats
Download as PDF, TXT or read online on Scribd
32 views

L15 Syllabus

The document discusses the constraints on cell size (dx) and time step (dt) for stability in finite-difference time-domain (FDTD) simulations. It states that dt must be less than or equal to dx/cmax to ensure stability, where cmax is the maximum wave propagation speed in the model. It also notes that numerical dispersion, where wave propagation speed varies with frequency and direction, can occur due to the finite dx and errors from numerical differentiation in the FDTD method. The document recommends choosing a dx no larger than λ/10 to limit numerical dispersion errors to less than 10%.

Uploaded by

Wasyhun Asefa
Copyright
© Attribution Non-Commercial (BY-NC)
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

ECE 5340 / 6340 FDTD: Stability and Dispersion (Constraints on Cell Size and Time Step)

For FDTD Stability we have to choose the time step (dt) small enough. If the simulation becomes unstable, it will blow up. For stability, dt is related to dx.

How far will wave travel in 1 time step? dx = (dt) (c) c is the speed of light. Usually we use the speed of light in vacuum, because that is the fastest it will go. For a simulation that does not have any air cells, we can let c be the fastest velocity of propagation in the model. This is typically in the cell with the lowest conductivity and permittivity. It is OK to sample more often in time, but not less. dx (dt)(c) ; dt dx /c

In 2D: dt 1/ (cmax sqrt( 1/dx2 + 1/dy2) ) For dx = dy dt dx/ (sqrt(2) cmax ) In 3D: dt 1/ (cmax sqrt( 1/dx2 + 1/dy2 + 1/dz2 ) ) For dx = dy = dz dt dx/ (sqrt(3) cmax ) Numerical Dispersion: (dx) Dispersion: wave propagation velocity in numerical simulation may vary with frequency, direction of propagation, and dx. happens naturally in non-air
1
Dr. Cynthia Furse UNIVERSITY OF UTAH DEPARTMENT OF ELECTRICAL AND COMPUTER ENGINEERING 50 S. Central Campus Dr | Salt Lake City, UT 84112-9206 | Phone: (801) 581-6941 | Fax: (801) 581-5281 | www.ece.utah.edu

ECE 5340 / 6340 FDTD: Stability and Dispersion (Constraints on Cell Size and Time Step)

happens numerically (error) in differential forms because of error in numerical differentiation Example: Pulse dispersion

See Dispersion curves in Taflove FDTD handout. dx / 10 gives less than 10% error. This is similar to what you observed in the numerical differentiation homework.

2
Dr. Cynthia Furse UNIVERSITY OF UTAH DEPARTMENT OF ELECTRICAL AND COMPUTER ENGINEERING 50 S. Central Campus Dr | Salt Lake City, UT 84112-9206 | Phone: (801) 581-6941 | Fax: (801) 581-5281 | www.ece.utah.edu

You might also like