How can I avoid errors due to division by zero in Simulink?

I am using a simple model in Simulink in which I use a division on two input values using a 'Divide' block.

During my simulation, there might be a zero value fed to the denominator of the 'Divide' block. This often causes a warning, an error message, or erroneous results.

How can I avoid these problems?

Answer 1

[MathWorks Support Team]

Numerical division by zero is a common issue in programming, and its exact solution often depends on the particular application. In almost all cases, the best approach is to change the model never feed zero to a division block.

Two possible workarounds are as follows. For clarity purposes, let us call the original signal in the denominator as 'u'.

1. Use a 'switch' block to pass 'eps' instead of 'u' to the 'divide' denominator. The 'switch' must only be activated when the signal 'u' is zero.
2. Use a 'MATLAB Function' block to implement a zero-avoiding condition, such as:

u + (u==0)*eps

or, if the signal 'u' is real:

u + eps*(0^u)

The second workaround is demonstrated in the attached model 'example_no_divide_by_zeroFcn'.

[Walter Roberson]

At one time, MATLAB Fcn Block was called "Embedded MATLAB Block"

Answer 2

[madhan ravi]

Or making the result to be zero when the divisor is zero using a Switch block