standard deviation determination for normal distribution

I have 3 intervals :

I have 3 treshold values within the interval [0,1]: 0.05, 0.1 and 0.2. I would like to generate 50 random numbers around this tresholds. These are used later as input for my model. My aim is to generate deviations from the treshold values to see if the response changes.

For now i have used normal distribution to generate these 50 random values.

se.sw <- rnorm(50, 0.005, 0.005/3)
se.m <- rnorm(50, 0.1, 0.095/3)
se.st <- rnorm(50, 0.2, 0.1/3)

My questions are:

- Is there a sensible way to specify the distribution "proporional" to the range between the tresholds?