Burak - I think that you may want a larger population. Ten members may not be enough to introduce enough "genetic diversity" into the population especially with only 25 iterations.
What are the intervals for your four variables? What is the crossover method that you are using - one point or two point?
What have you observed after the 25 generations have completed - did your population converge quickly to the best solution, or do you think that a few more generations may have been helpful?
Can you optimize your (fitness) function so that it takes less than the four minutes per iteration?
