This is a well known problem in utilitarianism, primarily because both maximizing the total and maximizing the average can lead to weird answers at extremes, or "repugnant conclusions". If I just do the normal thing, though, and accept what sounds more sensible, I don't have an answer on whether it can be ethical to have children because I'm really just deciding on my ethics based on how I want that question to be answered. So I want a reasonable way to decide what to maximize.

Average utilitarianism would say that if there were two planets that happened to have sentient life, on opposite sides of the galaxy with no chance of ever interacting, then an event which completely and instantaneously destroyed the planet with lower average happiness would be a good thing. That sounds very wrong.

The usual reason for not adopting total utilitarianism is that it suggests that it's better to have large numbers of only slightly happy people than smaller numbers of happy ones. Looking at numbers, I'm not sure that's really a problem. Total utilitarianism would say that in order to have ten times the population, people should be at least 1/10th as happy. That seems reasonable to me. At some point you have so many people that adding another person is more likely to decrease happiness than increase it. So that's the ideal peak population.

[1] This is also a problem with outcomes that involve decreasing the number of people. When someone dies, though, there are generally people left behind who are sad that they are gone. This is much less true with babies people decide not to have.

[2] This is not to say that it's right or wrong for *specific* people to have children. A true utilitarian would likely be wrong to have children because when you have children you need to spend money on them, which is money that if it went to other less fortunate people might have a much larger effect on their happiness. This could be negated if you thought you could raise your children to be very generous.