ACS data are not aggregated. They are estimates based on 5-years worth of surveys. So essentially they are a period estimate that is benchmarked to the average annual population estimate over the given time period.
There are no one-year population estimates at the block group level, unless some 3rd party has developed something and I would not recommend using that unless you are confident in their methodology. The margin of error (MOE) for block groups is already pretty high and really block group data from the ACS 5-year should only be used for creating estimates of larger areas that aggregate block data (see the Census technical documentation on this).
You will need to either use Decennial census data - which are actual counts of the population every 10 years or ACS 5-year estimates data. Decennial Census data may be your best option. If you use the 5-year ACS data you would need to use period estimates for non-overlapping periods which means that right now you can only get 2014-2018 (most current) and 2009-2013.