There are two aspects to this question, future growth and fire station siting.
Firstly, lets look at future growth. You can use the Census 2010 data in conjunction with the ACS data to attempt to spot recent trends in growth. The ACS data must be treated carefully and in many cases is not directly comparable to the Census data - see the Census ACS guidance for details on reliability.
You should also go backward and get older Census data as well to build up a long term trend, however the Census data may have to be modified to be directly comparable (depending on the summary level used). This blog has a good write up of some of the issues faced in calculating a simple growth rate for a county, I am sure there are plenty of others.
Test your methodology with older data (say 1990-2000, or 2000-2010) and see if it seems a reasonable predictor of the actual values. Also compare the ACS data with the actual Census data for those periods, this will help you develop your methodology for handling your post 2010 period (assuming you are looking 2013-2023). that should give you an estimate of the "how many" aspect.
You may also need other information to help you narrow down the "where". Planning departments may be able to help you identify areas currently zoned for future residential development (ie an indication of where population growth may reside) and commercial development. Also real estate companies can be a useful source for gauging housing demand and the desirability of areas, which may factor into your second part. Existing road networks and underutilized developments may also give an indication of where future growth could occur first. Or you could assume a degree of infill + development growth and estimate family/house sizes and the like to get a rough land area needed for the extra population. The census geometries might also help you here.
The second part, siting fire stations should be reasonably straightforward once you have created your estimated population geometries and gathered the appropriate supporting datasets. ESRI have a white-paper (http://www.esri.com/library/whitepapers/pdfs/gis-for-fire.pdf), and there are examples in the ESRI Spatial Analysis Workbook or in the online help (spatial analyst tutorial). You will have to decide the variables that are most important; response time is a common one but requires a good network dataset, otherwise you might choose distance as a rough proxy (eg a buffer with 1 mile = 3 minutes for example) but be sure to stress the limitations of such an approach. You might be bound by restrictions that you will have to try and take into account such as existing stations (Is it feasible to move an existing station? Zoning restrictions?) and possibly capacity (how many incidents they can respond to?). Ideally you will be able to source data on response types and times for the existing fire stations to base this on.