I'm working with a dataset in ArcGIS 10.1 that has an approximate range of 650 km N-S, and 400 km E-W, across southern BC, Canada and northern WA, USA. I need to find a projection that will conserve distance information within this range, as I will be doing analysis/calculations regarding distances. I had recieved some advice to make a custom projection, starting with the NAD 1983 BC Environment Albers projection. So I did this by setting the central median to the midpoint of my longitudes and the two standard latitudes to a bit below (1/6 of total spread) my northmost latitude and above my southmost latitude. The data is originally unprojected, with a Geographic Coordinate System of WGS84, so I changed the projection's GCS from NAD83 to WGS84, as well, to avoid transformations of lat/long.
One of my primary concerns with the projection is that Albers is equal-area, whereas I had originally assumed that an equidistant projection must be the best choice. Equidistant projections tend to only conserve distance information from one point, though, and I will need to look at distance information between a number of arbitrary points, so I now feel like equidistant may not be the best. As well, Albers is conic, which I have read may be better for data that is more spread out E-W than N-S, like mine is, but the alternatives to Albers that have been suggested are usually a UTM projection, which is a conformal projection. As I understand it, this would be a bad choice of projection for doing most kinds of measurement/analysis.
Given this information, can someone tell me if I have made an appropriate choice of projection? Any feedback/comment is much appreciated! Let me know if any additional information would help.