In my opinion, this is within the normal range and you should not worry.
First of all, I don't agree with the 19.4 target value. This number was seen often here but was based on assumptions about the usable percentage. Now that we have OBD scanners, it appears that the usable range is fromapproximately 90% to approximately 8%. 82% of 23 kWh is 18.9 kWh. I would consider this my current estimate of the nominal capacity when measured in a specific manner.
Then, your specific result is influenced by the details of your test. Key factors include:
1. Battery temperature during the test...lower temps give lower available energy
2. Power levels during the test...higher power levels turn more stored energy into heat, resulting in less delivered energy
3. Amount of regeneration during the test. Only some portion of the regenerated energy actually gets stored back in the battery, and some portion gets lost in the charging process. The trip meter sees regeneration flowing into the battery terminals, but it doesn't know how much of this energy is converted into chemical energy and how much is lost to heat.
The only way to get a solid number is to use a battery tester and perform the test under specified conditions. Using the trip energy meter is a good approximation but is not exact. Another approximation is the energy to empty value shown in the OBD scanner.
If you really want to know your battery's capacity, buy one of these:
http://www.avinc.com/testsystems/product_matrix/abc-150/#product
Otherwise, don't worry about small differences between your trip meter and what we believe to be the expected capacity. If it makes you feel better, do the test:
1. With the battery at 98 F
2. On a level course with no regeneration, driving maybe 30 mph
I think this will give you a slightly higher number.