During model development, a number of experiments were performed in which particular importance was placed on vapor generation. Key tests were aimed at validating this aspect of the model under quiescent, engine-off (soak/diurnal) conditions; during refueling; and under high-temperature engine-on conditions.

Diurnal canister loading—A tank and a number of canisters were subjected to a 72-h diurnal-temperature cycling (22-35&#deg;C or 72-95&#deg;F) in a mini-SHED (Sealed Housing for Evaporative Determination). An 81-L (21-gal) tank was filled to 40% with 60 kPa (8.7 psi) Reid Vapor Pressure (RVP) certification test fuel and connected to an electronic scale by a balanced tube arrangement so that the weight gain of the canister could be monitored during the test.

Good agreement for the canister weight indicated that the vapor generation submodel is accurate under these conditions at relatively low temperatures when the fuel does not boil. Poor agreement for the exact magnitude of the SHED emissions reflects the sensitivity of the vapor migration to the amount of purge under conditions near breakthrough as well as two-dimensional effects, which were not taken into account in the model. The results are qualitatively acceptable and permit useful parametric studies.

Refueling emissions—The model can also compute the emissions from the tank during refueling. During a fill event, the vapor-air mixture in the tank is displaced by the incoming liquid and is trapped by the evaporative emission control system. In addition to that in the tank, some air is entrained within the nozzle as part of the shutoff mechanism and sweeps additional vapor out of the tank. Predicted vapor emissions are compared with experimental data for refueling (60 kPa (8.7 psi) RVP fuel at temperatures of 19&#deg; and 27&#deg;C (66&#deg; and 81&#deg;F)) into a tank (initially at 27&#deg;C (81&#deg;F)) as a function of entrained airflow.

Good agreement is obtained for the lower dispensed temperature case and for low entrained air rates. The model fails to replicate the slope at the higher fill temperature, meaning the air leaving the tank has a higher concentration of vapor in it than the model predicts. Since the model assumes continuous equilibrium between vapor space and liquid, this suggests possibly that there is some liquid droplet creation during the fill event, which would produce higher than equilibrium vapor concentrations.

Vapor generation (running loss)—Under the nonboiling conditions of the previous section the emissions are governed by the temperature-induced swing in vapor pressure of the fuel and the size of the vapor space that expands and contracts above the fuel. When the vehicle is running under generally higher heat loading, the fuel may enter the boiling regime resulting in significant evaporation. In this case, evaporation depends primarily on the distillation characteristics of the fuel and the quantity of liquid fuel in the tank.

A subset of data from a special Ford test rig, with separate controls over tank and rail heating and fuel rail flow rate. Under conditions of high rail temperature and flow, the fuel reservoir region was significantly hotter than the bulk fuel temperature. Although bulk liquid temperature may give a good indication of vapor generation under diurnal conditions, it can be misleading in the case of engine-on running conditions.