Two Fuel Tank Temperature Profile cycles—Heat transfer calibration was carried out for a vehicle operated over two back-to-back Fuel Tank Temperature Profile (FTTP) cycles. For this calibration, experimental measurements of representative underhood and underbody temperatures were obtained within 25 mm (1 in.) of the fuel rail and 25 mm (1 in.) below the gas tank, respectively. With heat transfer areas known from the vehicle geometry, heat transfer coefficients for the various parts of the cycle were then chosen to match the experimentally determined fuel tank temperature.

The long drive cycle is useful for calibration because the tank temperature reaches a steady-state value late in the test, allowing a determination of the ratio of the underhood and underbody heat transfer coefficients. The magnitude of the coefficients determines the rate of temperature rise. In most calibrations, good agreement was found by using approximately equal heat transfer coefficients. However, in this case, best results were obtained with a significantly larger underhood coefficient resulting from the increased thermal contact between rail and manifold.

Three-day federal test—A complete three-day EPA test procedure was simulated to illustrate the full potential of the FVSMOD. This test began with the FTP-75 emissions test procedure at 24&#deg;C (75&#deg;F) followed by a 4-h soak at 35&#deg;C (95&#deg;F). The vehicle was then driven over the 72-min., 39-km (24-mi) run-loss evaporative emissions cycle consisting of Urban Dynamometer Driving Schedule (UDDS)-New York City Cycle (NYCC)-NYCC-UDDS in a SHED at 35&#deg;C (95&#deg;F). This was followed by a soak of 1 h at 35&#deg;C (95&#deg;F) and 6 h at 22&#deg;C (72&#deg;F) before the vehicle re-entered the shed for three 24-h diurnal temperature cycles over the range of 22-35&#deg;C (72-95&#deg;F). Evaporative HC emissions were measured during both the run-loss and the diurnal portions of the test—they did not exceed 0.05 g/mi in the run-loss test, while the 24-h diurnal period plus soak emissions did not exceed 2 g (0.07 oz). At the beginning of the test, the fuel tank was 40% filled with 62 kPa (8.7 psi) RVP fuel and the canister was saturated with a 50/50 (by volume) mixture of n-butane and air.

To pass the test and prepare for the diurnal cycles, the canister must be purged early. During the hot run-loss test, vapors must be consumed as much as possible as they are generated. During one of the purge-off intervals near 350 min., the canister breaks through with an escaping vapor mass of about 10 g (0.35 oz). The breakthrough was surprising in that the canister has only about 130 g (4.6 oz) of vapor on it, significantly less than the initial saturated value of 212 g or 7.5 oz. Breakthrough occurred because the canister was much hotter than the initial state, due partly to the 35&#deg;C (95&#deg;F) ambient temperature and to the heat of vaporization released during the rapid vapor loading. The capacity of carbon canisters was reduced at higher loading rates due to the inability of the canister to release the heat of adsorption.

Increasing the purge by 20% to 1160 L (306 gal) reduced emissions during the breakthrough to 0.02 g/mi, a value below the maximum allowed value of 0.05. Reducing the purge by 27% to 700 L (185 gal) doubled the breakthrough to 23 g (0.81 oz).

In all cases, the maximum 24-h emissions were less than 2 g (0.07 oz), indicating a successful test, although the low purge case was marginal at 1.5 g or 0.05 oz. Note that the canister was trapping over 90% of the emissions during each diurnal cycle and was not near saturation. HC emissions to the atmosphere were due to diffusion along the carbon bed during the long time period available in the diurnal test. This can be reduced greatly by increasing the total purge prior to the test.

The FVSMOD describes the effects of fuel volatility, vapor generation, vent system design, and carbon canister loading and purging. Although heat is the driving force for vapor generation, the focus of this model was on the fuel and fuel vapor and how it is generated, stored, and consumed. Modeling of the heat transfer environment was handled by empirical calibration of fuel temperature rise rate from representative data or from prior experience.