EVAP Testing: Working Behind the Scenes

This was originally published July 2006.

The newest EVAP leak detection system is designed to perform its tests right there on the driveway.

Before cars had emission controls, gasoline evaporating from the fuel system generated almost as much unburned hydrocarbon emissions as the engine did. Sealing the fuel tank to prevent evaporative emissions was not a solution because pressure builds up with heat, making the emissions problem potentially a lot worse.

Today, the pressure is vented through a tube that leads to a canister filled with activated charcoal that absorbs fuel vapors. When the engine is running, the vapors are drawn into the intake manifold and burned in the engine.

It seems simple enough, but as it turns out, engineering an effective and legal evaporative emissions control system (EVAP) is not as easy as designing a leak-free fuel system. That's because the EVAP system is opened at both ends at least once every drive cycle to purge the vapor storage canister. It's also opened every time the tank is refilled, introducing the opportunity for malfunction caused by human error. Add in the on-board refueling vapor recovery system (OVR), and the on-board diagnostics (OBD) system that checks the whole EVAP system for leaks, and things can get complicated in a hurry.

THE LOGIC

Designing any system that can be tested while it's actually in use is a big challenge. Powertrain engineers will tell you that's why testing the EVAP system is the most difficult part of OBD. OBD II checks for leaks by putting the EVAP system under vacuum, closing all the valves to seal the system and then measuring the rate of vacuum decay.

Test vacuum is easy to generate, using either manifold vacuum or a leak detection pump (LDP), but the test vacuum must be very small to avoid the possibility of creating a leak in an otherwise tight system. Test vacuum is typically about 7 inches of water (H2O), and because the OBD system is looking for a leak equivalent to a hole only 20 thousandths of an inch, measuring vacuum decay requires very specific test conditions that are hard to meet in the real world.

In general, those conditions are:

• Fuel level between 15 percent and 85 percent.
• Ambient temperature between 40°F and 86°F.
• Coolant temperature close to ambient (cold start).
• Altitude below 8,000 feet (barometer greater than 75 kpa).
• Engine at idle or steady speed/load for a specific period of time.

Because the EVAP system is tested with vacuum, the test must run under conditions least likely to generate fuel vapor in the tank, which would increase pressure in the fuel tank and cause vacuum decay that is unrelated to a possible leak. Fuel vapors are generated when warm fuel is returned to the tank after passing through the fuel rail of a warm engine. Vapor pressure also is increased by fuel sloshing around in the tank when the vehicle is driven in stop-and-go traffic. Finally, fuel evaporates more easily at high ambient temperature and at low atmospheric pressure found at higher altitudes. The time required to run the test to completion – attain the target vacuums – varies with fuel level, and test accuracy cannot be assured when the level is above or below certain limits.

Depending on where and how the vehicle is typically used, it's common that the EVAP monitor will not run to completion for months at a time, and it's even possible that the test will never be completed. If the monitor does run and detect a small leak, it must do so twice before it will set a hard fault code and turn on the malfunction indicator lamp (MIL).

However, there is one way to leak test an EVAP system using test conditions that are very easy to meet. Called Natural Vacuum Leak Detection (NVLD), it was introduced by DaimlerChrysler (DCX) on the 2002 LH platform. DCX and other manufacturers are gradually switching over to this type of EVAP test as new models are introduced. Like existing Ford and General Motors (GM) EVAP systems, it uses solenoid valves at each end of the system, so DCX will gradually phase out the old Leak Detection Pump (LDP). The big advantage of NVLD is that it can detect small leaks with great accuracy because the test runs with the vehicle parked and the engine turned off.

NVLD takes advantage of Boyle's Gas Law, which describes the relationship between volume, temperature and pressure of a gas. Basically, if volume is held constant and either of the other two changes, so will the other. In a fuel system, the volume of air in the tank is held constant when the engine is off and the EVAP system solenoid valves are all closed. If temperature increases or decreases, so does pressure in the tank, as long as the EVAP system is not leaking. Engineers have learned over the years that the vapor pressure in the fuel tank drops slightly in the first 20 minutes or so after shut-down. This is true even on a cool day and on vehicles with a returnless fuel system because the fuel stops sloshing around in the tank. So all that's needed to leak-test an EVAP system is to detect the resulting pressure drop with the ignition switched off. To be sure, that pressure drop can be miniscule and last a very short period of time, especially in warm ambient conditions. However, it occurs far more often than under the general conditions mentioned earlier.

THE HARDWARE

The NVLD assembly can be mounted anywhere, but on DCX's LH platform, it's mounted directly on the EVAP charcoal canister. That canister is mounted under the car right next to the fuel tank. The assembly includes a diaphragm-operated switch, two check valves and the fuel system vent valve, which is a normally closed solenoid valve.

One of the check valves is a pressure relief valve for the entire fuel storage system. It opens at a pressure of only 0.5 inch H2O so air can escape during refueling and/or during heat-soak expansion of the fuel in the tank. Air escaping through that check valve has already passed through the charcoal canister, so fuel vapors are captured. The other check valve is sucked off its seat when vacuum in the fuel tank exceeds 3 inches to 6 inches H2O. This relief valve limits the amount of vacuum in the tank during engine-off cold-soak, and it also acts as a backup in case the vent valve fails. The solenoid-operated vent valve is opened by the powertrain control module (PCM) during EVAP canister purge and to vent air into the fuel storage system while the engine is running. Power is turned off to close the valve for certain OBD leak tests and purge flow tests.

There are three wires connected to the NVLD housing. The white/dark green wire (PCM pin C3-8) carries 12 volts to the vent valve solenoid from the PCM. The red wire (PCM pin C3-35) carries a 5-volt reference signal from the PCM to the NVLD switch; this is the switch sense circuit. The black wire is a chassis ground for both circuits. Depending on the model, there might be a resistor or even a diode on the switch, so checking for continuity between the switch circuit and ground with the NVLD switch closed will yield some resistance. A scan tool that can read the switch circuit will simply report the switch open or closed.

Unlike Ford and GM models, which read fuel tank pressure with a sensor, the NVLD system uses a simple on-off switch operated by a diaphragm. The normally open NVLD switch will close with a vacuum of only 1 inch H2O. When checking for large leaks, the test strategy is similar to that used on other EVAP monitors: Close the vent valve and open the purge valve, using manifold vacuum to pull a vacuum on the EVAP system. The PCM merely needs to see the NVLD switch close to know that the system is under vacuum.

Remember, there's a check valve that opens at 3 to 6 inches H2O, so vacuum will never build any higher than that, but it's quite sufficient for the large-leak test. However, the PCM won't bother running the large-leak test – that requires very specific operating conditions – if the system passes the small-leak test, which is done with the engine off. And that's the real advantage of NVLD.

THE TEST

All that's needed for the small-leak EVAP test to run is an ambient temperature greater than 40°F and fuel level less than 85 percent. When the ignition is switched off, a switch monitor in the PCM remains powered up. As the fuel in the tank settles or cools down, vapor pressure decreases and creates a natural vacuum in the tank. Remember, the purge valve and vent valve are closed with the key off.

This vacuum will be slight, but the NVLD switch only needs a vacuum of 1 inch H2O to change state, so if the EVAP system is not leaking, the switch will close. The switch monitor will note the switch closure, record the amount of time since key-off and then shut itself down. The next time the ignition switch is turned on, the switch monitor will report the information to the OBD system, the timers will be reset, the test will logged as "passed" and the PCM will not bother running the large-leak test during this drive cycle.

In general, after the engine is turned off, it should take only about five minutes to build 1 inch H2O vacuum in the fuel tank, but if the NVLD switch never closes, the switch monitor might run for half an hour before aborting the test. When testing for parasitic battery draw after engine shutdown, the PCM (switch monitor) will continue to draw current until the small-leak test is completed or aborted.

Obviously, there are some operating or ambient conditions that can prevent a build-up of natural vacuum in the tank, even if the system is tight. If that happens, the OBD system will run the small-leak test at each shutdown for one week without a single "pass" before it decides to turn on the MIL. The OBD system has timers to monitor engine-off time and engine-run time, giving it the ability to roughly calculate real time.

To make sure the NVLD switch is working, the PCM runs a "rationality test" with each drive cycle. When the ignition is first switched on, the solenoid vent valve in the NVLD assembly is opened to vent any vacuum in the system. The NVLD switch should open. When the PCM starts purging the EVAP canister after closed-loop begins, the vent valve will be closed briefly to draw a slight vacuum on the system, and the NVLD switch should close. If the switch doesn't operate as expected, a pending code is set. If it fails in two trips, the MIL will be turned on and P0452 (switch sense circuit low) or P0453 (switch sense circuit high) will be set. There may be other codes too, because the actual failure could be electrical, electronic or an EVAP leak.

The NVLD diaphragm is in the top of the NVLD assembly housing, and it pulls up to close the switch. A passage transfers vacuum from the inlet on the bottom of the housing to the top of the diaphragm. If the NVLD housing is damaged, even at the electrical connector, there will be no vacuum transmitted to the diaphragm. In fact, the EVAP system will leak.

As noted previously, the NVLD assembly on some models is mounted directly on the EVAP canister, but it also can be mounted anywhere on the vehicle. If you decide to use a hand vacuum pump to test for switch closure, remember that the switch is supposed to close with only 1 inch H2O of vacuum, and that it's protected with a check valve that opens at 3 to 6 inches H2O.

WHAT'S NEXT?

GM introduced Engine Off Natural Vacuum (EONV) testing in its 2003 trucks, and Ford began using EONV in 2005 F-Series trucks sold in California and other "green" states. Toyota, Mazda and other automakers are also using it in some their models. This system uses a pressure sensor instead of a simple switch, and the enable criteria are stricter because the monitor must calibrate itself for ambient conditions. Also, because the PCM may remain active for up to 45 minutes after engine shutdown, it's being introduced as each manufacturer introduces new low-current-draw PCM platforms.

It will take several years and the strategy may see some other changes, but natural vacuum leak detection is proving to be so reliable, eventually all manufacturers will use it for small-leak EVAP testing.