Friday, March 16, 2012

Carburetor Bowl Vent Canister

rev. 4-29-2012

Surely it appears I'm going overboard on this topic, but I love experiments.

If a carbon canister can reduce fuel loss and water gain in a tank, why not address the carburetor? Wouldn't this reduce gum formation and ease starting? And since it won't cost me anything but time in the shop, isn't it worth a go? Tiny carbon canisters have been fitted to California lawn tractors for some time, to meet stringent air regulations.

In automobiles evaporation emission systems tied the carburetor vent--back when we had carburetors on cars--to the main canister and flushed the system with intake air during operation. More efficient in terms of hydrocarbon destruction, but more complex and only effective if the car ran nearly every day. In fact, those systems did not reduce carburetor bowl evaporation, since they would never reach equilibrium. They actually drew gasoline from the bowl, much as calcium chloride draws water from the air. I've taken a different approach, since evaporation reduction is a primary goal, and that requires that the carbon reach equilibrium with the carburetor bowl while the engine is running, exactly the opposite sequence from what automobiles systems achieve. Since the carbon is saturated during operation--a warm engine encourages plenty of evaporation--when fresh air tries to enter the carburetor when it cools and at night, that fresh air will become saturated with gasoline vapors and will not contribute to further evaporation in the morning. This process of self-regeneration is about 65% efficient in reducing emissions for long time periods, according to EPA testing (page 5-120).

The carbon canister is above the in-line fuel 
filter and is connected to the carburetor vent hose.
Click to enlarge.

The canisters I installed are retrofitted from small in-line gas filters from the might-need box. Originally, they contained fritted bronze elements, which I had discarded (these were replaced by Raycor filters). I fitted each end with a carbon support screen cut from beach cat trampoline fabric. They hold 6 ml of granular carbon of a type optimized for low-flow gas phase  hydrocarbon adsorption. Though small, the carbon:gasoline volume ratio is much greater than used on tank vent filters, this is intentionally so; these will see much greater temperature operating temperature range because they are mounted beside the engine. Even the diurnal cycle will be greater, since the carburetor lacks the thermal mass of a fuel tank. They are NOT secured directly to the engine, as the high-frequency vibration is not good for the carbon. It is extended away from the carburetor, horizontally and slightly down hill, with 6 inches of hose to prevent fines from migrating back into the bowl. This is the existing hose, previously open ended and routed downwards between the carb and engine block.

What do I expect? Easier starting. Less idle jet gum formation and fewer stuck floats. I'll let you know.


Follow-up reports: 
  • Baseline, March 18. Both engines require primer to be pumped up if the boat has been help for a few days. Port requires 1-2 pumps of the throttle (Yamaha 9.9s have an accelerator pump). Starboard requires several pumps and several efforts, followed by an advanced throttle for a few minutes. Engines re-start within 18 hours, even if cold, with a single crank and idle well. Both engines are identical, purchased at the same time except for the carburetor; the starboard is a fixed idle jet US carb, and the port is Canadian with an adjustable idle screw.
  • March 25. Easier starting after a week away seems to be confirmed. Both engines start on the first crank, without pumping up the primer or pumping the throttle. The starboard engine did not require the throttle to be advanced during warm-up and did not stall.
  • April 20. Up-graded to 1/8-inch thick FRP filter with silica gel/alumina fill. This adsorbent has been performing better in bench trials when high humidity is considered. Very sturdy, better fit, easier to mass produce, twice the adsorbent volume. Non-refillable, but projected life is 5 years.


  1. Drew, I am confused. As you state, incoming air will get essentially saturated with gasoline vapor going thru the canister. But when the engine is started and heats up, the outgoing air will carry an equivalent amount of gasoline vapor out, won't it?

    Without a mechanism to desorb at least some of the gasoline vapor, won't this just be the equivalent of a longer vent pipe?


  2. By EPA's own calculations, passive diurnal breathing canisters are about 65% efficient even with no special regeneration step. Thus, the canister has a 7-10 year life expectancy.
    (see page 5-120 of the document, or page 122 of the PDF viewer)

    The nightly inflow of air washes some portion (65%) of the gasoline off the carbon and back into the tank or carb bowl, where it will either condense or simply prevent evaporation the next day. Much of this is actually displacement of gasoline by water vapor, which is also thus kept out of the tank, which is also a good thing.

    Yes, sounds too good to be true and it sounds like we should have done this a long times ago. In fact, I have installed systems like this on big gasoline tanks (20,000 to 100,000 gallons) for this purpose, paying for the projects with the savings. These projects were always in response to regulatory mandate; I just never made the connection to boats, since it only became vitally important with e10 and biodiesel.