Ejector Flare Test

Date: 2/5/09

Purpose:

test an ejector replacement for GEK fan
optimize combustion to reduce CO (per Crispin Pemberton-Pigott's comments/recommendations) with proper fuel air mix
minimize emissions by premixing propane into flare gas for startup
lower priority: test use of 1/4-1/8" char filter media

Photos

Testing setup. Compressed air and propane flow metered by rotameter.

Ejector mounted above filter. T and ball valve used before the ejector to change fuel/air ratio.

Flare and ejector. Ejector made from reducing 2"->1" 90 with hole drilled on center of 1" outlet for compressed air inlet (far left). Length of 1" pipe to 1"->2" reducer. Reducer tack welded onto flare.

Inlet tube : .25" OD, 0.172" ID. Back pressure at 100 lpm = 10 psi. The run always used less than 100 lpm (below the range of the rotameter).

Rotameters (poorly aligned in photo). Left rotameter for compressed air (100-500 lpm), right for propane admix (5-25 lpm). Air rotameter was out of range (general use was under 100 lpm).

Flare running on pure wood gas. Note temp gradient at location of perf metal. With good mixture, combustion occured largely within reducer (note highest temps opposite of reducer).

View of flare with perf stainless. The perf stainless was not used for the whole run, but did contain heat inside the lower end of the flare. Presumably this would limit CO emissions. Tests need to be run with a combustion analysizer/CO meter.

Start Up

Step 1: Low flow of compressed air introduced into ejector

Step 2: Torch placed in front of flare and propane added until it lit. Mixture adjusted to get complete combustion at bottom of flare (~20 lpm propane)

Step 3: Starting bed of char prepared in GEK (to air nozzle level). Small amount of mineral spirits (~10 mL) added above reduction cone. Torch used to ignite and briefly heat above reduction. Char above reduction glowing in ~30 sec.

Step 4: 2 gallons of wood pellets added above char bed.

Step 5: Within approx. 30 seconds, flame starting to exit flare (fuel/air ration increasing). Reduced propane admix by ~5 lpm. Repeated until propane was off. (Clean burn through whole process).

Results

Ejector worked very well. No quantitative information yet (working on it), but got quite good flow. At some points optimal combustion occured with ball valve fully off. It would be good to optimize the ejector so that we can have some leakage, allowing a switch to a butterfly valve (eventually servo controlled).

Combustion was maintained well within the flare, which should reduce emissions. A set of tests with a combustion analyzer would be good.

Propane premix fully eliminated any visible emissions during start up. Can also eliminate most during shutdown by gradually backing off of flow rate allowing the reactor to slowly cool (water injection/quench may be an option for faster shutdown). Maintaining a slight vacuum on the system keeps leaks to a minimum.

Piezo spark was tested with pure woodgas to test ignitability. It worked.

Filter media was surprisingly dry with very little coating with tar. Perhaps some channeling occurs or it wasn't very absorbent. Perhaps moistening it slightly would aid filtration.

Additional Information

Back of the envelope calculations:

One US gallon of liquid propane = 1030 liters gaseous propane.

Start up may have used on the order of 20-50 liters propane (with automation/more runs this could be optimized).

5 gallons propane (l) * 1030 L propane (g)/1 gal propane (l) = 5150 l propane (g) per 5 gallon canister

5150 L/50 L per start up = 103 start ups on one 5 gallon canister.