You are on page 1of 9

2001-01-0733

Impact and Control of Canister Bleed Emissions


Roger S. Williams and C. Reid Clontz
Westvaco Corporation, Covington, VA

Copyright 2001 Society of Automotive Engineers, Inc.

ABSTRACT
Current EPA and CARB regulations allow a maximum of 2.0 g/test for Hot Soak + Diurnal evaporative emissions. The State of California has adopted LEV II regulations that will decrease the evaporative emissions standard to 0.5 g/test starting in the 2004 model year. These regulations also include a Zero Emission Vehicle or ZEV program. The ZEV program allows car manufacturers to substitute vehicles that meet the SULEV tail pipe emission standards and have zero fuel evaporative emissions for electric vehicles. The increased stringency of these regulations has necessitated significant decreases in hydrocarbon emissions from evaporative emission canisters. For example, canister vent emissions may be at levels of 100-300 mg/test for a vehicle that meets the current regulations. However, canister emission targets should be 50 mg/test and less for LEV II and 10 mg/test and less for zero evaporative emission vehicles. Emissions at this level are not due to a lack of adsorptive capacity in the canister, but rather are due to diffusion of hydrocarbon species. These emissions are often referred to as bleed emissions. A technique was developed to study the level of bleed emissions specifically from canister vent ports. Canister design and purge volume were shown to have a significant impact on bleed emissions. Further, the incorporation of a small auxiliary chamber in series with the primary canister was shown to decrease bleed emissions significantly.

INTRODUCTION
Activated carbon has been used since the early 1970s to capture gasoline vapors from vehicle fuel vapor systems. The activated carbon is housed in canisters and is required to adsorb gasoline vapors at high efficiency and release them during the purge regeneration cycle. The canisters are used as part of a system to control evaporative emissions. This paper is concerned with hydrocarbon emissions from evaporative emission canister atmosphere ports specifically during diurnal testing. Diurnal emissions occur while a vehicle is parked and the fuel tank is

heated due to daily temperature changes. Both the California Air Resources Board (CARB) and the Environmental Protection Agency (EPA) have established test standards that measure and control hot soak and diurnal emissions during real-time, two- or three-day procedures. These procedures, adopted for 1995 (CARB) and 1996 (EPA) model year vehicles, have identical standards of 2.0 g/test maximum emissions for the two and three-day tests in addition to the SHED measured emissions from a 1 hour Hot Soak test. The procedures differ somewhat in their temperature profile and test fuel volatility. Recently, CARB adopted LEV II regulations that will decrease the evaporative emissions standard for all cars and light trucks to as low as 0.5 g/test starting with the 2004 model year. Additionally, the LEV II program calls for 10% of the 2003 model year fleet to be Zero Emissions Vehicles (ZEV or battery electric vehicles). The law allows the car manufacturer to substitute vehicles with very low tailpipe emissions (SULEV) and zero fuel evaporative emissions to satisfy some of the ZEV requirement. The increased stringency of these regulations has necessitated significant decreases in hydrocarbon emissions from evaporative emissions canisters. For example, under the current standard, emissions from carbon canisters could be as high as 0.1 - 0.3 g/test for an in-compliance vehicle. However, canister emissions targets for near-zero evaporative emissions vehicles (0.5 g/test standard) should be 50 mg/test and less. Target canister emissions levels for zero-evap vehicles (0.0 g/test) are in the 3-10 mg/test range. Emissions at this level are not due to a lack of adsorptive capacity in the canister, but are due to diffusion. These emissions are often referred to as bleed emissions. Bleed emissions are those emissions from the canister that occur prior to breakthrough. Breakthrough is defined by ASTM D 2652 as the first appearance in the effluent of an adsorbate of interest under specified conditions. In the automotive carbon application, breakthrough is usually considered to be that point when 2.0 g of hydrocarbon have been emitted from the canister or when the total hydrocarbon concentration in the effluent reaches 5000 ppm. Canister bleed emissions during diurnal testing are rarely due to carbon bed inefficiencies but are rather due to diffusion. That is, the bleed emissions most likely consist of hydrocarbons that were already adsorbed in the canister prior to the start of the diurnal. Repeated loading/purging of carbon canisters produce a hydrocarbon concentration gradient within the carbon bed (1,2). The adsorbed hydrocarbon concentration will tend to be higher near the fuel tank port of the canister (vapor inlet) and gradually decreases as the vapor flow path approaches and exits the canister atmosphere port (purge inlet). There will exist some concentration of hydrocarbon at the atmosphere port, even after extended

purging. The amount of hydrocarbon remaining after purge is referred to as hydrocarbon heel. During extended soak periods, the hydrocarbon vapors will tend to diffuse from areas of higher concentration to areas of lower concentration. Thus, the concentration of hydrocarbons in the section of the carbon bed near the atmosphere port will increase with time. During diurnal loading, a mixture of air and gasoline vapor enters the canister through the tank port. Virtually all of the gasoline vapor is adsorbed by the carbon within a specific region of the bed known as the mass transfer or adsorption zone. The adsorption zone will move in the direction of vapor flow with time. On the downstream side of the adsorption zone, a small volume of clean air will move through the remainder of the bed, potentially causing bleed emissions. Diffusion can be described by Ficks Law, one form of which is: Amount of Diffusion = dn/dt = - A * D * dC/dx , Where, n = moles of a species t = time A = cross sectional area D = diffusion coefficient of a species (temperature dependent) C = concentration x = diffusion path length Although an exhaustive examination of this law is not within the scope of this paper, inspection of it with respect to evaporative emissions canisters yields several characteristics that affect bleed emissions. The first characteristic that affects bleed emissions is time. Bleed emissions are important in evaporative emissions testing because of the potential for long soak times prior to and during the real-time diurnal. Although important, most of the time elements in the procedure are fixed and cannot be changed. Another characteristic that affects bleed emissions is the concentration gradient. This can be viewed in two ways. First, the concentration gradient from one portion of the canister to another affects the diffusion rate. This is controlled primarily by the amount of purge used. Second, the hydrocarbon concentration gradient between the hydrocarbons adsorbed in the carbon pores and the hydrocarbons in the gas phase affects the diffusion rate between the phases. Although this is also controlled by the amount of purge used, the pore size distribution of the carbon used will influence the equilibrium. Thus, a well-purged canister filled with a high-capacity, low-heel carbon is desired to reduce bleed emissions. Two characteristics that can be considered together are the diffusion path length and the cross-sectional area.

Both are elements of canister design. Decreasing the cross-sectional area can be accomplished by simply narrowing the canister bed profile or effectively by placing a plate with restricted open area across the carbon bed. Both of these options will increase the pressure drop, a critical canister parameter particularly important for Onboard Refueling Vapor Recovery and should be considered carefully. Therefore, a canister with an optimized length-to-diameter ratio is desired to reduce bleed emissions. Two final characteristics that affect bleed emissions are temperature and hydrocarbon species. The diffusion coefficient is temperature-dependent and is unique for a given species. Since the temperatures of the test conditions are fixed, there is little reason to consider their effect. However, the effect of the hydrocarbon species is an important consideration. For like species, the diffusion coefficient will decrease with increasing molecular weight. For example, n-butane will have a higher diffusion coefficient than will n-pentane. Prior to the diurnal test, the canister is preconditioned by loading to either breakthrough or 1.5 times breakthrough with 50% n-butane. Butane is the lightest major component of gasoline vapor (3) and has a diffusion coefficient greater than that of the other major components of gasoline vapor. That is, butane will have a higher diffusion rate compared to gasoline vapor. Thus, it is quite important to purge as much of the butane loaded during preconditioning as possible. In this paper, we will examine the effects of soak time, purge volume, canister geometry, and carbon type on bleed emissions.

volume, typically 300 bed volumes, using dry air at a rate of 15 liters per minute. The soak time between load and purge events was no more than 5 minutes. The ambient temperature of the equipment, including the canister storage compartment, was maintained at 25C during all stages of the testing. Butane vapor cycling was also performed using automated cycle test equipment. Butane, diluted with laboratory air to a concentration of 50% by volume, was delivered to the canister at a rate of 40 g/hr. The breakthrough concentration used was 5000 ppm. After breakthrough was detected, the canister was purged with dry air, typically for 300 bed volumes, at a rate of 15 l/min. The soak time between load and purge events was no more than 5 minutes. The ambient temperature of the equipment, including the butane gas cylinder, was maintained at 25C during all stages of the testing. BWC BLEED TEST - The BWC Bleed Test involved measuring emissions from the atmosphere port of the canister using a flame ionization detector while loading the canister with butane. The goal in performing the BWC bleed test was to evaluate carbon bleed performance under controlled conditions. A 1.0-liter test canister was used for all BWC bleed testing. The properties of the test canister are described elsewhere in the experimental methods. The temperature of the entire system including canister and butane fuel source was controlled at 25C. Canister preconditioning included three butane adsorb/purge cycles with loading to 5000-ppm breakthrough and 300 bed volumes of purge according to the procedures described above. Following the three conditioning cycles, the canister was soaked for a specified length of time. After the soak, the canister was again loaded with 50% butane vapors to breakthrough and the canister emissions were measured and recorded during the course of the loading using a flame ionization detector. DIURNAL TESTS - Two separate test procedures were developed for measuring bleed emissions under diurnal temperature conditions. Both of these tests involved measuring emissions directly from the canister vent rather than from the entire vehicle as in a SHED test, or from an entire fuel system, as with a mini-SHED. The first test, referred to as the compressed diurnal test, was developed for measuring canister bleed emissions during a one-day diurnal using the 11-hour temperature ramp segment of the CARB temperature profile. A 1.0liter test canister was used in combination with a 60-liter fuel tank and 9.0 R.V.P. fuel in order to load the carbon bed to a near-breakthrough condition within an 11-hour test period. The compressed test introduced the CARB temperature profile as a variable without the time requirements of the two- and three-day diurnal test. However, the results are useful only on a relative basis,

EXPERIMENTAL METHODS
Bleed emissions were evaluated using three different test procedures. Each test featured the following common elements: an initial canister pre-conditioning, vapor load, purge followed by a time-controlled soak, and emissions measurement during a final canister vapor load simulating a diurnal loading event GASOLINE VAPOR AND BUTANE CYCLING - Gasoline vapor cycling was performed using automated cycle test equipment that precisely controlled and monitored all testing conditions. Gasoline vapors were generated by bubbling air at a rate of 200 ml/min through 2 liters of 9.0 RVP-certified test gasoline heated to 36C. Under these conditions, the vapor generation rate was approximately 40 g/hr and the hydrocarbon concentration was approximately 50% by volume. The vapors were sent to the canister until a breakthrough concentration of 5000 ppm was detected using a flame ionization detector (FID). If breakthrough was not detected after 100 minutes of vapor loading, the liquid gasoline was replaced with a fresh 2-liter aliquot. After breakthrough was detected, the canister was purged for a specified

because the canister and tank size combinations were not realistic and the test time was shortened. Also, the vapor generation rates were much higher using 9.0 RVP fuel as compared to using 7.0 RVP fuel. Pre-conditioning for the compressed diurnal test consisted of six gasoline vapor load / purge cycles with 300 bed volumes of purge and a loading rate of 40 g/hr as described previously. Within 2 hours of the final gasoline vapor purge, the canister was loaded with 50% butane vapors in air at a rate of 40 g/hr until a breakthrough of 5000 ppm was measured by a flame ionization detector. The canister was soaked for 60 minutes before the final purge. The final purge was performed at a rate of 15 liters per minute with dry air for 300 bed volumes. Following the purge, the canister was soaked at 20 C for 24 hours before starting the diurnal test.

temperature ramp from 18.3 C to 40.6 C, the canister weight and emissions were measured several times. During the portion of the diurnal when the temperature decreased, the Tedlar bag was removed in order to allow the system to back purge. The removed Tedlar bag was filled to a known volume with nitrogen. The hydrocarbon concentration was determined by evacuation of the bag contents into an FID. Once the concentration and volume were determined, the mass of hydrocarbon was calculated and recorded.

A second diurnal test procedure, referred to as the simulated real-time diurnal test, subjected a fuel tank and canister to the entire two-day or three-day CARB diurnal temperature profile. A 60-liter fuel tank was used in combination with commercial automotive canisters, typically with a nominal size of about 2 liters. The fuel used for the simulated real-time diurnal test was a 7.0 RVP Phase II fuel as specified in the CARB test procedures. This test was designed to simulate the diurnal portion of the CARB vehicle emissions test procedure in order to generate quantitative canister emissions data. Pre-conditioning for the simulated real-time diurnal test included multiple gasoline vapor load / purge cycles as described above. The loading rate for the gasoline cycles was 40 g/hr, and the most typical purge was 400 bed volumes. Following the gasoline cycles, the canister was loaded with 50% butane vapor to 5000 ppm breakthrough in preparation for a two-day diurnal test and was loaded to 1.5 times the nominal butane working capacity in preparation for a three day diurnal test. After the butane load, the canister was allowed to soak for 60 minutes before purging for a specified volume. Typical purge volumes were 400 bed volumes in preparation for the three-day diurnal test and 150 bed volumes for the two-day diurnal test. Following the purge, the canister was soaked at 20 C for 24 hours before starting the diurnal test. After the 24-hour soak, for both the compressed and real time diurnal tests, the canister was attached to a commercial 60-liter, steel fuel tank containing 24 liters of certified test fuel, previously equilibrated to 18.3 C overnight. A 9.0 RVP fuel was used for the compressed diurnal test and a 7.0 RVP fuel was used for the real-time diurnal test. A Tedlar bag was attached to the atmosphere port of the canister as shown in Figure 1 to collect the hydrocarbon emissions. During the 11-hour

Figure 1. CARB temperature profile and simulated diurnal test setup.


IN-USE CANISTERS - In some cases, in-use evaporative emission canisters were used to measure diurnal bleed emissions. Automotive Testing Laboratories Inc. was contracted for procurement of the canisters. The canisters were obtained from in-use vehicles, with odometer readings ranging from 16,000 km to 53,000 km. The nominal size of the canisters tested was about 2.0 liters. The canisters were filled with woodbased carbon from Westvaco. The carbon had a nominal specified new ASTM butane working capacity (BWC) of 15 g/dl.

TEST CANISTERS
In some cases, test canisters were filled with virgin carbon samples and evaluated for bleed emissions performance. The test canisters were 1.0 liter in size. Each of these canisters contained a partition separating the carbon into two beds. The length-to-diameter (L/D) ratio for the test canisters was 3.5 unless otherwise stated. The partition was located either in the middle of the canister creating equal vapor and atmosphere chambers or was located in the canister so that the ratio between vapor and atmosphere chambers was 2:1. Placement of the partition in the canister is referenced along with data obtained using the test canisters.

bed leading to the emission of butane that was adsorbed on the carbon during previous loading.

EFFECT OF PURGE
The simulated real-time diurnal test was used to study the effect of purge volume on bleed emissions. A twoliter, in-use canister was preconditioned as described above for the 2-day simulated diurnal. The final purge volume after the butane load to breakthrough was varied from 200 to 1200 bed volumes using a constant rate of 15 l/min. The test results are shown in Figure 3.

RESULTS AND DISCUSSION


EFFECT OF SOAK TIME The BWC Bleed Test was used to examine the effect of soak time. The results from the test are shown in Figure 2. A 1.0-liter test canister was exposed to repeated butane load/purge cycles under the conditions described above. At the end of a purge cycle, the canister was allowed to soak for 1 hour prior to the next load step. The plot shows the emissions as a function of canister weight gain. The data for the 1-hour soak is a typical breakthrough curve for a carbon canister. That is, the emissions from the canister vent are essentially nondetectable during loading. This is consistently true in the absence of long soak times after purge and is evidence that bleed emissions are not the result of canister inefficiencies.

Figure 3. Simulated real-time diurnal; effect of purge volume. The graph shows the bleed emissions as a function of purge volume after 24 hours of soak time for a 2-day simulated diurnal test. The data show the strong influence of purge on bleed emissions. At 200 bed volumes purge, the bleed emissions for the second day were 270 mg. The emissions decreased to 31.6 mg at 400 bed volumes purge. Further decreases to levels of 12.9, 6.9, and 2.1 mg were found for 600, 800, and 1200 bed volumes purge respectively. The effect of purge volume on bleed emissions is greater than its effect on working capacity. For example, increasing the purge volume from 200 to 400 bed volumes will increase the working capacity of the canister by about 20% (4). However, this same increase in purge volume will decrease bleed emissions by nearly 90%. Further, it appears that meeting canister emissions targets for LEV II can be achieved by increasing purge volume alone. Though it should be noted that increased purge volume could lead to drivability and tailpipe emissions concerns. Increased purge volume requires increased purge time and/or increased purge rate. On the vehicle, purge occurs during the drive cycle portions of the evaporative

Figure 2. BWC bleed test effect of soak time. The second set of data in Figure 2 show the emissions for the same canister when the soak time after purge was lengthened to 24 hours. In this case, the emissions from the canister during loading were immediate and somewhat consistent until breakthrough. Due to the long soak time, significant diffusion occurred within the carbon

emissions test. These drive cycles are fixed, making increased purge time difficult. Considering this, the effect of purge rate on bleed emissions was examined using a simulated 3-day diurnal test. Figure 4 gives the bleed emissions after 24 hours of soak time for an in-use canister purged for 400 and 600 bed volumes at different rates. The purge rates were adjusted so that the different final purge volumes were achieved in the same time. A rate of 15 l/min was used for 400 bed volumes and a rate of 22.5 l/min was used for 600 bed volumes. The bleed emissions after 400 bed volumes purge was 60.3 mg on the third day as compared to 15.7 mg for 600 bed volumes purge. The data indicate that increased purge volumes achieved through increased purge rates reduce bleed emissions significantly.

The first canister was rectangular with a partition completely separating two carbon beds of equal volume. The length-to-diameter (L/D) ratio of the canister was 3.5. The second canister also had a partition completely and equally separating the two carbon beds. The L/D of this canister was increased to 4.5. The third canister had the same dimensions as the first one, except that the partition was moved such that the vapor inlet carbon section had twice the volume as that of the purge inlet section. The same 15 g/dl BWC wood-based carbon was used in all of the canisters. The results of the diurnal test are shown in Figure 6. The equally-partitioned canister with an L/D of 3.5 had bleed emissions of 290 mg under the conditions used. Increasing the L/D to 4.5 was found to decrease the bleed emissions to 154 mg. Moving the partition in the L/D=3.5 canister from 50:50 to 2/3:1/3 further decreased the bleed emissions to 121 mg. Please note that bleed emissions results obtained under the compressed diurnal test conditions are somewhat higher than those found using the simulated diurnal test. Nonetheless, the data reflect the significant impact that canister geometry has on bleed emissions.

Figure 4. Simulated real-time diurnal test; effect of purge rate and volume.

EFFECT OF CANISTER GEOMETRY


The compressed diurnal test was used to study the effect of canister geometry on bleed emissions. Three one-liter canisters with different geometries were used (Figure 5). Figure 6. Compressed diurnal test effect of canister geometry. It is interesting to note that the effect of canister L/D is different for bleed emissions than it is for working capacity. Previous test results showed, that as the canister L/D is increased from 1.0 to 3.5, the gasoline working capacity (40 g/hr loading rate) is increased by nearly 40% for 2 mm carbon pellets (5). However, increased canister L/Ds above 3.5 had much less of an impact. The effect appears to be loading rate dependent. Under ORVR conditions (50 g/min loading rate), an increase in L/D from 1.0 to 3.5 resulted in only a 10% gain in working capacity (6). Based on these data, it was assumed that an increase in canister L/D from 1.0 to 3.5 would impact bleed emissions as it did working capacity. Therefore, an investigation of the effect of geometry for L/Ds greater than 3.5 was performed.

Figure 5. 1.0-liter test canisters.

Additional optimization of canister geometry may yield further reductions in bleed emissions.

EFFECT OF CARBON TYPE


The compressed diurnal test was used to study the effect of carbon type on bleed emissions. Three different carbons, all with ASTM Butane Working Capacities in the 11.0 12.0 g/dl range, were used. One liter of each carbon type was subjected to 6 adsorb/purge cycles using gasoline vapors, followed by preconditioning with butane as previously described. The heel was determined after the butane preconditioning. The results are shown in Figure 7.

Also under consideration was the influence of aging and heel composition on bleed emissions. A previous study showed that continued laboratory load/purge cycles with gasoline vapors alter the composition of the heel (2). Specifically, it was shown that the average molecular weight of the hydrocarbon heel increased with increased aging. As discussed earlier, higher molecular weight hydrocarbons tend to have lower diffusion coefficients as compared to lower molecular weight hydrocarbons. Thus, carbon canisters exposed to significant gasoline vapor aging should have higher hydrocarbon heels and lower bleed emissions than canisters exposed to less aging. Bleed emissions and hydrocarbon heels were determined for samples of wood-based carbons with BWCs of 11 and 15 g/dl using the compressed diurnal test. The results were compared after 6 and 200 gasoline vapor aging cycles and are displayed in Figure 8. After 6 cycles, the heel and bleed emissions for the 15-g/dl BWC carbon were 85.1 g/l and 154 mg. These values are higher than the 48-g/l heel and 60.8 mg of bleed emissions found for the 11-g/dl BWC carbon. However, after 200 aging cycles, the heel and bleed emissions found for the 15 g/dl BWC carbon were 121 g/l and 34 mg. The heel and bleed emissions for the 11-g/dl BWC carbon were 60 g/l and 41 mg after aging. As expected, the heel levels for both carbons increased while the bleed emissions decreased with aging. Also, the bleed emissions after aging were similar for both carbons.

The three carbon types differed in the raw material used and in the activation conditions. This resulted in different pore size distributions. It has been shown that carbon pore size distribution directly affects both working capacity and hydrocarbon heel development (1,4).. The wood-based carbon had a hydrocarbon heel of 58 g/l after cycling, significantly lower than the heels for the coal-based and olive-based carbons. The wood-based carbon also had the lowest measured bleed emissions, at 121 mg as compared to 179 and 299 mg for the coalbased and olive-based carbons respectively. The data suggests that there is a relationship between the hydrocarbon heel and bleed emissions.

Figure 7. Compressed diurnal test effect of carbon type. It is often necessary to utilize an activated carbon with very high working capacity. Currently, the highest working capacity carbon available commercially is a 15 g/dl BWC wood-based carbon. High working capacity is accomplished by increasing the pore volume in the critical pore size range (1). However, this can also result in an increase in pore volume of pore sizes associated with heel. For this reason, the effect of increased working capacity and heel on bleed emissions was studied.

Figure 8. Compressed diurnal test effect of aging. Although hydrocarbon heel increases with aging, working capacity is not significantly affected for wood based carbons. In one study, multiple in-use canisters were removed from vehicles and tested for GWC by performing 6 load/purge cycles as described above. The data obtained with the in-use canisters were compared to new canisters exposed to 25 gasoline load/purge cycles. The data is shown in Figure 9. The data indicate that,

for the wood-based carbons tested, the working capacity is consistent and independent of vehicle use. The data also indicate increased heel weights are due to heel composition changes rather than hydrocarbon occupation of additional pore volume.

Under the 2-day test procedure, the average maximum bleed emissions for the in-use canisters were 196 mg. The average maximum bleed emissions were reduced to 15.2 mg with the simple addition of the auxiliary chamber. Under the 3-day test procedure, the average maximum bleed emissions for the in-use canisters were 148.5 mg. The average maximum bleed emissions were reduced to 16.6 mg with the addition of the auxiliary chamber.

Figure 9. In-use canister - gasoline working capacity.

EFFECT OF AUXILIARY CHAMBER


The simulated real-time diurnal test was used to study the effect of an additional auxiliary chamber containing activated carbon on bleed emissions. In order to approximate real-world conditions, in-use canisters were used in the study. Fifty cubic centimeter auxiliary chambers were attached to the in-use canisters as displayed in Figure 10. Figure 11. Simulated real-time diurnal; effect of auxiliary chamber. The effect of the auxiliary chamber geometry was studied using the simulated real-time 2-day diurnal test. An inuse canister was equipped with auxiliary chambers of the same volume (50 cc) but having different L/Ds. The results are shown in Figure 12. The average maximum bleed emissions for the canister with no auxiliary chamber were 196 mg. When the canister was equipped with an auxiliary canister with an L/D=1, the maximum bleed emissions were 135 mg, significantly higher than the 15.2 mg of bleed emissions found for the canister equipped with the L/D=4 auxiliary chamber.

Figure 10. Auxiliary chamber concept. Bleed emissions were measured using both 2- and 3-day procedures. Multiple runs were made with each condition, and the average results are shown in Figure 11. The final purge volume used was 150 bed volumes for the 2-day test and 400 bed volumes for the 3-day test.

Figure 12. Simulated real-time diurnal test; effect of auxiliary chamber geometry. The data indicate that the addition of an auxiliary chamber to the vent side of the primary canister substantially reduces bleed emissions. The simple addition of an auxiliary chamber may help achieve canister emissions targets for LEV II applications without a significant increase in purge. The data also suggest that the design of the auxiliary chamber will need to be optimized for best performance. One aspect of auxiliary canister design is the form of activated carbon used. Carbon pellets, such as those used in automotive canisters, can be used to lower bleed emissions. However, their use in auxiliary chambers may also introduce a significant amount of flow restriction, which may be unsuitable for ORVR applications. As an alternative, high carbon-content monoliths, or honeycombs, have been developed by Westvaco for use in auxiliary chambers. The honeycombs have very low flow restriction and have been demonstrated to significantly reduce bleed emissions. In fact, the data presented in Figures 11 and 12 were developed using carbon honeycombs.

ACKNOWLEDGMENT
The authors wish to express their sincere appreciation to Joe W. Snead, who aided in the design, operation and maintenance of the equipment used in the study. Jerry Rucker provided assistance performing laboratory testing when needed. Dr. H. Ray Johnson provided support and encouragement. Catherine K. Morgan provided great help in the coordination of the manuscript with SAE. Erin Higinbotham of Automotive Testing Laboratories was very helpful in providing in-use canisters.

REFERENCES
1. H. R. Johnson and R. S. Williams, Performance of Activated Carbon in Evaporative Loss Control Systems, SAE Technical Paper 902119, October 1990. 2. H. Itakura, N. Kato, T. Kohama, Y. Hyoudou, and T. Murai, Studies on Carbon Canisters to Satisfy LEV II EVAP Regulations, SAE Technical Paper 2000-010895. 3. R. L. Furey and K. L. Perry, Composition and Reactivity of Fuel Vapor Emissions from GasolineOxygenate Blends, SAE Technical Paper 912429, October 1991. 4. P. J. Johnson, D. J. Setsuda, and R. S. Williams, Activated Carbons for Automotive Applications, from Carbon Materials for Advanced Technologies, Chap. XII, Ed. by T. D. Burchell, Pergamon Press, Oxford U.K., 1999. 5. R. S. Williams, Westvaco Evaporative Emissions Seminar, 1991, copies of data available upon request. 6. R. S. Williams and C. Reid Clontz, Westvaco Evaporative Emissions Seminar, 1998, copies of data available upon request.

CONCLUSION
LEV II regulations necessitate significant decreases in hydrocarbon emissions from evaporative emission canisters. The emissions from canisters are not due to a lack of adsorptive capacity, but are due to bleed emissions. Bleed emissions can be reduced by

Increasing the purge volume Optimizing the canister geometry Using high-capacity/low-heel carbons Utilizing an auxiliary chamber Optimizing the auxiliary chamber wood-based

You might also like