Hydrocarbon Traps for the Air Intake System: Component Test Rig and SHED Test Procedure for Determining Their Efficiencies

Abstract

Hydrocarbon traps in the air intake system (AIS) are a common method for controlling evaporative emissions from the air intake path. Several different systems are available, but there is no standard method for determining their efficiencies. Therefore, a component test rig for hydrocarbon traps was developed. Some optimizations were necessary to achieve emission characteristics observed in engine measurements. Using this setup, several measurements were performed on four different hydrocarbon traps. The results were in reasonable agreement with those from engine measurements. Two different hydrocarbon (HC) traps were selected for further studies. In these studies, the repeatability and the dependency of the emission mass level were investigated. Furthermore, the hydrocarbon concentration in the air filter box was determined using point source flame ionization detector (FID) sampling and a metal oxide semiconductor (MOS) sensor. The data showed a correlation with the emission mass determined in a sealed housing emission determination (SHED) test.

1. Introduction

Evaporative emissions are a significant source of volatile organic compounds (VOCs) in urban areas [1,2,3]. Several VOCs are harmful to health and act as precursors for ozone formation [4,5,6]. Many countries have established regulatory limits for evaporative emissions that automobile manufacturers must comply with. Various legislations, such as China 6, U.S. Tier 3, CARB LEV III, and EU6, outline specific test procedures and limit values [1,7]. The basic principle of these test procedures is similar: A car is placed in a sealed housing for emission determination (SHED) and emitted hydrocarbons are measured using a flame ionization detector (FID). The final result is the emission mass in milligrams, calculated using SHED and FID data according to equations provided by test procedures (e.g., [8,9]).

Vehicles have multiple sources of evaporative emissions. Nonfuel evaporative emissions originate from plastic parts and rubber materials, while fuel-related emissions stem from fuel tanks, fuel lines, and the internal combustion engine (ICE).

For engine evaporative emissions, the air intake system (AIS) is one critical pathway. Chinnasamy et al. [10] provided a comprehensive overview of potential sources, contributing to this pathway, including the fuel injection system (direct injection (DI), manifold injection, or even both systems in modern engines), the crankcase ventilation system, the purge air system for the fuel tank, and sometimes the exhaust gas recirculation system. The location and the connection of these systems to the AIS are crucial. For example, hydrocarbons emitted from a DI injector have a longer distance to the air inlet of the AIS, whereas the connection of the crankcase ventilation may be closer to the air inlet.

If hydrocarbons remain within the AIS, there is no issue. Only those escaping the system increase hydrocarbon concentrations in the surrounding environment or during a SHED test. Various effects like pressure gradients, thermal expansion, or diffusion [10] can drive hydrocarbons out of the air inlet.

To avoid or at least to reduce this contribution, HC traps in the AIS have been developed. Early systems were so-called flow-through or cross-flow models containing zeolite or activated carbon as an adsorption medium. Later, bypass traps became preferred, due to their lower airflow restriction. Moyer et al. [11] summarized the evolution of HC traps, comparing different types based on weight, flow restrictions, durability, costs, and media types (e.g., activated carbon and zeolite).

However, limited information is currently available on the efficiency of these HC traps. Reports suggest that flow-through traps are most efficient for the adsorption and desorption of hydrocarbons, whereas bypass traps are less effective [11]. Furthermore, it is mentioned that all presented types of HC traps were effective in complying with evaporation emission regulations. But no data or assessments indicate the range of their efficiency. One reason for this lack of information could be the absence of a standardized test method.

Schaffer et al. [12] reported in 2007 that each HC trap supplier and automobile manufacturer uses different test methods, highlighting the need for a uniform test. Yet, as of August 2024, no such test exists. Schaffer’s article describes a test procedure using a part of the AIS and a known amount of n-butane as a test medium, placed in a SHED for measurement. While useful for comparative studies, no numerical values are presented.

Maeda K. et al. [13] described a similar test procedure, comparing three types of HC traps (two flow-through and one bypass). The HC emission rates ranged from about 10% to 40%, corresponding to efficiencies between about 90% and 60%. Unfortunately, the article lacks details about the test procedure, especially the kind of “HC vapor” used, and does not address the precision and reproducibility of the test.

Chinnasamy et al. [10] described a test for hydrocarbon adsorbers using butane and employing a low flow of hydrocarbons instead of diffusion to shorten the test duration. This test was suitable for comparing cross-flow adsorbers, but a fair evaluation of bypass adsorbers was not possible. Again no detailed numerical results are presented.

In summary, there are promising ideas for an appropriate test setup, but due to missing numerical results, it is difficult to assess how well they are working. Reliable data require determining the precision and reproducibility of a test procedure, which are essential for evaluating individual results. Furthermore, our procedure aims to use fuel as the hydrocarbon source, although butane is an alternative easier to handle. However, butane is not representative of the wide variety of hydrocarbons in gasoline. The latter typically contains hydrocarbons with boiling points ranging from about 40 °C (initial boiling point) to about 200 °C (final boiling point), whereas the boiling point of butane is −0.5 °C. Gasoline also contains olefins, aromatics, and often ethanol, which significantly changes the interaction between gasoline vapor or butane and the adsorbent material in an HC trap [14].

Our goal is to develop a component test rig and a SHED test procedure to determine the efficiency of different HC traps—both bypass and cross-flow types—using gasoline as the hydrocarbon source and assess the reliability of the results. Initially, four different HC traps were measured using the new test setup, and these results were compared to the results of the same parts measured on an engine. Subsequently, two HC traps were selected to investigate the reproducibility of the test procedure. Basic parameters were varied, and several repetitions were measured with identical setups.

Additionally, we aimed to determine the hydrocarbon concentration inside the air filter box near the adsorption material, using a metal oxide semiconductor (MOS) sensor and point-source FID sampling. The resulting data were checked for a possible correlation with the SHED test results.

2. Materials and Methods

2.1. Component Test Rig

All parts of an AIS, from the air inlet to the turbocharger connection, were selected for the component test rig. Additional ports, such as those for crankcase ventilation and fuel tank purge air, were sealed with appropriate plugs. Instead of the turbocharger, a connector was fabricated to provide a vacuum flange (DN 50 ISO-KF) on the other side. This allowed the use of standard vacuum parts and FKM seals to create an extended air path. At the end of this path, a fuel reservoir was installed and connected via an orifice. The orifice is used to control the emission characteristics of the fuel reservoir. It consisted of two blank flanges, each containing a bulkhead fitting, connected by a stainless steel tube. By varying the diameter and length of this tube, the emission characteristics could be adjusted. Figure 1 shows an image of the orifice connected to the AIS and the fuel reservoir.

The entire assembly was installed in a frame made of aluminum profiles, providing stability and ease of handling. Figure 2 shows the complete AIS component test rig.

2.2. SHED Test Procedure

The SHED test procedure must meet two main requirements. First and foremost, its results must be comparable to the results of our engine test procedure. Additionally, time and cost efficiency are key considerations.

Engine SHED measurements have been conducted at our institute for over 15 years. No official test procedures are available for engines. Different legislation (e.g., CARB, EU6) only set limits and test procedures for whole vehicles. Nevertheless, the results for an engine are useful to check for any issues regarding its evaporative emissions. Over the years, a specific workflow has been established at our institute, which is based on CARB test protocols [8]. Furthermore, it considers the observation that in most cases the maximum emission mass of an engine in a SHED test (diurnal soak) is obtained shortly after reaching the maximum temperature. Afterward, the emission mass typically remains quite constant and only rises again with the temperature increase during the second day of the test. Usually, the emission mass of the first day is higher than that of the second day. For an optional third day, the behavior is similar. (This is not valid for engines with certain problems, e.g., some kind of leakage!) Figure 3 shows the result of a three-day Diurnal Soak SHED test of an engine. The conclusion of this behavior was that a kind of worst-case result for the diurnal soak emission mass is obtained after about 14 to 16 h of testing. Therefore, our typical measurement procedure for engines starts with conditioning the engine at a test dynamometer (FTP75, about 45 min). Then we perform a Hot Soak (1 h) test. After a soak period of 6 h, a Diurnal Soak test is started. It is stopped after 16 h because typically the worst-case value has already been reached (as explained above). Thus, the complete test procedure can be carried out in one day. Consequently, one result can be obtained each day, improving our time and cost efficiency.

Tests of different HC traps on engines further showed that the influence of an HC trap on Hot Soak test results is negligible in most cases. In contrast, their effect on Diurnal Soak test results was significant for certain engines. Therefore, for our component test rig, only a SHED test using the Diurnal Soak temperature profile is meaningful. The Hot Soak test can be neglected. To obtain results comparable to our engine SHED tests, the test period is similarly limited to 16 h. Considering that a short thermal desorption cycle of the SHED takes about 4 h, it is possible to carry out one test per working day.

The complete workflow for the test procedure was as follows: Each measurement began with purging the AIS component test rig. A standard industry vacuum cleaner was connected to the AIS setup instead of the fuel reservoir. The purpose of this step was to achieve a purging effect, which normally occurs by driving the vehicle or running the engine. The air volume flow rate was about 1.5 m³/min. In parallel, the SHED was thermally desorbed (heating to about 60 °C, cooling down to 18 °C while flushing with clean air, time requirement of about 4 h). As soon as the temperature of 18 °C was reached, the component test rig was placed in the SHED. After a period of at least 30 min, a specific amount of fuel (CARB LEV III test fuel) was injected into the fuel reservoir using a gas-tight syringe. The reservoir was mounted on the AIS assembly using an orifice. The next step was to start the CARB Diurnal Soak temperature profile [8]. The resulting emission mass [mg] of a test was calculated using the recorded data and the equation provided by CARB [8].

2.3. Measurement Devices

The concentration of hydrocarbons was measured using a flame ionization detector (FID) (Testa 123). It had a low sampling flow rate of approximately 12 mL/min. The FID was connected to a multiplexer system, which also provided temperature and barometric pressure measurements. This system was originally developed for component measurements in a micro SHED system [15].

The FID/multiplexer system allowed for measuring the HC concentration within the SHED system and, additionally, within the clean air section of the air filter box. The connections between the sampling sites (SHED and air filter box) were established using Teflon lines with an inner diameter of 1 mm. To minimize the impact of sampling, it was performed only once per hour, with each sampling interval set to 150 s. This duration was necessary to ensure sufficient flushing of the sampling lines. The recorded HC concentration was obtained by calculating the arithmetic mean of the last 20 s of each sampling interval. Each sampling began with the SHED and subsequently switched to the air filter box.

Some measurements were accompanied by an MOS sensor (Figaro TGS 822). The resistance of the MOS sensor (RMOS) correlates with the HC concentration. Generally, the calibration stability of such devices is quite poor, making them unsuitable for determining absolute values. Ten calibrations (#1–#10) were performed, showing the behavior depicted in Figure 4. The FID system was used to get reference values for the HC concentration (cFID). A double logarithmic plot showed a fairly linear correlation. The calibration experiments were carried out approximately six months before the MOS sensor was deployed in the AIS component test rig.

However, MOS sensors are very well suited for monitoring changes since data are recorded continuously. To overcome the lack of calibration stability, the FID data measured in parallel could be used in most cases. Assuming the FID data to be more reliable, it was possible to calculate correction factors for the hourly FID values and their corresponding MOS sensor values. The mean of all correction factors for one measurement was used for the entire data set of MOS sensor values. Typically, these corrected values agreed well with the hourly FID data. An example is shown in Figure 5.

The HC concentrations calculated using the calibration data for the MOS sensor differed significantly from the hourly FID values. The use of the correction factor significantly improved the fit (MOS corr). The corrected values of the MOS sensor provided a good indication of the changes in HC concentration in the air filter box.

3. Results and Discussion

3.1. Background Check Component Test Rig

The first step to verify the basic applicability of the component test rig was to demonstrate that the entire assembly, particularly, the vacuum flanges and seals, contributed only insignificantly to the resulting emission mass in a SHED test. A background measurement of the empty SHED was compared to measurements of the entire test rig and of the entire test rig including a volume of 0.2 mL LEV III test fuel. In the latter experiment, it was necessary to avoid emissions emitted from the air filter box opening (air inlet of AIS). Therefore, the air filter box was replaced by an aluminum adapter connected to a stainless-steel corrugated pipe. The other side of this pipe was connected to an outlet of the SHED system, thus preventing emissions from this pathway. Consequently, only emissions from potential leakages or permeation of the AIS assembly could affect the SHED test result. Figure 6 shows the results of these three measurements.

The background of the SHED showed almost no emission mass. The background test using the test rig without fuel (AIS background) resulted in an emission mass of about 0.4 mg. The measurement using a fuel volume of 0.2 mL injected into the fuel reservoir and a corrugated pipe to prevent emissions from the air inlet yielded a value comparable to the background of the test rig without fuel. The background of the test rig itself was quite low, and injecting fuel into the reservoir had no detectable effect. Thus, the contribution of the test rig to the resulting emission mass in a SHED test is negligible.

3.2. Emission Level and Gradient for SHED Test Procedure

To achieve an appropriate emission mass level and gradient for the SHED test procedure, data from engine SHED measurements were used as a starting point.

In a previous study, different HC traps were tested on one engine. To obtain a constant emission level from the AIS path, an engine with very low emissions originating from this source was selected. Before each SHED measurement, a specific amount of LEV III test fuel was injected into its intake manifold. This procedure resulted in an increased total emission mass, due to a higher contribution from the AIS pathway. Several data sets with different setups were available for this engine. Consequently, it was possible to calculate the emission mass emitted from the AIS path by subtracting the emission mass of the engine with prevented AIS emissions (AIS connected to SHED outlet) from the results including the AIS pathway. Figure 7 shows the calculated emission mass of the AIS path for this engine. Due to our typical measurement procedure for engines (FTP75—1 h Hot Soak—6 h Soak—16 h Diurnal Soak), data are only available for a period of 16 h.

The calculated data for the emission mass from the AIS showed an almost linear increase.

The next step was to find suitable parameters for the AIS component test rig to achieve similar qualitative (gradient of emission mass) and quantitative results (emission mass level) in the SHED test procedure. Different options for fuel reservoir (reservoir volume, amount of injected fuel, position of the reservoir) and different orifices were tested using varying stainless steel tubes (inner diameter: 3 mm, 6 mm, 8 mm, 12 mm diameter; length between 2.5 cm and 10 cm). The length of the elongated AIS path, which consisted of vacuum parts, was also varied.

Figure 8 shows the results of four different configurations. Configuration test rig AIS #4 achieved very good agreement with the data from the engine measurements. This configuration became the default setting, and further variations in the emission mass level were made only by changing the injected fuel volume.

Table 1. Parameters for four different test configurations.

Configuration	Fuel Volume [µL]	Orifice Diameter [nm]	Orifice Length [cm]
test rig AIS #1	150	3	2.0
test rig AIS #2	200	12	2.5
test rig AIS #3	200	12	7.0
test rig AIS #4	200	8	2.5

summarizes the parameters of the different configurations.

3.3. Influence of MOS Sensor and Air Filter Box FID Sampling

As mentioned earlier, an MOS sensor and/or FID sampling were used in the air filter box during several measurements to assess the HC concentration at this location. Figure 9 shows the upper part of the air filter box modified to include an MOS sensor and an FID sampling point. Both were mounted at the center of the upper part of the filter box. Bulkhead fittings were used for the cable and piping lead-throughs to prevent any leakage caused by these modifications.

It is important to consider that this modification might have affected the resulting SHED values. To check this assumption, the same setup and test rig were measured with the MOS sensor and FID sampling, with the MOS sensor but without FID sampling and without both. The resulting emission mass values for the SHED tests are summarized in

Table 2. SHED test results for the same setup, only varying the additional use of the MOS sensor and FID sampling within the upper part of the air filter box (“X” indicates that this item was used).

Test Number	MOS Sensor	FID Sampling	Emission Mass [mg]
#1	X	X	46.6
#2	X	X	49.0
#3	X	X	48.3
#4	X	X	46.9
#5	X	X	47.6
#6	X	-	47.1
#7	X	-	46.8
#8	X	-	47.0
#9	-	-	44.7
#10	-	-	47.9

.

The resulting emission masses within the SHED showed no significant changes, regardless of whether the MOS sensor or FID sampling was used. Therefore, it was assumed that there was no significant influence caused by the MOS sensor and FID sampling.

3.4. First Measurements of HC Traps, Comparison to Engine Measurements

Three different types of HC traps were investigated. Sample A was a cross-flow type HC trap, a non-woven fabric containing activated carbon. Sample B was a bypass-type, teabag-style HC trap. Samples C and D were also bypass-type traps, both containing carbon paper attached directly to the surface of the upper part of the air filter box. Sample C had a larger amount (or surface area) of carbon paper, as it covers an additional area compared to Sample D. Figure 10 shows photos of all four samples.

Table 3. Results of SHED tests of four different HC traps and without HC trap.

Test Number	Emission Mass [mg]
	Sample A	Sample B	Sample C	Sample D	No HC Trap
#1	3.8	17.9	14.5	14.9	27.8
#2	4.7	18.9	14.5	16.3	28.2
#3	4.6	18.6	15.6	16.1	30.1
#4	4.0	-	-	-	31.2
#5	4.6	-	-	-	30.5
Mean	4.3	18.5	14.9	15.8	29.6

presents the results for these four samples and for the measurements without an HC trap.

As expected, the cross-flow HC trap (Sample A) resulted in quite low emission masses, with a mean value of 4.3 mg. Sample B, the teabag-style HC trap, resulted in considerably higher emission masses, with a mean value of 18.5 mg. Samples C and D provided mean emission masses of 14.9 mg and 15.8 mg, respectively. The difference between these two samples correlated with their amount (or surface area) of carbon paper.

For all four samples, the efficiencies were calculated using Equation (1):

$$\mathrm{efficiency} =1-{\displaystyle \frac{{\mathrm{m}}_{\mathrm{noTrap}}-{\mathrm{m}}_{\mathrm{SampleX}}}{{\mathrm{m}}_{\mathrm{noTrap}}}}$$

(1)

where m_noTrap is the mean value for the emission mass without HC trap, and m_sampleX is the mean value for each sample (A to D).

Table 4. Efficiencies for HC traps A, B, C, and D determined using the AIS component test rig and engine measurements.

HC Trap	Efficiency
	AIS Component Test Rig	Engine
Sample A	85%	92%
Sample B	38%	41%
Sample C	47%	52%
Sample D	50%	52%

shows the results. Similar measurements were performed with all four samples using an engine (4 cylinder, 1998 cm³). The efficiencies were calculated according to the same principle.

Considering relative measurement uncertainties of about 5% to 10%, the agreement between the efficiencies determined by the AIS component test rig and those determined using the engine is quite good. This applies to both kinds of HC traps, cross-flow and bypass.

In conclusion, the AIS component test rig provides reasonable results for determining the efficiency of HC traps and is suitable for investigating different types of traps: cross-flow and bypass.

3.5. Further Measurement of HC Traps, Repeatability, and Varying Emission Mass

Two other HC traps, samples E1 and E2, were selected for further measurements. Both belong to a different AIS, so the component test rig was assembled with these parts. For confidentiality reasons, no pictures can be shown. In principle, the setup is comparable to the one shown in Figure 2. E1 and E2 were bypass-type HC traps containing carbon paper. The only difference between the two samples was the surface area of the carbon paper. Sample E1 has the entire lid surface covered (similar to sample C, but in one piece). Sample E2 was identical but included an additional sheet of carbon paper on one side of the upper part of the air filter box. Thus, the surface area of the carbon paper was about 122% compared to sample E1. The carbon papers were directly attached to the lid surface by ultrasonic welding at several points.

For the first measurements, 200 µL LEV III test fuel was used again. A total of 17 measurements were performed without an HC trap (see

Table 5. SHED test results for 17 experiments using identical parameters and no HC trap.

Number	Fuel Volume [µL]	Emission Mass [mg]
#1	200	31.3
#2	200	31.9
#3	200	31.8
#4	200	28.1
#5	200	31.0
#6	200	30.5
#7	200	32.1
#8	200	31.9
#9	200	33.1
#10	200	32.0
#11	200	32.6
#12	200	28.9
#13	200	30.1
#14	200	28.5
#15	200	26.6
#16	200	28.9
#17	200	27.7
Mean	-	30.2
Std. dev.	-	2.2

). Experiments #1 to #11 took place within one month, while experiments #12 to #17 were conducted about one month later over two weeks. The mean values and standard deviations were 31.5 mg ± 1.3 mg and 28.5 mg ± 1.2 mg, respectively. The overall mean value was 30.2 mg with a standard deviation of 2.2 mg. These data indicate that repeatability is significantly better within shorter time periods.

Measurements of the HC traps E1 and E2 were performed in between experiments #11 and #12 using the same setup. Results are summarized in

Table 6. SHED test results for experiments using HC trap E1 and E2.

Number	Fuel Volume [µL]	Emission Mass [mg]
		HC Trap E1	HC Trap E2
#1	200	11.1	6.1
#2	200	10.9	6.3
#3	200	10.0	6.5
#4	200	9.6	6.8
#5	200	9.7	-
#6	200	10.2	-
Mean	-	10.3	6.4
Std. dev.	-	0.6	0.3

.

The efficiencies of E1 and E2 were calculated using their mean emission mass and the mean emission mass of all 17 measurements without an HC trap, according to Equation (1). The results were 66% and 79% for E1 and E2, respectively. Both HC traps were also tested on the engine, yielding results of 61% (E1) and 69% (E2). While the values for E1 agreed quite well, the deviation for E2 was larger. It is also interesting to note that the efficiencies determined with the AIS test rig for these two HC traps were higher than those determined on the engine. For samples A to D, it was vice versa.

For the next experiments, the emission mass level was varied to observe any influence on the determined efficiencies. As mentioned previously, the test rig setup was kept unchanged, but the volume of LEV III test fuel was adjusted. A lower emission mass level was achieved by reducing the volume to 100 µL, and a higher emission mass level by increasing the volume to 300 µL. The emission mass decreased to about 17 mg for the lower fuel volume and increased to about 40 mg for the higher fuel volume. According to our knowledge, this is a reasonable range for modern engines.

Table 7. SHED test results for experiments with no HC trap and HC trap E1 and E2 using a test fuel volume of 100 µL.

Number	Emission Mass [mg]
	No HC Trap	HC Trap E1	HC Trap E2
#1	17.5	5.4	3.8
#2	16.7	6.0	4.3
#3	17.1	4.7	4.1
#4	14.9	5.6	-
#5	16.9	5.4	-
#6	16.0	-	-
#7	16.6	-	-
#8	16.8	-	-
#9	17.3	-	-
#10	15.9	-	-
Mean	16.6	5.4	4.1
Std. dev.	0.8	0.5	0.3

summarizes the results of the measurements for the volume of 100 µL LEV III test fuel, and

Table 8. SHED test results for experiments with no HC trap and HC trap E1 and E2 using a test fuel volume of 300 µL.

Number	Emission Mass [mg]
	No HC Trap	HC Trap E1	HC Trap E2
#1	39.7	11.2	9.0
#2	38.9	11.3	10.2
#3	39.3	11.9	10.6
#4	37.3	13.1	12.3
#5	40.2	12.2	9.0
#6	40.4	11.8	10.1
#7	38.8	13.6	10.1
#8	40.4	-	11.8
#9	41.3	-	-
Mean	39.6	12.2	10.4
Std. dev.	1.2	0.9	1.2

shows the results for a fuel volume of 300 µL.

Again, the efficiencies for HC traps E1 and E2 were calculated using Equation (1). For a fuel volume of 100 µL, values of 67% and 76% were obtained. For a volume of 300 µL, the efficiencies were 69% for E1 and 74% for E2. A more detailed discussion of these results is given after checking repeatability.

Most of the measurements shown in this section were repeated several times to evaluate repeatability and assess the uncertainty of the results. The highest number of repetitions was performed for the measurement with a volume of 200 µL fuel and no HC trap. As mentioned previously, there may be a small difference between measurements number #1 to #11 (first time period) and #12 to #17 (second time period), but for further consideration, only the mean and standard deviation of all values, 30.2 mg and 2.2 mg, were used.

Table 9. Mean values, standard deviations, and the resulting relative standard deviations for different scenarios.

HC Trap	Fuel Volume [µL]	Number of Measurements	Mean Emission Mass [mg]	Std. Dev. Emission Mass [mg]	Relative Std. Dev.
No trap	200	17	30.2	2.2	7%
No trap	100	10	16.6	0.9	6%
No trap	300	9	39.6	1.2	3%
E2	300	8	10.4	1.2	11%
E1	300	7	12.2	0.9	7%
E1	200	6	10.3	0.6	6%
E1	100	5	5.4	0.5	9%
E2	200	4	6.4	0.3	5%
E2	100	3	4.1	0.3	6%

shows all mean values and standard deviations of the different scenarios sorted by their total number of measurements. The relative standard deviations are also presented in this table, ranging from 3% to 11%, with most values at 6% or 7%.

Based on these data, the uncertainty of this type of measurement was assessed to be about 7%.

The standard deviations for the different scenarios presented in Table 9 were further used to calculate the uncertainties for the determined efficiencies of the HC traps. The diagram in Figure 11 shows the results.

Obviously, the uncertainties in relation to the differences in efficiencies for the three scenarios were quite large. For HC trap E1, the efficiencies were 67% ± 7%, 66% ± 6%, and 69% ± 6% for 100 µL, 200 µL, and 300 µL of test fuel, respectively. Considering the uncertainties, it was not possible to identify any differences between the efficiencies for the three scenarios.

For HC trap E2, the result was comparable. The efficiencies were 76% ± 6%, 79% ± 7%, and 74% ± 9% (100 µL, 200 µL, and 300 µL). Again, it was not possible to determine differences in the efficiencies for varying test fuel volumes.

Altogether, the efficiency of at least these two HC traps seemed to be independent of the emission mass level. However, it should be noted that the emission mass varied in a range from about 17 mg (100 µL test fuel) to about 40 mg (300 µL). For higher values, it might be different.

Results also indicate that comparing HC traps with quite similar efficiencies is challenging. On the one hand, all three scenarios resulted in lower efficiencies for HC trap E1. This correlated with the area of carbon paper. HC trap E2 contained a surface area of carbon paper of about 122% compared to E1. Therefore, higher efficiencies were expected for E2. On the other hand, the uncertainties of the data were so high that this difference could not be clearly revealed.

3.6. HC Measurements within the Air Filter Box

In many experiments, the HC concentration was measured not only in the SHED system but also in the clean air section of the air filter box. This was performed by MOS sensor and/or point source FID sampling. As mentioned in the method section, only FID data provided meaningful absolute values. MOS sensor data were solely for verifying qualitative changes in the HC concentration.

Figure 12 shows all measurements with a 200 µL test fuel volume. As expected, experiments without an HC trap exhibited the highest HC concentrations, typically above 60 ppm, with a maximum of around 120 ppm. Only at the beginning of an experiment, the HC concentration was lower. The results for measurements with HC traps E1 and E2 were significantly lower. The maximum value for E1 was about 20 ppm, and for E2, it was about 15 ppm. Values for experiments without an HC trap were quite unsteady. In contrast, the data with HC traps showed a more uniform gradient.

Figure 13 presents the recorded data for the MOS sensor measurements. As mentioned in the method section, the quantitative information of the basic MOS sensor data is poor. Thus, only a qualitative comparison with the FID values shown in Figure 12 was meaningful. In principle, the agreement between FID and MOS sensor data was quite good. Only the maximum values for measurements without an HC trap at approximately 15 h seemed to be significantly increased for the FID data. Furthermore, some measurements without an HC trap (#6, #7, #8) were conducted using the MOS sensor but without FID sampling. The trends of these measurements did not differ considerably from those with additional FID sampling. Thus, FID sampling seemed to have no detectable effect on the measured HC concentrations.

The data for experiments with 100 µL and 300 µL test fuel volumes were qualitatively identical, differing only in absolute values.

Moreover, a correlation was identified between the HC concentrations determined in the air filter box by FID sampling and the resulting emission mass within the SHED. The arithmetic mean of all HC concentrations measured in the air filter box (time period 0–16 h) was calculated for each experiment. Figure 14 shows these values plotted against the corresponding SHED test result (emission mass [mg]). Overall, it appears to be a linear correlation. However, most data of the experiments shown in Figure 14 were without an HC trap, especially for emission masses above 20 mg. Data from measurements with HC trap were limited to HC traps E1 and E2 and a corresponding emission mass level below 20 mg. Considering this, a linear correlation was evident for experiments without an HC trap. For experiments with HC traps E1 and E2, the correlation was not as clear, but the trend appeared to be analogous.

This correlation might be useful for SHED tests of entire vehicles. Additional point source measurement within the air intake system could indicate whether there is an elevated contribution of the AIS pathway. This information may help to localize specific failures (e.g., leaky injector(s), problems with crankcase ventilation) in case of an increased result for the entire vehicle.

4. Conclusions

Our primary objective was to develop a component test rig and a SHED test procedure to determine the efficiency of HC traps in the AIS of an engine. A review of the literature revealed only a few similar approaches [10,12,13]. Unfortunately, the descriptions of these systems are limited, allowing us to adopt only some basic ideas. In particular, details of the test procedures and results were either missing or extremely limited.

A component test rig was constructed using all relevant parts of an AIS (from the air inlet to the turbocharger connection) and vacuum parts (DN 50 ISO-KF) for elongating the pathway. Initially, basic tests, such as checking the leak tightness and the background emissions of the component test rig, were conducted to check the basic useability of this approach. The results showed no obvious problems.

The next step was to achieve an emission behavior analogous to that of a complete engine, to establish the most realistic test scenario possible. The modular approach of the elongated air path using standard vacuum parts allowed for quick and flexible adjustments. Additionally, modifying the parameters of the orifice (diameter, length) provided further degrees of freedom. These details were very helpful in making quick and easy adjustments, finally leading to a setup that resulted in emissions quite similar to those of an engine. Thus, it should also be possible to match the emission behavior of other engines by simply modifying the elongated air path and/or the orifice to the fuel reservoir.

The measurement of four different HC traps in an initial test produced plausible values for their efficiencies. These four traps were also tested on an engine, yielding similar efficiencies. The flow-through HC trap (sample A) showed the highest efficiency of about 90%. The carbon paper containing traps (samples C and D) had considerably lower values of about 50% and the teabag-style HC trap (sample B) had an efficiency of only about 40%. Literature does not provide many values for comparison. Moyer et al. [10] only mentioned that flow-through traps are the most efficient design, while bypass traps are less effective, which aligns with our findings. Maeda et al. [13] reported emission rates of 10% to 40%. The corresponding efficiencies of 60% to 90% are in the same range as our results, even though further details are missing.

The repeatability of the entire procedure was assessed by measuring two other HC traps (samples E1 and E2) and the test rig without an HC trap. The measurements without the HC trap were conducted over a longer time range in two periods. Results showed a slight difference between the first and the second set of values. The first set yielded a mean value of 31.5 mg with a standard deviation of 1.3 mg, while the second set yielded a mean of 28.5 mg with a standard deviation of 1.2 mg. No specific reason for this difference could be identified. The test rig and procedure were identical, and the measurement system showed no anomalies. Until another reason is found, it is recommended to perform tests in a short time range to optimize repeatability or to consider higher uncertainties for longer time ranges. Considering all values, the relative uncertainty of the whole procedure was about 7% (confidence level 65%). This limits the significance of results for HC traps with quite similar efficiencies (efficiency difference less than 10%).

Another important parameter is the emission mass level, as it can affect the efficiency of HC traps. Three different levels were selected by injecting 100 µL, 200 µL, and 300 µL of test fuel. The resulting emission masses without HC trap were approximately 17 mg, 30 mg, and almost 40 mg. According to our knowledge, this is a meaningful range for modern engines. Unfortunately, no other values could be found in the literature to support our data. The determined values for the efficiencies of two HC traps (E1 and E2) showed no dependence on the emission mass level, at least within this range.

In many SHED tests, the hydrocarbon concentration within the air filter box was measured using point source FID sampling and an MOS sensor. Without an HC trap, the hydrocarbon concentrations were quite high (up to 120 ppm) and unsteady. With an HC trap, the concentrations were considerably lower and more stable. As expected, the absolute values recorded with the MOS sensor were not meaningful due to its low calibration stability. However, the qualitative changes in the values recorded by point source FID sampling and MOS sensor were in good agreement. Furthermore, there seems to be a correlation between the hydrocarbon concentration within the air filter box and the SHED test result. The arithmetic mean of the hydrocarbon concentrations in the air filter box was plotted against the emission mass of the SHED test, resulting in an approximate linear correlation. This might be useful for measurements of the entire vehicle. By additionally recording the hydrocarbon concentration within the air filter box, the contribution of the AIS path can be assessed. This information may be useful for locating or narrowing down emission sources, especially in case of too high overall values.

In summary, it was possible to develop a suitable component test rig for HC traps in the AIS. It can be adjusted to achieve similar emission characteristics compared to an engine and uses gasoline fuel as the emission source, increasing comparability with vehicle conditions. The repeatability of the entire procedure was assessed to be about 7%, making it possible to evaluate the significance of results when comparing the efficiency of different HC traps. This method can help OEMs, and suppliers improve the efficiency of their HC traps, resulting in lower overall evaporative emissions from their vehicles and a reduced environmental footprint. Furthermore, this method might be useful to determine the deterioration effects of HC traps. Up to now, this topic has not been addressed by regulatory standards, unlike carbon canisters for the fuel tank. For the latter test, procedures for aging are given, e.g., canister fuel aging in Commission Regulation (EU) 2017/1221 [9]. By determining the BWC (butane working capacity) at the beginning and at the end of the procedure, a potential aging effect of the canister can be detected. Similarly, this would be possible for an HC trap. The efficiency can be determined by the procedure described here. Additionally, a loading and unloading procedure (fuel aging) for HC traps would be needed. By comparing the efficiency before and after the fuel aging, a deterioration effect of an HC trap can be determined.

For future work, two topics are interesting. The first is improving the repeatability. This would be helpful to determine differences between HC traps with similar efficiencies. The single steps of the test procedure need to be analyzed to find certain steps, which have a significant effect on the repeatability. The second topic is the potential deterioration effects of HC traps. For that, a fuel aging procedure needs to be established. Limits for evaporative emission must be fulfilled during a long period of a car’s lifetime (e.g., CARB: 15 years or 150,000 miles [16]). So possible deterioration effects of HC traps should be interesting for OEM and suppliers but also for government agencies.