Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
45 Chapter 4 Bench Scale Testing of Open Graded Friction Course A limited laboratory testing program was completed to evaluate the effectiveness of ferric oxide media in removing dissolved metals in open-graded friction course (OGFC) pavement under laboratory-controlled conditions. Results from the laboratory study described below showed the potential for ferric oxide media to remove dissolved metals in a typical OGFC pavement for a range of aggregate types. The bench scale testing scope was designed to demonstrate âproof of conceptâ rather than provide enough information for engineering design of this BMP type for treatment of stormwater runoff. 4.1 Experimental Design 4.1.1 Testing Apparatus Figure 4-1 Testing apparatus to evaluate ferric oxide media in OGFC. The bench scale testing apparatus consists of a plastic column with OGFC specimen and synthetic highway stormwater as shown schematically in Figure 4-1. The flow rate through the column is controlled by an adjustable valve under the apparatus. To simulate field conditions on a typical OGFC pavement section (Figure 4-1), typical contact time (i.e., residence time of a dissolved metal particle in pavement layer) was estimated and several circulation cycles of the synthetic highway stormwater were completed for a duration corresponding to the estimated contact time. To avoid contamination in setup, the apparatus was cleaned by rinsing with several volumes of diluted acid followed by several volumes of distilled water prior to use in the experiments.
46 4.1.2 OGFC Specimen Preparation For bench scale testing, two OGFC mix designs were completed by American Engineering Testing (AET) following Superpave and other relevant standards as shown in Appendix A. Two non-carbonate aggregates: (a) granite and (b) quartzite, were selected for this laboratory study. Target air void for the design and preparation of the laboratory specimens was 18% which corresponds to the average value used by transportation agencies in the US for OGFC (Putman, 2012). A typical asphalt binder commonly used in pavement projects in the Midwest (i.e., PG 64-22) was used to prepare all OGFC samples. A total of 14 OGFC cylindrical specimens of 6-in diameter and 4.5 inches of height were prepared by AET using a ServoPac Gyratory Compactor (SGC). Eight specimens (four for each aggregate source) used for development of experimental setup and six specimens (three for each aggregate source) for actual completion of bench scale testing scope. OGFC specimens for the granite and quartzite aggregates are shown Figure 4-2 and Figure 4-3, respectively. Figure 4-2 Granite OGFC sample. Figure 4-3 Quartzite OGFC sample. 4.1.3 Experimental Design The following variables were studied in the OGFC bench scale experiment: 1. Aggregate: Two non-carbonate aggregates (i.e., granite and quartzite) with different properties (e.g., grain size distribution, absorption, angularity, and mineralogy) that are commonly used for OGFC projects in the Midwest.
47 2. Control: A control test was conducted for one aggregate and contact time combination. The control test consisted of testing the OGFC sample without ferric oxide. The control test was conducted following the same procedure as the tests conducted with ferric oxide. 3. Contact time: Three contact times (high, medium, and low) were investigated. Selection of contact times were based on typical OGFC pavement geometry as shown in Figure 4-4 and typical OGFC permeability (i.e., k=100 to 250 inches per hour). Contact times were accounted for in the bench scale experiment by cycling the synthetic highway stormwater as shown in Figure 4-1. The three contact times were estimated using the following typical OGFC pavement geometry (Figure 4-4). ï· t=1.5 in. ï· ï¡=2% ï· L= 12 ft (high contact time â particle A), 6 ft (medium contact time â particle B), 3 ft (low contact time â Particle C) Figure 4-4 Contact times for three dissolved metal particles (A, B, and C) in OGFC pavement. To estimate contact times, typical rainfall intensity (I) was selected from available data at the National Oceanic and Atmospheric Administration (NOAA) website (https://hdsc.nws.noaa.gov/hdsc/pfds/). Rainfall data from three states (i.e., Connecticut, Indiana, and California) was used to cover representative precipitation conditions in the US. Two duration events (i.e., 3 and 24 hr) with one year recurrence was selected for estimation of typical average rainfall intensity as shown in Table 4-1. Table 4-1 Selection of Rainfall Intensity from NOAA. Recurrence Interval (y), Duration (hr) State (in hr -1) Average (in hr-1) Connecticut Indiana California 1 year, 24 hr 0.108 0.102 0.121 0.110 1 year, 3 hr 0.482 0.592 0.316 0.463 The two selected rainfall intensities shown in Table 4-1 were used in Finite-Element (FE) based particle transport simulations to estimate the high, medium, and low contact times.
48 FE-based particle transport simulations were completed using GeoStudio Suite, which is a software package developed by GEO-SLOPE International. A subprogram of the suite can be used to model seepage flow through porous media and particle transport. The typical OGFC pavement geometry shown in Figure 4-4 and the rainfall intensities listed in Table 4-1 were used in the FE-based particle transport simulations. Typical seepage and particle transport simulation results are shown in Figure 4-5 and Figure 4-6, respectively. Figure 4-5 Typical FE seepage simulation results. Figure 4-6 Typical FE-based particle transport simulation for contact time estimation. The summary of results from the FE simulations to estimate the contact times is shown in Table 4-2. A high and low permeability value was used in the simulations to cover typical values for OGFC pavements. Two rainfall intensities were used to cover representative precipitation conditions in the US and particles at three different locations (12, 6, and 3 feet) were released in the simulations for estimation of the contact times presented in Table 4-2.
49 Table 4-2 Contact Time Estimation Based on FE Simulations. Particle k= 250 in hr-1 k=100 in hr-1 I = 0.46 in hr-1 I = 0.11 in hr-1 I = 0.46 in hr-1 I = 0.11 in hr-1 Contact Time (Travel Time), hr A (12 ft) 4.7 5.9 11.4 12.4 B (6 ft) 2.0 1.8 5.3 4.9 C (3 ft) 0.7 0.7 2.1 1.6 Selected contact times for the bench scale testing program correspond to the conservative scenario where metal particles are out of the pavement the fastest and thus have less interaction time with the ferric oxide media. The high, medium and low contact times used in this study are 4.7, 2.0, and 0.7 hours, respectively. The experimental matrix for the bench scale test is shown in Table 4-3. A total of 18 experiments were conducted in this laboratory program to evaluate effectiveness of ferric oxide media in removing dissolved metals in runoff stormwater flowing through OGFC pavements. Table 4-3 Experimental Design for OGFC Samples Containing Ferric Oxide Media. FactorÂ LevelsÂ DescriptionÂ AsphaltÂ BinderÂ 1Â PGÂ 64â22Â AggregateÂ 2Â GraniteÂ andÂ QuartziteÂ (nonâcarbonateÂ aggregates)Â AirÂ VoidÂ inÂ MixÂ 1Â 18%Â ReplicatesÂ 3Â OneÂ controlÂ (noÂ iron)Â andÂ twoÂ testsÂ withÂ ironÂ ContactÂ timeÂ 3Â HighÂ (4.7Â hr),Â MediumÂ (2.0Â hr),Â andÂ LowÂ (0.7Â hr)Â basedÂ onÂ typicalÂ rainfallÂ andÂ OGFCÂ geometryÂ andÂ permeabilityÂ 4.1.4 Synthetic Highway Stormwater Synthetic highway stormwater was prepared for the laboratory experiment based on the national highway runoff quality data available in the Stochastic Empirical Loading and Dilution Model (SELDM) for stormwater quality risk assessment developed by Granato, G.E. (2013) at USGS. SELDM contains a national database with 54,384 event mean concentrations (EMC) measurements of 194 water quality constituents monitored at 117 sites during 4,186 storm events (Granato, 2013). The database in SELDM was used to select the national cumulative probability distributions for each parameter and constituent (Figure 4-7, Figure 4-8, and Figure 4-9).
50 Figure 4-7 Cumulative distribution for pH from SELDM. Figure 4-8 Cumulative distribution for hardness from SELDM. 0 10 20 30 40 50 60 70 80 90 100 5.0 5.5 6.0 6.5 7.0 7.5 8.0 8.5 9.0 Pr ob ab ili ty Â o fÂ E xc ed ee nc eÂ (% ) pHÂ (s.u.) 0 10 20 30 40 50 60 70 80 90 100 1.0Eâ01 1.0E+01 1.0E+03 1.0E+05 Pr ob ab ili ty Â o fÂ E xc ed ee nc eÂ (% ) ConstituentÂ ConcentrationÂ (mg/LÂ asÂ CalciumÂ Cabonate)
51 Figure 4-9 Cumulative Distribution for Metals from SELDM. The national cumulative distributions shown above were used to estimate median storm event concentrations for pH, hardness, chromium, copper, lead, and zinc. SELDM does not provide estimates for nickel and arsenic and hence estimates for these values were determined from a published study that used highway runoff data from Minnesota (Thomson et al., 1996). Estimated composition for the synthetic highway runoff used in the laboratory experiments is shown in Table 4-4. The solution was buffered with sodium bicarbonate to approximately achieve the targeted synthetic water pH. Table 4-4 Synthetic Highway Runoff Composition. Metals are nominal. ConstituentÂ ConcentrationÂ UnitsÂ SourceÂ pHÂ 7.2Â s.u.Â SELDMÂ HardnessÂ 13.2Â mgÂ Lâ1Â asÂ CaCO3Â ClÂ 30.0Â mgÂ Lâ1Â CrÂ 12.0Â ÂµgÂ Lâ1Â CuÂ 26.9Â ÂµgÂ Lâ1Â PbÂ 8.7Â ÂµgÂ Lâ1Â ZnÂ 123.0Â ÂµgÂ Lâ1Â NiÂ 10.4Â ÂµgÂ Lâ1Â ThomsonÂ etÂ al.Â 1996Â AsÂ 19.3Â ÂµgÂ Lâ1Â 4.1.5 Ferric Oxide Media Application Ferric oxide media was obtained in granulated form from Brightwell Aquatics (Figure 4-10). In order to incorporate the ferric oxide into the OGFC sample, a mixture of water and crushed granulated ferric oxide media was circulated several cycles through the pore spaces of OGFC sample until the ferric oxide was retained by the OGFC. 0 10 20 30 40 50 60 70 80 90 100 1.0Eâ01 1.0E+01 1.0E+03 Pr ob ab ili ty Â o fÂ E xc ed ee nc eÂ (% ) ConstituentÂ ConcentrationÂ (ïg/L) Chromium Copper Lead Zinc
52 Figure 4-10 Ferric oxide media. The concentration of ferric oxide media used in the experiments was estimated from the maximum adsorptive capacity of granulated ferric oxide and zinc reported in NCHRP 767 study (Barrett et al., 2014). The selected iron concentration used was 7.8 grams of ferric oxide per OGFC sample, which corresponds to a design life of approximately 3 years for the chemistry in Table 4-4. 4.2 Experimental Procedure The bench scale testing apparatus schematically shown in Figure 4-1 was used to investigate the effect of the aforementioned factors on the effectiveness of ferric oxide media to remove dissolved metals from runoff stormwater. The effectiveness of the ferric oxide media was evaluated by measuring the quality of a specific volume of synthetic highway stormwater before and after it cycles through a laboratory-prepared OGFC specimen. The experiment consisted of the following steps: 1. Clean testing system with several volumes of diluted acid followed by several volumes of distilled water. 2. Prepare synthetic highway stormwater using concentrations shown in Section 4.1.4. 3. Collect untreated samples of synthetic highway sample for chemical testing. 4. Complete control testing (i.e., no ferric oxide media added to sample) for high contact time. a. Adjust flow rate in the test based on sample volume, porosity (18%), available stormwater in the apparatus (0.8 L), number of cycles (i.e., typical pavement cross section length/ sample thickness), and contact time selected in Section 4.1.3. b. Collect samples for chemical analysis of control testing to quantify metals removal by OGFC without ferric oxide.
53 5. Add ferric oxide media to OGFC sample using concentration indicated in Section 4.1.5 by completing several circulation cycles of a water-ferric oxide mix. 6. Complete testing for high, medium and low contact times. a. Adjust flow rates as indicated in Step 4a. b. Collect samples of synthetic highway sample for chemical testing. All samples were filtered through a 0.45 Âµm polyethersulfone filter prior to submittal for analysis. Dissolved metals were analyzed for each collected sample and included copper, zinc, lead, nickel, chromium, iron, and arsenic. General chemical analysis included alkalinity, hardness, chloride, pH, specific conductance and dissolved oxygen. A single composite sample was collected for each of the experimental runs. Stock synthetic stormwater was collected and analyzed for each test. Two laboratory equipment blanks were collected and processed using the sample filtration apparatus and collection bottles to identify any potential for contamination with the apparatus. Equipment blank results were either below or near the reporting limits (e.g., within two times the reporting limits) indicating that the testing apparatus did not leach metals or adversely affect the results. 4.3 Results Metals were removed from synthetic stormwater for tests conducted with OGFC and no ferric oxide. As a consequence the results of the tests with OGFC and ferric oxide are provided uncorrected and corrected. The corrected test results reflect metals removal only by the ferric oxide material. Percent removals were calculated for the uncorrected results using the stock solution as the untreated metals concentration. The concentration removed by the OGFC material only was calculated using the difference between the treated concentration data generated for tests with OGFC and no ferric oxide and the stock solution. This difference was used to correct the percent removals for the tests with OGFC and ferric oxide to calculate removal only by the ferric oxide material. The correction was conducted to estimate the incremental benefit of using ferric oxide in OGFC. It should also be noted that the data discussed in the following paragraphs is the average of the duplicate tests except for the quartzite OGFC and lead as there was a suspected lead analysis error for the untreated stock solution. Uncorrected testing results (i.e., removal by the OGFC and ferric oxide) are shown in Figure 4-11 for quartzite OGFC and Figure 4-12 for granite OGFC. Comparison of the results in Figure 4-11 and Figure 4-12 shows that dissolved metals removal was not affected by aggregate material (e.g., quartzite or granite). Metals removal was high for all of the metals tested with the possible exception of chromium which also appeared to be most affected by the range of contact times tested. Arsenic, copper and zinc were largely unaffected by contact time whereas there were some effects for lead, nickel, and chromium. Overall, the results suggest that ferric oxide in OGFC will be effective for a wide range of contact times and storm event sizes. Corrected testing results (e.g., removal by ferric oxide only) are shown in Figure 4-13 for quartzite OGFC and Figure 4-14 or granite OGFC. Comparison of the results in Figure 4-13 and Figure 4-14 shows that
54 dissolved metals removal was not affected by aggregate material (e.g., quartzite or granite) for all of the metals tested except for nickel. It appears that copper removal by the ferric oxide and OGFC was largely due to the OGFC material rather than ferric oxide. This is consistent with the testing results at the Woodlynn Avenue and the Highway 36/61 ferric oxide sand filters where dissolved copper removal was highly variable. Also consistent with the Woodlynn Avenue and the Highway 36/61 ferric oxide sand filter results, removal was greatest for arsenic and zinc. Laboratory and general chemistry measurements taken for the tests are included in Table 4-5. There was an increase in specific conductance in the treated waters compared to the stock synthetic stormwater and an increase in contact time lead to an increase in specific conductance. The specific conductance increase corresponded with an increase in hardness. The other parameters were mostly stable during the tests and differed only minimally from the stock synthetic stormwater. Figure 4-11 Uncorrected percent removal of dissolved metals for quartzite OGFC for tests conducted with 0.7, 2, and 4.7 hours contact time. Â 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% As Cr Cu Ni Pb Zn Pe rc en tÂ R em ov al 0.7Â Hours 2.0Â Hours 4.7Â Hours
55 Figure 4-12 Uncorrected percent removal of dissolved metals for granite OGFC for tests conducted with 0.7, 2.0, and 4.7 hours contact time. Figure 4-13 Corrected percent removal of dissolved metals for quartzite OGFC for tests conducted with 0.7, 2, and 4.7 hours contact time. 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% As Cr Cu Ni Pb Zn Pe rc en tÂ R em ov al 0.7Â Hours 2.0Â Hours 4.7Â Hours 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% As Cr Cu Ni Zn Pb Pe rc en tÂ R em ov al 0.7Â Hours 2.0Â Hours 4.7Â Hours
56 Figure 4-14 Corrected percent removal of dissolved metals for granite OGFC for tests conducted with 0.7, 2.0, and 4.7 hours contact time. When interpreting these results it is important to note that testing was limited and thus no statistical analysis was completed. This laboratory program was intended as a âproof of conceptâ. There is no literature available regarding the use of ferric oxide in asphalt pavements and thus the effect of ferric oxide on the mechanical performance in terms of rutting, fatigue, and thermal cracking of open-graded mixtures is unknown. Full scale testing is needed before widespread application. Table 4-5 Average general chemical and physical conditions measured for the OGFC tests with quartzite and granite. ParameterÂ ContactÂ TimeÂ (hr)Â AverageÂ SyntheticÂ StormwaterÂ QuartziteÂ GraniteÂ 0.7Â 2.0Â 4.7Â 0.7Â 2.0Â 4.7Â TemperatureÂ (oC)Â 22Â 22Â 21Â 22Â 22Â 22Â 23Â DOÂ (mgÂ Lâ1)Â 6.3Â 6.1Â 6.5Â 6.2Â 6.2Â 6.4Â 4.9Â pHÂ (s.u.)Â 7.5Â 7.6Â 8.0Â 7.4Â 7.5Â 7.7Â 7.8Â SpecificÂ ConductanceÂ (ÂµsÂ cmâ1)Â 245Â 387Â 916Â 236Â 362Â 968Â 157Â TotalÂ AlkalinityÂ asÂ CaCO3Â (mg/L)Â 24Â 21Â 21Â 25Â 24Â 24Â 24Â ChlorideÂ (mg/L)Â 30Â 30Â 31Â 30Â 30Â 32Â 29Â HardnessÂ CaCO3Â (mg/L)Â 70Â 105Â 420Â 33Â 58Â 470Â 20Â 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% As Cr Cu Ni Zn Pb Pe rc en tÂ R em ov al 0.7Â Hours 2.0Â Hours 4.7Â Hours