This study was performed by the University of California Pavement Research Center (UCPRC) at the Davis campus for the Caltrans Division of Environmental Analysis. The focus of this study was to measure relevant parameters in the laboratory and use them as input for a computer model which simulated the structural and hydrologic performance of permeable pavements (particularly under medium speed and heavy load). Model results were used to evaluate structural and hydraulic performance. Results related to structural design, hydrologic design, and clogging and maintenance are summarized below.
7.1.1 Structural Performance
In developing pavement designs the mechanistic-empirical (ME) approach was used rather than the R-value design method as assumptions in the latter are not appropriate for fully permeable pavements (Jones et al. 2010, Li et al. 2012). Several design parameters were tested for subgrade soils and base course aggregate in the lab and used as input parameters in simulation studies.
These parameters include: Atterberg Limits, density-moisture relationships, permeability, and resilient modulus, among others. Also, tests were performed to determine permanent deformation of subgrade soils and dynamic cone penetrometer (DCP) tests were performed on base course materials.
Permeable pavements of both hot mix asphalt and Portland Cement concrete mixes were tested.
Different mix designs and aggregate gradations were tested for hot mix asphalt with regard to permeability, moisture sensitivity, rutting resistance, raveling resistance, fatigue cracking resistance, and flexural stiffness. Open-graded Portland Cement concrete mixes were tested for modulus of rupture.
In the modeling portion of the research, results of stress calculations in concrete and strain calculations in asphalt were used to estimate the thickness required to ensure the layer didn't fail due to fatigue. Nonlinear layer elastic theory was used to estimate the stiffness of the granular base. This was then used to estimate the shear stress to strength ratios in the subgrade. These results were used to develop structural design tables that can be used with hydraulic performance results (see below) to determine the required layer thickness. Design input variables were
subgrade permeability, truck traffic level (i.e. traffic index), climate/region, traffic speed, design storm for the climates/regions, and the number of adjacent impermeable lanes.
Two recommended design layouts were proposed and, based on that, numerous simulations were performed. Results from the computer modeling indicated that:
• Using the mechanistic-empirical design equations can be an effective way of determining the required thickness of fully permeable pavements so they are strong enough to carry heavy truck traffic,
• All required pavement structures were less than 5 ft in total thickness and most concrete slabs were less than 1.5 ft for the heaviest traffic. Thus, all pavements were considered feasible,
• Design cross-sections for shoulder retrofits of highways and low speed traffic areas are feasible as determined by construction and maintenance experts after review.
7.1.2 Hydrologic Performance
The focus of the hydraulic performance study was to determine the required aggregate depth of highway shoulders in order to provide adequate hydraulic capacity to capture the generated design rainfall volume. The hydraulic performance was assessed by solving Richards’ equation using a commercially available HYDRUS software program that is based on unsaturated flow theory. Rainstorms of 2, 50, and 100 year return periods were modeled and simulations were performed for three different climate regions (north with high rainfall, central with medium rainfall, and south with low rainfall) in California. The aggregate permeability was assumed to be constant at 10-1 cm/s and the subgrade permeability varied from constant values of 10-3 cm/s to 10-6 cm/s. Rainfall data, soil data, and other parameters were used as input data for HYDRUS in order to determine the required aggregate depth to capture the runoff volume from expected design storms. When performing the hydraulic simulations the following assumptions were made:
• Rain water infiltrated downward in the vertical direction with no lateral flow,
• Traffic lanes were impervious and all rainfall was directed towards permeable shoulders,
• Travel times between the location of runoff generation and infiltration were small compared to infiltration times,
• Rain intercepted by vehicles, lost as spray or evaporation was ignored,
• The water table was low and did not impact infiltration,
• There was no clogging of the surface layer,
• The top surface was at atmospheric conditions,
• There was no flux through the right or left side (i.e. rain water infiltrated vertically downward),
• At the bottom of the permeable pavements there was no pressure head and thus the system experienced free drainage,
• The initial water content of the soil was or was close to the residual water content.
Results obtained from hydraulic performance simulations showed that:
1. For most average storm designs an aggregate thickness of about 1 m was sufficient to capture the entire runoff generated from average rainfall in California. The highest aggregate depth under high rainfall was found to be about 3 m,
2. Required aggregate thickness was influenced by climate, storm recurrence interval, subgrade soil saturated hydraulic conductivity, aggregate void ratio, number of traffic lanes, and boundary conditions,
3. Higher rainfall amounts (and longer recurrence intervals) required larger aggregate thickness depths but the difference in required aggregate thickness for the 50 and 100 year storms was not significant,
4. Natural rainfall data input generated slightly thicker aggregate base depths as compared to synthetic rainfall data,
5. Native subgrade soil saturated hydraulic conductivity (soil permeability) was the factor that had the greatest impact on calculating the subgrade aggregate thickness. Native sub-grade soil saturated hydraulic conductivity values less than 10-5 cm/s was an important factor that made full depth permeable pavement impractical,
6. Highway surface area and number of traffic lanes also impacted the required aggregate thickness. Increasing traffic lanes from 2 to 4, increased the required aggregate thickness by 100%.
Additional detail information on hydraulic performance evaluation can be obtained from Kayhanian et al. (2010) and Chai et al. (2012).
7.1.3 Clogging Simulation Performance
To perform this simulation, three different clogging profiles were considered and it was assumed that the surface permeability of the upper 5 cm (2 in.) of HMA-O and PCC-O pavement was reduced to 1/100 of its original value. It was assumed that if there was no reduction in porosity in the upper surface of the pavement, the average porosity would have been about 25 percent. With this porosity the saturated hydraulic conductivity of PCC-O was estimated to be about 0.1 cm/sec (Kayhanian et al. 2010). In theory, a rainfall of about 360 cm/hr (141 in. /hr) would be needed to create surface runoff from either the HMA-O or PCC-O pavements. This amount of rainfall is unusual and will likely never occur. Surface runoff, however, can be generated when a significant portion of surface air voids are reduced due to clogging.
The permeability of twenty-three porous asphalt and pervious concrete parking lots was
measured and used as a basis to determine the clogging and surface infiltration of each parking lot. The parking lot age varied from less than a year to eight years old. Large variability was observed in permeability measurements within each parking lot and among all parking lots. In general, the average permeability in older parking lots was lower than newer parking lots;
indicating possible surface clogging and the importance of maintenance (Kayhanian et al. 2012a).
Theoretically, surface overflow will occur when the porosity of the upper surface layer of pavement decreases and when the surface hydraulic conductivity reaches a value less than the
rainfall intensity. The maximum rainfall intensity for the Sacramento area is about 10-3 cm/sec.
Therefore, the surface pavement hydraulic conductivity must decrease to 1 percent of its original values in order for surface runoff to occur, if there is no run-on water (permeable shoulders). If the ratio of impermeable to permeable area is three with permeable shoulders, the surface pavement hydraulic conductivity would need to decrease to 4 percent of its original value in order for surface runoff from the shoulder to occur. To investigate this aspect of the clogging issues, several simulations without run-on water were performed with surface saturated
conductivity values ranging from 1 to 5 percent of their original values. The clogging simulation results showed that no overflow failure occurred until the saturated hydraulic conductivity reached about 1 percent of the original saturated hydraulic conductivity value.
Since there was no field investigation involved with this project, no specific maintenance schedule was recommended. It was specified, however, that the full depth permeable pavement must be regularly cleaned in order to assure continuous permeability and infiltration of surface runoff through the pavement and base layers. Additional detailed information on permeability measurements and scanning evaluation of core samples can be obtained from Kayhanian et al.
(2010), Kayhanian et al. (2012a), and Manahiloh et al. (2012).
7.1.4 Life Cycle Cost Analysis
A Life-Cycle Cost Analysis was performed to evaluate the net present value economic costs of stormwater management alternatives. When performing the life-cycle cost analysis it was assumed that the fully permeable pavements that were simulated carried out the same function with regards to stormwater treatment (both runoff volume and water quality) as the other alternative SCMs. The full depth permeable pavements shoulder retrofit was considered for a high and low speed highways. The life-cycle cost analysis determined the net present value of the basic elements and included the analysis period, discount rate, costs, and salvage value. Most of the material and construction cost related to permeable pavement shoulder retrofits was obtained from local pavement construction companies. All costs were converted to the net present value and compared to other currently available stormwater management practices. The cost of permeable pavement shoulder retrofits was compared with the cost of conventional SCMs that were obtained from Caltrans pilot SCM retrofit study (Caltrans 2003).
Results showed that fully permeable pavements for the shoulder retrofits are more cost-effective than currently practiced SCMs in most scenarios. Fully permeable shoulders draining one lane of conventional asphalt had a net present value of about two-thirds of the next lowest cost
alternative SCM and fully permeable shoulders draining three lanes were about half the cost. In addition, a study performed by Houle et al. (2013) concluded that low impact development systems including permeable pavements, as compared to conventional treatment systems, have generally lower marginal maintenance burdens (as measured by cost and personnel hours).