National Academies Press: OpenBook

Guidance to Predict and Mitigate Dynamic Hydroplaning on Roadways (2021)

Chapter: 4. Hydroplaning Potential Assessment Tool (Beta Version)

« Previous: 3. Integrated Hydroplaning Model
Page 70
Suggested Citation:"4. Hydroplaning Potential Assessment Tool (Beta Version)." National Academies of Sciences, Engineering, and Medicine. 2021. Guidance to Predict and Mitigate Dynamic Hydroplaning on Roadways. Washington, DC: The National Academies Press. doi: 10.17226/26287.
×
Page 70
Page 71
Suggested Citation:"4. Hydroplaning Potential Assessment Tool (Beta Version)." National Academies of Sciences, Engineering, and Medicine. 2021. Guidance to Predict and Mitigate Dynamic Hydroplaning on Roadways. Washington, DC: The National Academies Press. doi: 10.17226/26287.
×
Page 71
Page 72
Suggested Citation:"4. Hydroplaning Potential Assessment Tool (Beta Version)." National Academies of Sciences, Engineering, and Medicine. 2021. Guidance to Predict and Mitigate Dynamic Hydroplaning on Roadways. Washington, DC: The National Academies Press. doi: 10.17226/26287.
×
Page 72
Page 73
Suggested Citation:"4. Hydroplaning Potential Assessment Tool (Beta Version)." National Academies of Sciences, Engineering, and Medicine. 2021. Guidance to Predict and Mitigate Dynamic Hydroplaning on Roadways. Washington, DC: The National Academies Press. doi: 10.17226/26287.
×
Page 73
Page 74
Suggested Citation:"4. Hydroplaning Potential Assessment Tool (Beta Version)." National Academies of Sciences, Engineering, and Medicine. 2021. Guidance to Predict and Mitigate Dynamic Hydroplaning on Roadways. Washington, DC: The National Academies Press. doi: 10.17226/26287.
×
Page 74
Page 75
Suggested Citation:"4. Hydroplaning Potential Assessment Tool (Beta Version)." National Academies of Sciences, Engineering, and Medicine. 2021. Guidance to Predict and Mitigate Dynamic Hydroplaning on Roadways. Washington, DC: The National Academies Press. doi: 10.17226/26287.
×
Page 75
Page 76
Suggested Citation:"4. Hydroplaning Potential Assessment Tool (Beta Version)." National Academies of Sciences, Engineering, and Medicine. 2021. Guidance to Predict and Mitigate Dynamic Hydroplaning on Roadways. Washington, DC: The National Academies Press. doi: 10.17226/26287.
×
Page 76
Page 77
Suggested Citation:"4. Hydroplaning Potential Assessment Tool (Beta Version)." National Academies of Sciences, Engineering, and Medicine. 2021. Guidance to Predict and Mitigate Dynamic Hydroplaning on Roadways. Washington, DC: The National Academies Press. doi: 10.17226/26287.
×
Page 77
Page 78
Suggested Citation:"4. Hydroplaning Potential Assessment Tool (Beta Version)." National Academies of Sciences, Engineering, and Medicine. 2021. Guidance to Predict and Mitigate Dynamic Hydroplaning on Roadways. Washington, DC: The National Academies Press. doi: 10.17226/26287.
×
Page 78
Page 79
Suggested Citation:"4. Hydroplaning Potential Assessment Tool (Beta Version)." National Academies of Sciences, Engineering, and Medicine. 2021. Guidance to Predict and Mitigate Dynamic Hydroplaning on Roadways. Washington, DC: The National Academies Press. doi: 10.17226/26287.
×
Page 79
Page 80
Suggested Citation:"4. Hydroplaning Potential Assessment Tool (Beta Version)." National Academies of Sciences, Engineering, and Medicine. 2021. Guidance to Predict and Mitigate Dynamic Hydroplaning on Roadways. Washington, DC: The National Academies Press. doi: 10.17226/26287.
×
Page 80
Page 81
Suggested Citation:"4. Hydroplaning Potential Assessment Tool (Beta Version)." National Academies of Sciences, Engineering, and Medicine. 2021. Guidance to Predict and Mitigate Dynamic Hydroplaning on Roadways. Washington, DC: The National Academies Press. doi: 10.17226/26287.
×
Page 81
Page 82
Suggested Citation:"4. Hydroplaning Potential Assessment Tool (Beta Version)." National Academies of Sciences, Engineering, and Medicine. 2021. Guidance to Predict and Mitigate Dynamic Hydroplaning on Roadways. Washington, DC: The National Academies Press. doi: 10.17226/26287.
×
Page 82

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.

70 4. HYDROPLANING POTENTIAL ASSESSMENT TOOL (BETA VERSION) As introduced in Chapter 1, the research used the results of the integrated model to developed simplified relationships, which were implemented in a beta version of a Hydroplaning Potential Assessment Tool developed to apply the Integrated Hydroplaning Model without the need of commercial software. This section describes the development of this tool, which estimates the PM using the simplified relationships. The PM is used as an indicator of the HP as summarized in Figure 45. The tool is supported by an User Manual (provided as an appendix to this report), which summarizes the fundamentals of vehicle hydroplaning, introduces the new definition of HP, and explains the main principles and factors that impact the accumulation of water on the road surface and the response of the vehicle to the accumulated water. The User Manual also discusses the development, data needs, and use of the tool and provides additional practical examples including assessment of the effectivity of mitigation strategies. Figure 45. Hydroplaning Potential Assessment (beta version) Tool Flow Diagram. The use of the beta version of the tool includes the following steps (Figure 46): 1. Step 1: Select a file containing a prepared coarse grid for the alignment. Once the file is selected, the Tool calculates the flow streamlines. 2. Step 2:Add the main surface characteristics and road geometric characteristics: macrotexture, cross slope, grade, radius of curvature, and roughness profile (based on SAE classification) and select a design rainfall intensity based on the location, duration, and average recurrence interval. The Tool then estimates and displays the distribution of water on the section under analysis and the maximum WFT.

71 3. Step 3: Select the design speed and braking deceleration, design vehicle, and tire condition (or approve the default). The Tool uses these values to compute a PM. 4. Step 4: Once the Tool gives the computed PM, compare the computed PM to determine if it is appropriate for the type of roadway being analyzed. This step has to be carried out by the user outside the software tool. Figure 46. Hydroplaning Potential Assessment Tool (beta version) User Interface. For the last step, the agency will need to define acceptable values for the PM (or PM percentage) for different types of facilities. For example, the agency may define different PM criteria for interstate, primary, and secondary roads to balance the risk associated with traveling on each type of roadway. One possible approach for setting these criteria would be analyzing hydroplaning- related crashes. 4.1. ROAD SURFACE PROPERTIES The assessment of HP requires characterizing the road surface by the road geometry (number of lanes, curvature, and longitudinal and transversal slopes), smoothness, and macrotexture properties. These properties not only affect the accumulation of water on the roadway, but also the available traction for the vehicles that drive on the road segment(s) under investigation. The geometric information can be obtained from the design blueprints or geometric design software for new roads or using inertial surface mapping systems to assess existing roads. Any mesh in Excel or comma delimited text with the x, y, z coordinates of the surface in ft will work.

72 However, it is recommended that a relatively coarse resolution in the horizontal dimensions (e.g., about 2-12 ft) be used to speed up the processing. The road smoothness or roughness and macrotexture can be estimated for new roads based on the type of pavement and category of the road or measured using high-speed laser profilers or surface mapping systems. 4.2. SIMPLIFIED CALCULATION OF WATER FILM THICKNESS The tool implements a simplified procedure that estimates the WFT produced by the 3-D modeling using a three-step enhanced 2-D model, which: 1. Determines the streamlines along the road from the road geometry file; the streamlines are the paths of steepest descent along the road. 2. Calculates the WFT along each streamline using a modification of the Gallaway equation that allows a change in the slope along the streamline. 3. Makes 2-D adjustments with resolution reduction and Gaussian kernel smoothing to simulate the diffusion of the water that occurs in 2-D flow. The modified Gallaway equation and the diffusion simulated by the Gaussian kernel are calibrated with the results of the 3-D water accumulation simulations discussed in section 3.2.3. The details of the three steps used to calculate the WFT are given below. 4.2.1. Determine Streamlines Streamlines are the paths of steepest descent along the road. To determine these paths, a uniform grid of 1,000 points (50 by 20) is generated on the road. Starting from each of these points, the path of steepest descent is determined from the input road geometry file. This is done by calculating the derivative of the vertical coordinate z in the x and y directions. As an example, Figure 47 shows the streamlines obtained along a transition section on an eight-lane roadway. Figure 47. Streamlines for a transition section on an 8-lane highway.

73 4.2.2. Calculate the WFT Using a Modified Gallaway Equation The WFT is calculated for each streamline using a modified version of the Gallaway equation for 1-D flow. The Gallaway equation is for a 1-D flow path having a constant slope. However, the streamlines provide a 1-D path along which the slope varies. To calculate the WFT along the varying slope path, the equation is used in an algorithm that incrementally calculates the WFT along the path. The algorithm works as follows: 1. Divide the flow path into n segments having length Li and slope Si (i = 1, …, n). The WFT is calculated at the beginning and end of every segment. 2. For the first segment, calculate the WFT at the end of the segment using the Gallaway et al. (1979) equation as follows: 0.43 0.42 1 1 1WFT A L S C −= × − (44) where 3 0.11 0.593.38 10A MTD I−= × C MTD= L = length of drainage path (ft), I = rainfall intensity (in/h) S = cross slope (ft/ft), and MTD = mean texture depth (in). 3. For the remaining segments, calculate the WFT as follows: ( )0.43 0.42i i i iWFT A EL L S C−= × + − (45) where ( ) 1 0.430.42 0.43 1 i i i SEL WFT C A − − +   = +    if i > 1 (46) EL1 = 0 ( )1 1 1 1 20i i i i WFT W W WFT− + − −= − + (47) 0.43 0.42 1i i iW A EL S C − −= × − (47c) 0.43 0.42 1 1 1i i iW A EL S C − − − −= × − (47d) The main feature of the procedure is the effective flow path length, ELi. It is defined as the length that results in a WFT equal to WFTi-1+ in a 1-D flow with slope Si. Because section i has a slope Si and total length Li, then the WFT at the end of section I, WFTi, can be calculated using the 1-D flow (Gallaway equation) with slope Si and length L = (Eli-1 + Li). The WFT at the end of each

74 segment i of the streamline can therefore be calculated knowing the WFT at the end of segment i- 1 with the WFT at the end of the first segment calculated using the 1-D Gallaway equation. It might seem reasonable to assume that WFTi-1+ = WFTi-1; however, because of the change in slope between the two segments, the WFT at the beginning of segment i will be slightly different than the WFT at the end of segment i−1. It can be verified that if the slope does not change, the procedure is equivalent to the Gallaway equation. 4.2.3. Gaussian Kernel Smoothing to Simulate Diffusion The WFT calculated for the streamlines cannot be directly used as an approximation of the 2-D flow for the following reasons: 1. The starting point of the streamlines is based on the grid of points used to generate the streamlines. This does not necessarily correspond to the starting point of water flow. 2. The streamlines are 1-D and the water is constrained to flow along each streamline. In two dimensions, the water will diffuse to adjacent locations if the adjacent streamlines have a different calculated WFT. The first issue is best illustrated by looking at a figure of the streamline for a specific example. Figure 48 shows the streamline on a transition on a two-lane road (including the shoulder) using a point density lower than the one finally implemented to better visualize the various flow paths. (a) Streamlines (b) Detail Figure 48. Streamlines for a transition section on 2-lane roadway including the shoulders. The road grade is 1% and consists of a crown section with a cross slope of 2% in each direction transitioning into a 4% cross slope in one direction. The bottom left quarter of the figure best illustrate the issue and is shown in Figure 48(b). The figure shows long streamlines starting at the section crown and shorter streamlines starting away from the crown further to the side of the road (closer to the bottom of the figure). This does not represent the flow of water. The water flow for these shorter streamlines should actually also start at the crown and, thus, these streamlines will predict a much smaller WFT than the actual WFT. Therefore, at each point, the adopted approach estimates the WFT as the maximum WFT in a neighborhood defined by a square window that is large enough to at least encompass one of the long streamlines.

75 The second issue related to the 1-D aspect of the streamlines is that the flow is restricted to follow the road geometry. In regions where the geometry changes, adjacent streamlines can have different lengths, resulting in very different WFTs. This results in a very sudden drop in WFT that is not realistic. To reduce this effect, we use a Gaussian kernel to smooth the drop in WFT between adjacent areas. Note that in areas where the WFT is relatively constant, the Gaussian kernel has practically no effect on the calculated WFT (because it is averaging values that are numerically very close). However, in areas where there is a sharp drop between adjacent streamlines, the Gaussian kernel will make the drop smoother. The smoothness of the transition depends on the selected bandwidth of the kernel as explained in section 4.1.4. An example of calculated WFT with the simplified WFT model on an eight-lane road is shown Figure 49. Figure 49. Smoothed WFT along a transition on 8-lane roadway. 4.2.4. Calibration of the WFT Model The simplified WFT model was calibrated to the results of the numerical solution obtained with the 3-D WFT model. Calibration was performed by adjusting the following parameters to best reproduce the results of the 3-D model: 1. The constant corresponding to the flow length in the Gallaway model; 2. Neighborhood size to estimate the maximum WFT; 3. Bandwidth of the Gaussian kernel. For the constants in the Gallaway model, the calibration procedure resulted in the exponent of the flow length being changed from 0.43 in the original model to 0.38 in the simplified WFT model. This is because the simplified WFT model with the original coefficient tended to overestimate the maximum WFT compared to the 3-D WFT model. Similar results were obtained by scientists at the Argonne National Laboratory. Sitek et al. (2020) simulated the condition of Gallaway et al. (1971) experiments that simulated cross slopes of approximately 1%, 2%, and 4%, and a rain intensity of about 5 in/h, a texture depth of 0.076 mm,

76 and two water temperatures of 25 °C (77 °F) and 4 °C (39 °F). The results of the simulations are reproduced in Figure 50. The CFD results are slightly lower than the Gallaway’s measurements at 1% and 4% cross slopes. Figure 50. Simulated WFT across the roadway without a curb compared with the Gallaway equation and experimental results (Sitek et al. 2020) For the neighborhood size and Gaussian kernel, Figure 51 illustrate how these two parameters affect the calculated WFT. Because there are a finite number of streamlines generated, some of the grid cells will have no streamlines going through them, which results in a calculated WFT of zero. In reality, the rainfall covers the whole road section and these cells should have water flow and a measured WFT. Therefore, the preliminary (before applying the Gaussian kernel) WFT thickness in each cell is assigned the maximum calculated WFT determined in a neighborhood centered on that cell. Figure 51. Top grid view of road section; streamlines do not cross every cell in the grid.

77 The neighborhood size has to be large enough that it includes at least one cell with calculated WFT not equal to zero and representative of that cell (i.e., has a long streamline crossing the neighborhood). In addition, the neighborhood size cannot be too large, so that it does not assign a WFT from a cell that is too far from the cell of interest, which is the cell at the center of the neighborhood (see Figure 51). In the next step, a (two-dimensional) Gaussian kernel is used to obtain a more realistic 2-D water flow on the surface. The Gaussian kernel is used to approximate the 2-D interaction of water flow because the streamlines represent 1-D water flow. The Gaussian kernel has weights that decrease as the distance between the location of interest and adjacent locations increases, which implicitly accounts for the fact that locations that are farther away have less effect on the water flow and WFT. Note that the neighborhood step is essential before applying the Gaussian kernel. Otherwise, the Gaussian kernel will include many locations with zero calculated WFT (dry locations), which results in a significant drop in the WFT. Figure 52 shows the effect of neighborhood size on the calculated WFT distribution. Figure 52(a) shows the case of a small neighborhood size and small kernel bandwidth (neighborhood size of 1 and bandwidth of ω/15, where ω is the final selected bandwidth). The neighborhood size and kernel bandwidth are so small that the resulting flow is nearly the same as the 1-dimensional flow of each streamline. This is reflected in the figure, where many of the streamlines shown in Figure 47 can be discerned. Figure 52(b) shows the WFT for the same small neighborhood size but with the kernel bandwidth ω (the final selected bandwidth). The WFT distribution is qualitatively closer to the 3-D model distribution; however, the maximum WFT is significantly reduced to 1.7 mm, and the WFT distribution consists of isolated regions. This is because the neighborhood size is too small, resulting in many cells having zero WFT before smoothing with the Gaussian kernel. Figure 52(c) through Figure 52(f) show the effect of increasing the neighborhood size from 5 cells to 240 cells while keeping the kernel bandwidth constant. Initially, increasing the neighborhood size results in a more realistic WFT distribution until a point where the neighborhood size becomes too large and the WFT distribution starts approaching a uniform distribution. The final selected neighborhood size for the simplified tool is 40 cells, which gives a WFT distribution shown in Figure 49 that is between the cases shown in Figure 52(d) and Figure 52(e). Figure 53 illustrates the effect of kernel bandwidth on the WFT distribution for a neighborhood size of 40 cells, with the smallest bandwidth shown in Figure 53(a). The final selected bandwidth is shown in Figure 53(d) (the same as in Figure 49). The figure shows how Gaussian smoothing results in a WFT distribution that is closer to the one obtained from the 3-D model. However, too much smoothing (i.e., too large of a bandwidth) results in a WFT that, again approaches a uniform distribution.

78 (a) Neighborhood size = 1 cell, bandwidth = ω/15, max WFT = 3.1 mm (b) Neighborhood size = 1 cell, bandwidth = ω, max WFT = 1.7 mm (c) Neighborhood size = 5 cells, bandwidth = ω, max WFT = 2.1 mm (d) Neighborhood size = 20 cells Bandwidth = ω, max WFT = 2.6 mm (e) Neighborhood size = 80 cells, bandwidth = ω, max WFT = 3.2 mm (f) Neighborhood size = 240 cells, bandwidth = ω, max WFT = 3.5 mm Figure 52. Effect of neighborhood size on 2-D distribution of WFT and maximum WFT.

79 (a) Bandwidth = ω/15, max WFT = 3.5 mm (b) Bandwidth = ω/3, max WFT = 3.5 mm (c) Bandwidth = ω/3, max WFT = 3.4 mm (d) Bandwidth = ω, max WFT = 2.9 mm (e) Bandwidth = 2ω, max WFT = 2.3 mm (f) Bandwidth = 4ω, max WFT = 1.9 mm Figure 53. Effect of Gaussian kernel bandwidth compared to selected bandwidth ω on 2-D distribution of WFT and maximum WFT.

80 4.3. PERFORMANCE MARGIN ESTIMATION Once the Hydroplaning Potential Assessment Tool has estimated the maximum WFT, the user can confirm the defaults or enter the vehicle characteristics and operating conditions that would be used for the analysis. The tool then computes the PM and PM percentage by interpolating the results for the closet WFTs. The following sections illustrate the use of the tool with examples and additional ones are provided in Appendix A. 4.3.1. Speed Effect This resulting PM can be compared with the target adopted by the agency. For example, if we adopt a threshold PM of 0.10 for an SUV with a worn tire, Figure 54 shows that the selected critical vehicle will meet the criteria at a speed of 104 km/h (65 mph) but not at 136 km/h (85 mph) on an eight-lane transition area. (a) v = 104 km/h, PM = 0.108 (b) v = 136 km/h, PM = 0.093 Figure 54. Results of the Hydroplaning Potential Assessment Tool for two different speeds. 4.3.2. Longitudinal Slope Effect To illustrate the effect of other factors, Figure 55 shows the results of a transition similar to the one considered in Figure 54 but with only two lanes. In this case, the PM would be appropriate at 136 km/h (85 mph) if the same 0.10 threshold is used, suggesting that the HP is lower. However, if the longitudinal slope (Sx) is reduced to 0%, the WFT increases to more than 3 mm and the PM falls below the threshold, suggesting that the site should be investigated for HP. A more comprehensive sensitivity analysis for all the relevant variables is presented in the tool manual included as Appendix A.

81 (a) 2 lanes, Sx = 1%, v = 136 km/h, WFT = 1.9 mm, PM = 0.112 (b) 2 lanes, Sx = 0%, v = 136 km/h, WFT = 3.1 mm, PM = 0.09 Figure 55. Example results for 2-lane transitions with variable longitudinal slope 4.4. LIMITATIONS AND POSSIBLE IMPROVEMENTS Because of the complexities of the hydroplaning phenomena and all the factors involved in the interaction between the vehicle, water, and roadway surface, it was necessary to make many simplifications throughout the project. The primary ones are summarized as following: • The Integrated Hydroplaning Model required various simplifications and assumptions, including: - Although the original plan was to integrate the hydrology, tire-water-pavement, and vehicle dynamics submodels to run fully coupled, this was not possible. The need to use different software packages and the complexity of interconverting the inputs and outputs prevented an efficient real-time coupling. This was considered appropriate because the various processes occur at different time scales, as explained in section 3.1. - The 3-D water required a rigid non-frictional wall outside the shoulder along the section for the numerical solution to converge. This boundary condition created artificially high WFT on the side of the modeled area, but it did not significantly affect the WFT in the area of interest, as discussed in sections 3.2.4. - The maximum braking deceleration was assumed to be equal to the maximum cornering acceleration to simplify the development of the Performance Envelope. This was necessary because the research team encountered some difficulties while modeling the partial braking in the FSI submodel. - The number of vehicles and tire types and conditions considered in the study was limited because of the long times required to run the FSI simulations.

82 - Given the budget available for the project, only very limited additional field testing was conducted as part of the effort, and the majority of the verification and validation was based on results from previous research found in the literature. • Additional simplifications were needed to be able to run the practical hydroplaning risk Assessment Tool without the need to purchase licenses for simulation software and preparing complicated input files. - The water accumulation modeling was simplified using an enhanced version of the existing 1-D models and was calibrated based on the results of the formal 3-D simulations. - The analysis requires the user to enter a pre-prepared mesh representing the pavement surface, and it does not automatically compute any average or representative curvature or longitudinal and transversal slopes. These must be entered manually by the user. - The vehicle response analysis that calculates the PM implemented in the beta version of the tool is limited to three possible vehicles with either new or completely worn tires. - The user must enter a simple combination of braking and cornering (curvature and deceleration) conditions for the vehicle handing assessment (default values are provided), but these values are not checked for compatibility with the input road surface geometry. • Finally, the proposed risk analysis was not implemented as part of the project because it was not possible to get enough hydroplaning crash data representative of the entire country to verify and calibrate the risk functions. Thus, a simplified approach based on defining threshold PM values is proposed. Although the developed approach, models, and tool represent a significant advance in the understanding and prediction of HP based on all these needed simplifications, further improvements are possible, as discussed later in the report (section 6.2).

Next: 5. Hydroplaning Mitigation Solutions »
Guidance to Predict and Mitigate Dynamic Hydroplaning on Roadways Get This Book
×
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

Hydroplaning is a serious problem that is associated with a relatively small but significant number of crashes. Statistics from various parts of the world indicate that approximately 15% to 20% of all road traffic crashes occur in wet weather conditions.

The TRB National Cooperative Highway Research Program's NCHRP Web-Only Document 300: Guidance to Predict and Mitigate Dynamic Hydroplaning on Roadways provides a novel, transformational approach to estimate hydroplaning based on the physics behind it. Using advanced fluid dynamics, tire, and vehicle response models, the project has developed a new way to assess the safety risks associated with vehicle hydroplaning. This research represents one of the first attempts to significantly upgrade understanding and methods to predict hydroplaning potential since the 1970s.

Supplemental to the document is a Hydroplaning Potential Assessment Tool and Excel files.

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  6. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  7. ×

    View our suggested citation for this chapter.

    « Back Next »
  8. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!