Not for Sale

• #### Appendix D Model Evaluation 203-214

The National Academies of Sciences, Engineering, and Medicine
500 Fifth St. N.W. | Washington, D.C. 20001

Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 173
APPENDIX B REVIEW OF MODELS MODEL SELECTION AND DEVELOPMENT Models that can be used to predict hydroplaning speeds and the depth of sheet flow over a roadway surface have been published In the literature. These models are identified and discussed in detail in this appendix. Several types of models are available for predicting the depth of sheet flow: One-d~mensional models: Two~uneDsional models; Depth of flow over porous pavements Porous media flow models; and Other models. These models were discussed in the main body of this report, and additional information regarding these models is presented here. One-Dimensional Dow Models Russnam and Ross f34) presented a mode! for one~'mensional flow over highway pavements based on equations developed by Chezy and Manning for open channel flow. Both B-1

OCR for page 173
equations represent turbulent, one-~unensional, steady-state flow. If the slope of the energy grade fine is assumed to be equal to the slope of the flow path, they represent uniform flow. The authors also noted that even at velocities and depths when flow would normally be considered to be I~nar, the impact of rainfall on the fluid surface created conditions that were turbulent. The Maurung equation was simplified by assuming a wide channel approximation where the hydraulic radius, R. is equivalent to the depth of flow. This resulted In the following equation, Be-: where: K (L I'm h= S n h = K - 1 Depth of flow (cm) Empirically determined constant Length of the flow path (m) Rainfall intensity (cm/h) m,n - Empirically determined exponents S = Slope of the flow path (m/m) The values of K, m, and n were determined from data collected on a rolled asphalt pavement with chipp~ngs and on a brushed concrete pavement. This led to the general equation for both pavement surfaces, B-2: h = 0 0~7 (L i)O-47 so.2o B-2 (Bar) (B-2)

OCR for page 173
Measurements of Mann~g's roughness coefficient, n, were not made, and there was no statistical evaluation of equation B-2 comparing it to the observed data. The authors also offered a discussion of the effect of cross-sIope on flow over pavements, suggesting that the total length of the flow path could be determined by using the vector sum of the slopes to determine the direction of flow as presented in equation B-3: if = ~ where: 11 +(mi m2 1/2 Lf - Length of the flow path - Width of the pavement lane I/m ~- Cross slope of the pavement I/m2 = Longitudmal slope of the pavement (slope In Me direction of travel) The report also presented the results of field tests that were used to develop the ~-3) previous equations. Although the relationships developed by these authors were not used In this study, their work does represent a significant contribution to the literature, and their data were used to verify the models developed during this study. B-3

OCR for page 173
Empirical Flow Models In 1971 and again in 1979, GalIaway et al. (3, 4~46J at the Texas Transportation Institute developed a set of empirical equations for predicting water depths on road surfaces. The equations were one-~mensional, inasmuch as they were developed from data collected from surfaces with slopes in a single principal direction. The equations, based on a regression analysis, used plane length, rainfall intensity, texture depth, and pavement slope to predict water depth. Depths were observed at 20 locations on a pavement section. In the 1971 study, (3J nine pavement surfaces were used, along with six slopes and five rainfall rates. The 1979 study (4J added observations on Portland cement concrete pavements to those in the 1971 study to improve the regression. The coefficients of determination, r2, were 0.68 and 0.83 for the regression equations reported In the two studies, respectively. The regression equation wid1 the highest coefficient of determination and the one representing all 1,059 observations on the surfaces is shown in the following (Ball: WET = 0.003726 L0519 i0562 MTDo.l2s so. - - - MTD where (B4) WET Water film thickness (in) (1 in-25.4 mm) Plane length (ft) 1 Rainfall intensifier (in/h) MTD - Mean texture depth (in) B4 (1 It =305 mm)

OCR for page 173
h Pavement slope (m/m) Water depth (in) = WET + MID To date, Gallaway's work represents the single most comprehensive set of water depth data collected on different types of pavements. The equation, however, is without a fundamental resistance variable, such as Manning's n, and has been regressed by combining pavements with different types of surfaces. These models were not used In this study. Flow Models Based on Kinematic Wave Equation The kinematic wave approximation was discussed in ache body of this report (equations 4 through 5). To expand, steady-state or equilibrium conditions correspond to the greatest depths on a flow surface. Under these conditions, depths have increased to the point that inflow is equal to outflow. For the steady-state case, the term bu/6t is zero. On impervious surfaces, such as Portland cement concrete, the infiltration rate is zero, and the term f in equations 4 and 5 is dropped from the right-hand side of the equation. Thus, if infiltration is zero, the equation representing the conservation of momentum is: bu ah i u fix g fix h Ox fx The kinematic approximation assumes that the velocity terms In equation B-5 are negligible (i.e., gradually varied flow) and that the gravitational forces are equal to the frictional forces. Equation B-5 then reduces to the following, B-6: B-5 03-5,

OCR for page 173
S = Sf (B-6) The kinematic wave approximation has been shown to be valid for conditions when K is greater than 20, and the Froude number, NE, is greater than O.S, where: S L KOX = - hO NF and ho = normal depth. Additionally, when NF is less than 0.5, N2 K > S . Rainfall-induced flow on most pavement surfaces falls within the criteria established by 7) (B-~) equations B-7 and Bee. Further development results in equations 6 through ~ as discussed in the body of this report. Two-D'mensional Flow Models Flow on highway pavements is a two~,mensional phenomenon. A vertical component to flow exists that would add a Bird dimension. Yet, since the fluid depths are so small, B-6

OCR for page 173
variations In flow In the indirection can be averaged, and flow may be accurately represented by a two-dimensional model. Equations B-9 through B-~1 present the equations of state, continuity, and conservation of momentum for two-~mensional flow. Conservation of mass: Gh + 6(uh) + 6(vh) = i - f = fix BY where h Depth of flow u - Spatially averaged velocities (x - direction) v = Spatially averaged velocities (y - direction) i = Rainfall intensity over the domain f TnfiItration rate ~Incoming rainfall minus Infiltration into subsurface Conservation of momentum In the Erection leads to: (B-9) bu bu bu dh _ + u- + v , ~+ g is, = g(Sox ~ Set) (B-IO) u (i -f) h Or cost ~ - ~ ~ h whereas conservation of momentum In the y-direction leads to: B-7

OCR for page 173
By 8t where and TV TV Oh U- + V- ~ g- = g (Soy v (i - f) nr cos6', h h + vt - Sly) u, v, h, i, and f are the same as described in B-9 g - S Ox, S Oy - So,, Sly, vt ex,ey - (B-1 1) Acceleration due to gravity (32.17 ft/s2 or 9.806 m/s2) Slope of the flow path In the x and y directions, respectively Slope of the energy grade line In the x and y directions, respectively Terminal rainfall velocity Angle of rainfall input with respect to the x- and y-axes The equations are simplified, as were the one~mensional equations, by negating the force of raindrop impact and, if appropriate, the infiltration term. In most two~mensional models the remaking terms are retained, and the partial differential terms are approximated using either a finite difference or finite element scheme. In general, the system of nonlinear, partial differential equations has no analytical solution and must be solved by numerical methods. Zhang and Cundy (33) developed a two~mension~ model for computing water depths and velocities on a three~imensional surface. The model allows an inclined plane with B-8

OCR for page 173
irregular topography and is based upon a two-step, explicit solution of the finite difference approximation of the continuity and momentum equations shown earlier in equations B-9 through Bell. The rainfall intensity was considered constant over a finite length but, like infiltration, could vary over discrete lengths. The model also allowed spatial variations in plane characteristics, including surface roughness and topography. Tayfur et al. (47) applied an implicit solution scheme to Zhang and Cundy's model to improve the number of iterations necessary to reach equilibrium conditions. The model was applied to a planar surface with a relatively steep slope (eight percent). Due to the nature of nonlinear equations, stability and convergence remained a problem. Specifically, instability at lower slopes was noted. Froehlich (4&~) developed a two~unensional, free-surface model to analyze flows affecting roadway structures, such as culverts, embankments, and bridges. It has the capacity to analyze unsteady, nonuniform flows but is limited to situations where flow enters or leaves the flow domain at the boundaries and cannot account for infiltration or flows due to a spatially distributed source, such as rainfall, on a pavement surface. Huebaer (49) developed a two- dimensional, steady-state, finite element flow model for flow over highway pavements with irregular topography. The program produced acceptable results for planar surfaces but encountered problems with stability and convergence when irregular topographies were Introduced. The two- dimensional models are cited here for completeness. They were considered too cumbersome for use In the drain age guidelines as developed for this project. B-9

OCR for page 173
TRIS PAGE ~ENUONALLY LEFT BLAT