| 1 | !> @file timestep.f90 |
|---|
| 2 | !--------------------------------------------------------------------------------! |
|---|
| 3 | ! This file is part of PALM. |
|---|
| 4 | ! |
|---|
| 5 | ! PALM is free software: you can redistribute it and/or modify it under the terms |
|---|
| 6 | ! of the GNU General Public License as published by the Free Software Foundation, |
|---|
| 7 | ! either version 3 of the License, or (at your option) any later version. |
|---|
| 8 | ! |
|---|
| 9 | ! PALM is distributed in the hope that it will be useful, but WITHOUT ANY |
|---|
| 10 | ! WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR |
|---|
| 11 | ! A PARTICULAR PURPOSE. See the GNU General Public License for more details. |
|---|
| 12 | ! |
|---|
| 13 | ! You should have received a copy of the GNU General Public License along with |
|---|
| 14 | ! PALM. If not, see <http://www.gnu.org/licenses/>. |
|---|
| 15 | ! |
|---|
| 16 | ! Copyright 1997-2016 Leibniz Universitaet Hannover |
|---|
| 17 | !--------------------------------------------------------------------------------! |
|---|
| 18 | ! |
|---|
| 19 | ! Current revisions: |
|---|
| 20 | ! ------------------ |
|---|
| 21 | ! Adapted for modularization of microphysics |
|---|
| 22 | ! |
|---|
| 23 | ! Former revisions: |
|---|
| 24 | ! ----------------- |
|---|
| 25 | ! $Id: timestep.f90 1849 2016-04-08 11:33:18Z maronga $ |
|---|
| 26 | ! |
|---|
| 27 | ! 1682 2015-10-07 23:56:08Z knoop |
|---|
| 28 | ! Code annotations made doxygen readable |
|---|
| 29 | ! |
|---|
| 30 | ! 1484 2014-10-21 10:53:05Z kanani |
|---|
| 31 | ! Changes due to new module structure of the plant canopy model: |
|---|
| 32 | ! calculations and parameters related to the plant canopy model removed |
|---|
| 33 | ! (the limitation of the canopy drag, i.e. that the canopy drag itself should |
|---|
| 34 | ! not change the sign of the velocity components, is now assured for in the |
|---|
| 35 | ! calculation of the canopy tendency terms in subroutine plant_canopy_model) |
|---|
| 36 | ! |
|---|
| 37 | ! 1342 2014-03-26 17:04:47Z kanani |
|---|
| 38 | ! REAL constants defined as wp-kind |
|---|
| 39 | ! |
|---|
| 40 | ! 1322 2014-03-20 16:38:49Z raasch |
|---|
| 41 | ! REAL functions provided with KIND-attribute |
|---|
| 42 | ! |
|---|
| 43 | ! 1320 2014-03-20 08:40:49Z raasch |
|---|
| 44 | ! ONLY-attribute added to USE-statements, |
|---|
| 45 | ! kind-parameters added to all INTEGER and REAL declaration statements, |
|---|
| 46 | ! kinds are defined in new module kinds, |
|---|
| 47 | ! old module precision_kind is removed, |
|---|
| 48 | ! revision history before 2012 removed, |
|---|
| 49 | ! comment fields (!:) to be used for variable explanations added to |
|---|
| 50 | ! all variable declaration statements |
|---|
| 51 | ! |
|---|
| 52 | ! 1257 2013-11-08 15:18:40Z raasch |
|---|
| 53 | ! openacc porting |
|---|
| 54 | ! bugfix for calculation of advective timestep in case of vertically stretched |
|---|
| 55 | ! grids |
|---|
| 56 | ! |
|---|
| 57 | ! 1092 2013-02-02 11:24:22Z raasch |
|---|
| 58 | ! unused variables removed |
|---|
| 59 | ! |
|---|
| 60 | ! 1053 2012-11-13 17:11:03Z hoffmann |
|---|
| 61 | ! timestep is reduced in two-moment cloud scheme according to the maximum |
|---|
| 62 | ! terminal velocity of rain drops |
|---|
| 63 | ! |
|---|
| 64 | ! 1036 2012-10-22 13:43:42Z raasch |
|---|
| 65 | ! code put under GPL (PALM 3.9) |
|---|
| 66 | ! |
|---|
| 67 | ! 1001 2012-09-13 14:08:46Z raasch |
|---|
| 68 | ! all actions concerning leapfrog scheme removed |
|---|
| 69 | ! |
|---|
| 70 | ! 978 2012-08-09 08:28:32Z fricke |
|---|
| 71 | ! restriction of the outflow damping layer in the diffusion criterion removed |
|---|
| 72 | ! |
|---|
| 73 | ! 866 2012-03-28 06:44:41Z raasch |
|---|
| 74 | ! bugfix for timestep calculation in case of Galilei transformation, |
|---|
| 75 | ! special treatment in case of mirror velocity boundary condition removed |
|---|
| 76 | ! |
|---|
| 77 | ! Revision 1.1 1997/08/11 06:26:19 raasch |
|---|
| 78 | ! Initial revision |
|---|
| 79 | ! |
|---|
| 80 | ! |
|---|
| 81 | ! Description: |
|---|
| 82 | ! ------------ |
|---|
| 83 | !> Compute the time step under consideration of the FCL and diffusion criterion. |
|---|
| 84 | !------------------------------------------------------------------------------! |
|---|
| 85 | SUBROUTINE timestep |
|---|
| 86 | |
|---|
| 87 | |
|---|
| 88 | USE arrays_3d, & |
|---|
| 89 | ONLY: dzu, dzw, kh, km, u, v, w |
|---|
| 90 | |
|---|
| 91 | USE control_parameters, & |
|---|
| 92 | ONLY: cfl_factor, coupling_mode, dt_3d, dt_fixed, dt_max, & |
|---|
| 93 | galilei_transformation, old_dt, message_string, & |
|---|
| 94 | stop_dt, terminate_coupled, terminate_coupled_remote, & |
|---|
| 95 | timestep_reason, u_gtrans, use_ug_for_galilei_tr, v_gtrans |
|---|
| 96 | |
|---|
| 97 | USE cpulog, & |
|---|
| 98 | ONLY: cpu_log, log_point |
|---|
| 99 | |
|---|
| 100 | USE grid_variables, & |
|---|
| 101 | ONLY: dx, dx2, dy, dy2 |
|---|
| 102 | |
|---|
| 103 | USE indices, & |
|---|
| 104 | ONLY: nxl, nxlg, nxr, nxrg, nyn, nyng, nys, nysg, nzb, nzt |
|---|
| 105 | |
|---|
| 106 | USE interfaces |
|---|
| 107 | |
|---|
| 108 | USE kinds |
|---|
| 109 | |
|---|
| 110 | USE microphysics_mod, & |
|---|
| 111 | ONLY: dt_precipitation |
|---|
| 112 | |
|---|
| 113 | USE pegrid |
|---|
| 114 | |
|---|
| 115 | USE statistics, & |
|---|
| 116 | ONLY: flow_statistics_called, hom, u_max, u_max_ijk, v_max, v_max_ijk,& |
|---|
| 117 | w_max, w_max_ijk |
|---|
| 118 | |
|---|
| 119 | IMPLICIT NONE |
|---|
| 120 | |
|---|
| 121 | INTEGER(iwp) :: i !< |
|---|
| 122 | INTEGER(iwp) :: j !< |
|---|
| 123 | INTEGER(iwp) :: k !< |
|---|
| 124 | |
|---|
| 125 | REAL(wp) :: div !< |
|---|
| 126 | REAL(wp) :: dt_diff !< |
|---|
| 127 | REAL(wp) :: dt_diff_l !< |
|---|
| 128 | REAL(wp) :: dt_u !< |
|---|
| 129 | REAL(wp) :: dt_u_l !< |
|---|
| 130 | REAL(wp) :: dt_v !< |
|---|
| 131 | REAL(wp) :: dt_v_l !< |
|---|
| 132 | REAL(wp) :: dt_w !< |
|---|
| 133 | REAL(wp) :: dt_w_l !< |
|---|
| 134 | REAL(wp) :: u_gtrans_l !< |
|---|
| 135 | REAL(wp) :: u_max_l !< |
|---|
| 136 | REAL(wp) :: u_min_l !< |
|---|
| 137 | REAL(wp) :: value !< |
|---|
| 138 | REAL(wp) :: v_gtrans_l !< |
|---|
| 139 | REAL(wp) :: v_max_l !< |
|---|
| 140 | REAL(wp) :: v_min_l !< |
|---|
| 141 | REAL(wp) :: w_max_l !< |
|---|
| 142 | REAL(wp) :: w_min_l !< |
|---|
| 143 | |
|---|
| 144 | REAL(wp), DIMENSION(2) :: uv_gtrans !< |
|---|
| 145 | REAL(wp), DIMENSION(2) :: uv_gtrans_l !< |
|---|
| 146 | REAL(wp), DIMENSION(3) :: reduce !< |
|---|
| 147 | REAL(wp), DIMENSION(3) :: reduce_l !< |
|---|
| 148 | REAL(wp), DIMENSION(nzb+1:nzt) :: dxyz2_min !< |
|---|
| 149 | |
|---|
| 150 | |
|---|
| 151 | |
|---|
| 152 | CALL cpu_log( log_point(12), 'calculate_timestep', 'start' ) |
|---|
| 153 | |
|---|
| 154 | ! |
|---|
| 155 | !-- In case of Galilei-transform not using the geostrophic wind as translation |
|---|
| 156 | !-- velocity, compute the volume-averaged horizontal velocity components, which |
|---|
| 157 | !-- will then be subtracted from the horizontal wind for the time step and |
|---|
| 158 | !-- horizontal advection routines. |
|---|
| 159 | IF ( galilei_transformation .AND. .NOT. use_ug_for_galilei_tr ) THEN |
|---|
| 160 | IF ( flow_statistics_called ) THEN |
|---|
| 161 | ! |
|---|
| 162 | !-- Horizontal averages already existent, just need to average them |
|---|
| 163 | !-- vertically. |
|---|
| 164 | u_gtrans = 0.0_wp |
|---|
| 165 | v_gtrans = 0.0_wp |
|---|
| 166 | DO k = nzb+1, nzt |
|---|
| 167 | u_gtrans = u_gtrans + hom(k,1,1,0) |
|---|
| 168 | v_gtrans = v_gtrans + hom(k,1,2,0) |
|---|
| 169 | ENDDO |
|---|
| 170 | u_gtrans = u_gtrans / REAL( nzt - nzb, KIND=wp ) |
|---|
| 171 | v_gtrans = v_gtrans / REAL( nzt - nzb, KIND=wp ) |
|---|
| 172 | ELSE |
|---|
| 173 | ! |
|---|
| 174 | !-- Averaging over the entire model domain. |
|---|
| 175 | u_gtrans_l = 0.0_wp |
|---|
| 176 | v_gtrans_l = 0.0_wp |
|---|
| 177 | !$acc parallel present( u, v ) |
|---|
| 178 | DO i = nxl, nxr |
|---|
| 179 | DO j = nys, nyn |
|---|
| 180 | DO k = nzb+1, nzt |
|---|
| 181 | u_gtrans_l = u_gtrans_l + u(k,j,i) |
|---|
| 182 | v_gtrans_l = v_gtrans_l + v(k,j,i) |
|---|
| 183 | ENDDO |
|---|
| 184 | ENDDO |
|---|
| 185 | ENDDO |
|---|
| 186 | !$acc end parallel |
|---|
| 187 | uv_gtrans_l(1) = u_gtrans_l / REAL( (nxr-nxl+1)*(nyn-nys+1)*(nzt-nzb), KIND=wp ) |
|---|
| 188 | uv_gtrans_l(2) = v_gtrans_l / REAL( (nxr-nxl+1)*(nyn-nys+1)*(nzt-nzb), KIND=wp ) |
|---|
| 189 | #if defined( __parallel ) |
|---|
| 190 | IF ( collective_wait ) CALL MPI_BARRIER( comm2d, ierr ) |
|---|
| 191 | CALL MPI_ALLREDUCE( uv_gtrans_l, uv_gtrans, 2, MPI_REAL, MPI_SUM, & |
|---|
| 192 | comm2d, ierr ) |
|---|
| 193 | u_gtrans = uv_gtrans(1) / REAL( numprocs, KIND=wp ) |
|---|
| 194 | v_gtrans = uv_gtrans(2) / REAL( numprocs, KIND=wp ) |
|---|
| 195 | #else |
|---|
| 196 | u_gtrans = uv_gtrans_l(1) |
|---|
| 197 | v_gtrans = uv_gtrans_l(2) |
|---|
| 198 | #endif |
|---|
| 199 | ENDIF |
|---|
| 200 | ENDIF |
|---|
| 201 | |
|---|
| 202 | ! |
|---|
| 203 | !-- Determine the maxima of the velocity components, including their |
|---|
| 204 | !-- grid index positions. |
|---|
| 205 | #if defined( __openacc ) |
|---|
| 206 | IF ( dt_fixed ) THEN ! otherwise do it further below for better cache usage |
|---|
| 207 | u_max_l = -999999.9_wp |
|---|
| 208 | u_min_l = 999999.9_wp |
|---|
| 209 | v_max_l = -999999.9_wp |
|---|
| 210 | v_min_l = 999999.9_wp |
|---|
| 211 | w_max_l = -999999.9_wp |
|---|
| 212 | w_min_l = 999999.9_wp |
|---|
| 213 | !$acc parallel present( u, v, w ) |
|---|
| 214 | DO i = nxl, nxr |
|---|
| 215 | DO j = nys, nyn |
|---|
| 216 | DO k = nzb+1, nzt |
|---|
| 217 | u_max_l = MAX( u_max_l, u(k,j,i) ) |
|---|
| 218 | u_min_l = MIN( u_min_l, u(k,j,i) ) |
|---|
| 219 | v_max_l = MAX( v_max_l, v(k,j,i) ) |
|---|
| 220 | v_min_l = MIN( v_min_l, v(k,j,i) ) |
|---|
| 221 | w_max_l = MAX( w_max_l, w(k,j,i) ) |
|---|
| 222 | w_min_l = MIN( w_min_l, w(k,j,i) ) |
|---|
| 223 | ENDDO |
|---|
| 224 | ENDDO |
|---|
| 225 | ENDDO |
|---|
| 226 | !$acc end parallel |
|---|
| 227 | #if defined( __parallel ) |
|---|
| 228 | reduce_l(1) = u_max_l |
|---|
| 229 | reduce_l(2) = v_max_l |
|---|
| 230 | reduce_l(3) = w_max_l |
|---|
| 231 | IF ( collective_wait ) CALL MPI_BARRIER( comm2d, ierr ) |
|---|
| 232 | CALL MPI_ALLREDUCE( reduce_l, reduce, 3, MPI_REAL, MPI_MAX, comm2d, ierr ) |
|---|
| 233 | u_max = reduce(1) |
|---|
| 234 | v_max = reduce(2) |
|---|
| 235 | w_max = reduce(3) |
|---|
| 236 | reduce_l(1) = u_min_l |
|---|
| 237 | reduce_l(2) = v_min_l |
|---|
| 238 | reduce_l(3) = w_min_l |
|---|
| 239 | IF ( collective_wait ) CALL MPI_BARRIER( comm2d, ierr ) |
|---|
| 240 | CALL MPI_ALLREDUCE( reduce_l, reduce, 3, MPI_REAL, MPI_MIN, comm2d, ierr ) |
|---|
| 241 | IF ( ABS( reduce(1) ) > u_max ) u_max = reduce(1) |
|---|
| 242 | IF ( ABS( reduce(2) ) > v_max ) v_max = reduce(2) |
|---|
| 243 | IF ( ABS( reduce(3) ) > w_max ) w_max = reduce(3) |
|---|
| 244 | #else |
|---|
| 245 | IF ( ABS( u_min_l ) > u_max_l ) THEN |
|---|
| 246 | u_max = u_min_l |
|---|
| 247 | ELSE |
|---|
| 248 | u_max = u_max_l |
|---|
| 249 | ENDIF |
|---|
| 250 | IF ( ABS( v_min_l ) > v_max_l ) THEN |
|---|
| 251 | v_max = v_min_l |
|---|
| 252 | ELSE |
|---|
| 253 | v_max = v_max_l |
|---|
| 254 | ENDIF |
|---|
| 255 | IF ( ABS( w_min_l ) > w_max_l ) THEN |
|---|
| 256 | w_max = w_min_l |
|---|
| 257 | ELSE |
|---|
| 258 | w_max = w_max_l |
|---|
| 259 | ENDIF |
|---|
| 260 | #endif |
|---|
| 261 | ENDIF |
|---|
| 262 | #else |
|---|
| 263 | CALL global_min_max( nzb, nzt+1, nysg, nyng, nxlg, nxrg, u, 'abs', 0.0_wp, & |
|---|
| 264 | u_max, u_max_ijk ) |
|---|
| 265 | CALL global_min_max( nzb, nzt+1, nysg, nyng, nxlg, nxrg, v, 'abs', 0.0_wp, & |
|---|
| 266 | v_max, v_max_ijk ) |
|---|
| 267 | CALL global_min_max( nzb, nzt+1, nysg, nyng, nxlg, nxrg, w, 'abs', 0.0_wp, & |
|---|
| 268 | w_max, w_max_ijk ) |
|---|
| 269 | #endif |
|---|
| 270 | |
|---|
| 271 | IF ( .NOT. dt_fixed ) THEN |
|---|
| 272 | #if defined( __openacc ) |
|---|
| 273 | ! |
|---|
| 274 | !-- Variable time step: |
|---|
| 275 | !-- Calculate the maximum time step according to the CFL-criterion, |
|---|
| 276 | !-- individually for each velocity component |
|---|
| 277 | dt_u_l = 999999.9_wp |
|---|
| 278 | dt_v_l = 999999.9_wp |
|---|
| 279 | dt_w_l = 999999.9_wp |
|---|
| 280 | u_max_l = -999999.9_wp |
|---|
| 281 | u_min_l = 999999.9_wp |
|---|
| 282 | v_max_l = -999999.9_wp |
|---|
| 283 | v_min_l = 999999.9_wp |
|---|
| 284 | w_max_l = -999999.9_wp |
|---|
| 285 | w_min_l = 999999.9_wp |
|---|
| 286 | !$acc parallel loop collapse(3) present( u, v, w ) |
|---|
| 287 | DO i = nxl, nxr |
|---|
| 288 | DO j = nys, nyn |
|---|
| 289 | DO k = nzb+1, nzt |
|---|
| 290 | dt_u_l = MIN( dt_u_l, ( dx / ( ABS( u(k,j,i) - u_gtrans ) + 1.0E-10_wp ) ) ) |
|---|
| 291 | dt_v_l = MIN( dt_v_l, ( dy / ( ABS( v(k,j,i) - v_gtrans ) + 1.0E-10_wp ) ) ) |
|---|
| 292 | dt_w_l = MIN( dt_w_l, ( dzu(k) / ( ABS( w(k,j,i) ) + 1.0E-10_wp ) ) ) |
|---|
| 293 | u_max_l = MAX( u_max_l, u(k,j,i) ) |
|---|
| 294 | u_min_l = MIN( u_min_l, u(k,j,i) ) |
|---|
| 295 | v_max_l = MAX( v_max_l, v(k,j,i) ) |
|---|
| 296 | v_min_l = MIN( v_min_l, v(k,j,i) ) |
|---|
| 297 | w_max_l = MAX( w_max_l, w(k,j,i) ) |
|---|
| 298 | w_min_l = MIN( w_min_l, w(k,j,i) ) |
|---|
| 299 | ENDDO |
|---|
| 300 | ENDDO |
|---|
| 301 | ENDDO |
|---|
| 302 | !$acc end parallel |
|---|
| 303 | |
|---|
| 304 | #if defined( __parallel ) |
|---|
| 305 | reduce_l(1) = dt_u_l |
|---|
| 306 | reduce_l(2) = dt_v_l |
|---|
| 307 | reduce_l(3) = dt_w_l |
|---|
| 308 | IF ( collective_wait ) CALL MPI_BARRIER( comm2d, ierr ) |
|---|
| 309 | CALL MPI_ALLREDUCE( reduce_l, reduce, 3, MPI_REAL, MPI_MIN, comm2d, ierr ) |
|---|
| 310 | dt_u = reduce(1) |
|---|
| 311 | dt_v = reduce(2) |
|---|
| 312 | dt_w = reduce(3) |
|---|
| 313 | |
|---|
| 314 | reduce_l(1) = u_max_l |
|---|
| 315 | reduce_l(2) = v_max_l |
|---|
| 316 | reduce_l(3) = w_max_l |
|---|
| 317 | IF ( collective_wait ) CALL MPI_BARRIER( comm2d, ierr ) |
|---|
| 318 | CALL MPI_ALLREDUCE( reduce_l, reduce, 3, MPI_REAL, MPI_MAX, comm2d, ierr ) |
|---|
| 319 | u_max = reduce(1) |
|---|
| 320 | v_max = reduce(2) |
|---|
| 321 | w_max = reduce(3) |
|---|
| 322 | reduce_l(1) = u_min_l |
|---|
| 323 | reduce_l(2) = v_min_l |
|---|
| 324 | reduce_l(3) = w_min_l |
|---|
| 325 | IF ( collective_wait ) CALL MPI_BARRIER( comm2d, ierr ) |
|---|
| 326 | CALL MPI_ALLREDUCE( reduce_l, reduce, 3, MPI_REAL, MPI_MIN, comm2d, ierr ) |
|---|
| 327 | IF ( ABS( reduce(1) ) > u_max ) u_max = reduce(1) |
|---|
| 328 | IF ( ABS( reduce(2) ) > v_max ) v_max = reduce(2) |
|---|
| 329 | IF ( ABS( reduce(3) ) > w_max ) w_max = reduce(3) |
|---|
| 330 | #else |
|---|
| 331 | dt_u = dt_u_l |
|---|
| 332 | dt_v = dt_v_l |
|---|
| 333 | dt_w = dt_w_l |
|---|
| 334 | |
|---|
| 335 | IF ( ABS( u_min_l ) > u_max_l ) THEN |
|---|
| 336 | u_max = u_min_l |
|---|
| 337 | ELSE |
|---|
| 338 | u_max = u_max_l |
|---|
| 339 | ENDIF |
|---|
| 340 | IF ( ABS( v_min_l ) > v_max_l ) THEN |
|---|
| 341 | v_max = v_min_l |
|---|
| 342 | ELSE |
|---|
| 343 | v_max = v_max_l |
|---|
| 344 | ENDIF |
|---|
| 345 | IF ( ABS( w_min_l ) > w_max_l ) THEN |
|---|
| 346 | w_max = w_min_l |
|---|
| 347 | ELSE |
|---|
| 348 | w_max = w_max_l |
|---|
| 349 | ENDIF |
|---|
| 350 | #endif |
|---|
| 351 | |
|---|
| 352 | #else |
|---|
| 353 | ! |
|---|
| 354 | !-- Variable time step: |
|---|
| 355 | !-- Calculate the maximum time step according to the CFL-criterion, |
|---|
| 356 | !-- individually for each velocity component |
|---|
| 357 | dt_u_l = 999999.9_wp |
|---|
| 358 | dt_v_l = 999999.9_wp |
|---|
| 359 | dt_w_l = 999999.9_wp |
|---|
| 360 | DO i = nxl, nxr |
|---|
| 361 | DO j = nys, nyn |
|---|
| 362 | DO k = nzb+1, nzt |
|---|
| 363 | dt_u_l = MIN( dt_u_l, ( dx / ( ABS( u(k,j,i) - u_gtrans ) + 1.0E-10_wp ) ) ) |
|---|
| 364 | dt_v_l = MIN( dt_v_l, ( dy / ( ABS( v(k,j,i) - v_gtrans ) + 1.0E-10_wp ) ) ) |
|---|
| 365 | dt_w_l = MIN( dt_w_l, ( dzu(k) / ( ABS( w(k,j,i) ) + 1.0E-10_wp ) ) ) |
|---|
| 366 | ENDDO |
|---|
| 367 | ENDDO |
|---|
| 368 | ENDDO |
|---|
| 369 | |
|---|
| 370 | #if defined( __parallel ) |
|---|
| 371 | reduce_l(1) = dt_u_l |
|---|
| 372 | reduce_l(2) = dt_v_l |
|---|
| 373 | reduce_l(3) = dt_w_l |
|---|
| 374 | IF ( collective_wait ) CALL MPI_BARRIER( comm2d, ierr ) |
|---|
| 375 | CALL MPI_ALLREDUCE( reduce_l, reduce, 3, MPI_REAL, MPI_MIN, comm2d, ierr ) |
|---|
| 376 | dt_u = reduce(1) |
|---|
| 377 | dt_v = reduce(2) |
|---|
| 378 | dt_w = reduce(3) |
|---|
| 379 | #else |
|---|
| 380 | dt_u = dt_u_l |
|---|
| 381 | dt_v = dt_v_l |
|---|
| 382 | dt_w = dt_w_l |
|---|
| 383 | #endif |
|---|
| 384 | |
|---|
| 385 | #endif |
|---|
| 386 | |
|---|
| 387 | ! |
|---|
| 388 | !-- Compute time step according to the diffusion criterion. |
|---|
| 389 | !-- First calculate minimum grid spacing which only depends on index k |
|---|
| 390 | !-- Note: also at k=nzb+1 a full grid length is being assumed, although |
|---|
| 391 | !-- in the Prandtl-layer friction term only dz/2 is used. |
|---|
| 392 | !-- Experience from the old model seems to justify this. |
|---|
| 393 | dt_diff_l = 999999.0_wp |
|---|
| 394 | |
|---|
| 395 | DO k = nzb+1, nzt |
|---|
| 396 | dxyz2_min(k) = MIN( dx2, dy2, dzw(k)*dzw(k) ) * 0.125_wp |
|---|
| 397 | ENDDO |
|---|
| 398 | |
|---|
| 399 | !$OMP PARALLEL private(i,j,k,value) reduction(MIN: dt_diff_l) |
|---|
| 400 | !$OMP DO |
|---|
| 401 | !$acc parallel loop collapse(3) present( kh, km ) |
|---|
| 402 | DO i = nxl, nxr |
|---|
| 403 | DO j = nys, nyn |
|---|
| 404 | DO k = nzb+1, nzt |
|---|
| 405 | dt_diff_l = MIN( dt_diff_l, dxyz2_min(k) / & |
|---|
| 406 | ( MAX( kh(k,j,i), km(k,j,i) ) + 1E-20_wp ) ) |
|---|
| 407 | ENDDO |
|---|
| 408 | ENDDO |
|---|
| 409 | ENDDO |
|---|
| 410 | !$acc end parallel |
|---|
| 411 | !$OMP END PARALLEL |
|---|
| 412 | #if defined( __parallel ) |
|---|
| 413 | IF ( collective_wait ) CALL MPI_BARRIER( comm2d, ierr ) |
|---|
| 414 | CALL MPI_ALLREDUCE( dt_diff_l, dt_diff, 1, MPI_REAL, MPI_MIN, comm2d, & |
|---|
| 415 | ierr ) |
|---|
| 416 | #else |
|---|
| 417 | dt_diff = dt_diff_l |
|---|
| 418 | #endif |
|---|
| 419 | |
|---|
| 420 | ! |
|---|
| 421 | !-- The time step is the minimum of the 3-4 components and the diffusion time |
|---|
| 422 | !-- step minus a reduction (cfl_factor) to be on the safe side. |
|---|
| 423 | !-- The time step must not exceed the maximum allowed value. |
|---|
| 424 | dt_3d = cfl_factor * MIN( dt_diff, dt_u, dt_v, dt_w, & |
|---|
| 425 | dt_precipitation ) |
|---|
| 426 | dt_3d = MIN( dt_3d, dt_max ) |
|---|
| 427 | |
|---|
| 428 | ! |
|---|
| 429 | !-- Remember the restricting time step criterion for later output. |
|---|
| 430 | IF ( MIN( dt_u, dt_v, dt_w ) < dt_diff ) THEN |
|---|
| 431 | timestep_reason = 'A' |
|---|
| 432 | ELSE |
|---|
| 433 | timestep_reason = 'D' |
|---|
| 434 | ENDIF |
|---|
| 435 | |
|---|
| 436 | ! |
|---|
| 437 | !-- Set flag if the time step becomes too small. |
|---|
| 438 | IF ( dt_3d < ( 0.00001_wp * dt_max ) ) THEN |
|---|
| 439 | stop_dt = .TRUE. |
|---|
| 440 | |
|---|
| 441 | WRITE( message_string, * ) 'Time step has reached minimum limit.', & |
|---|
| 442 | '&dt = ', dt_3d, ' s Simulation is terminated.', & |
|---|
| 443 | '&old_dt = ', old_dt, ' s', & |
|---|
| 444 | '&dt_u = ', dt_u, ' s', & |
|---|
| 445 | '&dt_v = ', dt_v, ' s', & |
|---|
| 446 | '&dt_w = ', dt_w, ' s', & |
|---|
| 447 | '&dt_diff = ', dt_diff, ' s', & |
|---|
| 448 | '&u_max = ', u_max, ' m/s k=', u_max_ijk(1), & |
|---|
| 449 | ' j=', u_max_ijk(2), ' i=', u_max_ijk(3), & |
|---|
| 450 | '&v_max = ', v_max, ' m/s k=', v_max_ijk(1), & |
|---|
| 451 | ' j=', v_max_ijk(2), ' i=', v_max_ijk(3), & |
|---|
| 452 | '&w_max = ', w_max, ' m/s k=', w_max_ijk(1), & |
|---|
| 453 | ' j=', w_max_ijk(2), ' i=', w_max_ijk(3) |
|---|
| 454 | CALL message( 'timestep', 'PA0312', 0, 1, 0, 6, 0 ) |
|---|
| 455 | ! |
|---|
| 456 | !-- In case of coupled runs inform the remote model of the termination |
|---|
| 457 | !-- and its reason, provided the remote model has not already been |
|---|
| 458 | !-- informed of another termination reason (terminate_coupled > 0) before. |
|---|
| 459 | #if defined( __parallel ) |
|---|
| 460 | IF ( coupling_mode /= 'uncoupled' .AND. terminate_coupled == 0 ) THEN |
|---|
| 461 | terminate_coupled = 2 |
|---|
| 462 | IF ( myid == 0 ) THEN |
|---|
| 463 | CALL MPI_SENDRECV( & |
|---|
| 464 | terminate_coupled, 1, MPI_INTEGER, target_id, 0, & |
|---|
| 465 | terminate_coupled_remote, 1, MPI_INTEGER, target_id, 0, & |
|---|
| 466 | comm_inter, status, ierr ) |
|---|
| 467 | ENDIF |
|---|
| 468 | CALL MPI_BCAST( terminate_coupled_remote, 1, MPI_INTEGER, 0, comm2d, ierr) |
|---|
| 469 | ENDIF |
|---|
| 470 | #endif |
|---|
| 471 | ENDIF |
|---|
| 472 | |
|---|
| 473 | ! |
|---|
| 474 | !-- Ensure a smooth value (two significant digits) of the timestep. |
|---|
| 475 | div = 1000.0_wp |
|---|
| 476 | DO WHILE ( dt_3d < div ) |
|---|
| 477 | div = div / 10.0_wp |
|---|
| 478 | ENDDO |
|---|
| 479 | dt_3d = NINT( dt_3d * 100.0_wp / div ) * div / 100.0_wp |
|---|
| 480 | |
|---|
| 481 | ! |
|---|
| 482 | !-- Adjust the time step |
|---|
| 483 | old_dt = dt_3d |
|---|
| 484 | |
|---|
| 485 | ENDIF |
|---|
| 486 | |
|---|
| 487 | CALL cpu_log( log_point(12), 'calculate_timestep', 'stop' ) |
|---|
| 488 | |
|---|
| 489 | END SUBROUTINE timestep |
|---|
