| 1 | SUBROUTINE pres |
|---|
| 2 | |
|---|
| 3 | !------------------------------------------------------------------------------! |
|---|
| 4 | ! Current revisions: |
|---|
| 5 | ! ----------------- |
|---|
| 6 | ! |
|---|
| 7 | ! |
|---|
| 8 | ! Former revisions: |
|---|
| 9 | ! ----------------- |
|---|
| 10 | ! $Id: pres.f90 720 2011-04-06 13:10:30Z suehring $ |
|---|
| 11 | ! |
|---|
| 12 | ! 719 2011-04-06 13:05:23Z gryschka |
|---|
| 13 | ! Bugfix in volume flow control for double cyclic boundary conditions |
|---|
| 14 | ! |
|---|
| 15 | ! 709 2011-03-30 09:31:40Z raasch |
|---|
| 16 | ! formatting adjustments |
|---|
| 17 | ! |
|---|
| 18 | ! 707 2011-03-29 11:39:40Z raasch |
|---|
| 19 | ! Calculation of weighted average of p is now handled in the same way |
|---|
| 20 | ! regardless of the number of ghost layers (advection scheme), |
|---|
| 21 | ! multigrid and sor method are using p_loc for iterative advancements of |
|---|
| 22 | ! pressure, |
|---|
| 23 | ! localsum calculation modified for proper OpenMP reduction, |
|---|
| 24 | ! bc_lr/ns replaced by bc_lr/ns_cyc |
|---|
| 25 | ! |
|---|
| 26 | ! 693 2011-03-08 09:..:..Z raasch |
|---|
| 27 | ! bugfix: weighting coefficient added to ibm branch |
|---|
| 28 | ! |
|---|
| 29 | ! 680 2011-02-04 23:16:06Z gryschka |
|---|
| 30 | ! bugfix: collective_wait |
|---|
| 31 | ! |
|---|
| 32 | ! 675 2011-01-19 10:56:55Z suehring |
|---|
| 33 | ! Removed bugfix while copying tend. |
|---|
| 34 | ! |
|---|
| 35 | ! 673 2011-01-18 16:19:48Z suehring |
|---|
| 36 | ! Weighting coefficients added for right computation of the pressure during |
|---|
| 37 | ! Runge-Kutta substeps. |
|---|
| 38 | ! |
|---|
| 39 | ! 667 2010-12-23 12:06:00Z suehring/gryschka |
|---|
| 40 | ! New allocation of tend when ws-scheme and multigrid is used. This is due to |
|---|
| 41 | ! reasons of perforance of the data_exchange. The same is done with p after |
|---|
| 42 | ! poismg is called. |
|---|
| 43 | ! nxl-1, nxr+1, nys-1, nyn+1 replaced by nxlg, nxrg, nysg, nyng when no |
|---|
| 44 | ! multigrid is used. Calls of exchange_horiz are modified. |
|---|
| 45 | ! bugfix: After pressure correction no volume flow correction in case of |
|---|
| 46 | ! non-cyclic boundary conditions |
|---|
| 47 | ! (has to be done only before pressure correction) |
|---|
| 48 | ! Call of SOR routine is referenced with ddzu_pres. |
|---|
| 49 | ! |
|---|
| 50 | ! 622 2010-12-10 08:08:13Z raasch |
|---|
| 51 | ! optional barriers included in order to speed up collective operations |
|---|
| 52 | ! |
|---|
| 53 | ! 151 2008-03-07 13:42:18Z raasch |
|---|
| 54 | ! Bugfix in volume flow control for non-cyclic boundary conditions |
|---|
| 55 | ! |
|---|
| 56 | ! 106 2007-08-16 14:30:26Z raasch |
|---|
| 57 | ! Volume flow conservation added for the remaining three outflow boundaries |
|---|
| 58 | ! |
|---|
| 59 | ! 85 2007-05-11 09:35:14Z raasch |
|---|
| 60 | ! Division through dt_3d replaced by multiplication of the inverse. |
|---|
| 61 | ! For performance optimisation, this is done in the loop calculating the |
|---|
| 62 | ! divergence instead of using a seperate loop. |
|---|
| 63 | ! |
|---|
| 64 | ! 75 2007-03-22 09:54:05Z raasch |
|---|
| 65 | ! Volume flow control for non-cyclic boundary conditions added (currently only |
|---|
| 66 | ! for the north boundary!!), 2nd+3rd argument removed from exchange horiz, |
|---|
| 67 | ! mean vertical velocity is removed in case of Neumann boundary conditions |
|---|
| 68 | ! both at the bottom and the top |
|---|
| 69 | ! |
|---|
| 70 | ! RCS Log replace by Id keyword, revision history cleaned up |
|---|
| 71 | ! |
|---|
| 72 | ! Revision 1.25 2006/04/26 13:26:12 raasch |
|---|
| 73 | ! OpenMP optimization (+localsum, threadsum) |
|---|
| 74 | ! |
|---|
| 75 | ! Revision 1.1 1997/07/24 11:24:44 raasch |
|---|
| 76 | ! Initial revision |
|---|
| 77 | ! |
|---|
| 78 | ! |
|---|
| 79 | ! Description: |
|---|
| 80 | ! ------------ |
|---|
| 81 | ! Compute the divergence of the provisional velocity field. Solve the Poisson |
|---|
| 82 | ! equation for the perturbation pressure. Compute the final velocities using |
|---|
| 83 | ! this perturbation pressure. Compute the remaining divergence. |
|---|
| 84 | !------------------------------------------------------------------------------! |
|---|
| 85 | |
|---|
| 86 | USE arrays_3d |
|---|
| 87 | USE constants |
|---|
| 88 | USE control_parameters |
|---|
| 89 | USE cpulog |
|---|
| 90 | USE grid_variables |
|---|
| 91 | USE indices |
|---|
| 92 | USE interfaces |
|---|
| 93 | USE pegrid |
|---|
| 94 | USE poisfft_mod |
|---|
| 95 | USE poisfft_hybrid_mod |
|---|
| 96 | USE statistics |
|---|
| 97 | |
|---|
| 98 | IMPLICIT NONE |
|---|
| 99 | |
|---|
| 100 | INTEGER :: i, j, k, sr |
|---|
| 101 | |
|---|
| 102 | REAL :: ddt_3d, localsum, threadsum, d_weight_pres |
|---|
| 103 | |
|---|
| 104 | REAL, DIMENSION(1:2) :: volume_flow_l, volume_flow_offset |
|---|
| 105 | REAL, DIMENSION(1:nzt) :: w_l, w_l_l |
|---|
| 106 | |
|---|
| 107 | |
|---|
| 108 | CALL cpu_log( log_point(8), 'pres', 'start' ) |
|---|
| 109 | |
|---|
| 110 | |
|---|
| 111 | ddt_3d = 1.0 / dt_3d |
|---|
| 112 | d_weight_pres = 1.0 / weight_pres(intermediate_timestep_count) |
|---|
| 113 | |
|---|
| 114 | ! |
|---|
| 115 | !-- Multigrid method expects array d to have one ghost layer. |
|---|
| 116 | !-- |
|---|
| 117 | IF ( psolver == 'multigrid' ) THEN |
|---|
| 118 | |
|---|
| 119 | DEALLOCATE( d ) |
|---|
| 120 | ALLOCATE( d(nzb:nzt+1,nys-1:nyn+1,nxl-1:nxr+1) ) |
|---|
| 121 | |
|---|
| 122 | ! |
|---|
| 123 | !-- Since p is later used to hold the weighted average of the substeps, it |
|---|
| 124 | !-- cannot be used in the iterative solver. Therefore, its initial value is |
|---|
| 125 | !-- stored on p_loc, which is then iteratively advanced in every substep. |
|---|
| 126 | IF ( intermediate_timestep_count == 1 ) THEN |
|---|
| 127 | DO i = nxl-1, nxr+1 |
|---|
| 128 | DO j = nys-1, nyn+1 |
|---|
| 129 | DO k = nzb, nzt+1 |
|---|
| 130 | p_loc(k,j,i) = p(k,j,i) |
|---|
| 131 | ENDDO |
|---|
| 132 | ENDDO |
|---|
| 133 | ENDDO |
|---|
| 134 | ENDIF |
|---|
| 135 | |
|---|
| 136 | ELSEIF ( psolver == 'sor' .AND. intermediate_timestep_count == 1 ) THEN |
|---|
| 137 | |
|---|
| 138 | ! |
|---|
| 139 | !-- Since p is later used to hold the weighted average of the substeps, it |
|---|
| 140 | !-- cannot be used in the iterative solver. Therefore, its initial value is |
|---|
| 141 | !-- stored on p_loc, which is then iteratively advanced in every substep. |
|---|
| 142 | p_loc = p |
|---|
| 143 | |
|---|
| 144 | ENDIF |
|---|
| 145 | |
|---|
| 146 | ! |
|---|
| 147 | !-- Conserve the volume flow at the outflow in case of non-cyclic lateral |
|---|
| 148 | !-- boundary conditions |
|---|
| 149 | !-- WARNING: so far, this conservation does not work at the left/south |
|---|
| 150 | !-- boundary if the topography at the inflow differs from that at the |
|---|
| 151 | !-- outflow! For this case, volume_flow_area needs adjustment! |
|---|
| 152 | ! |
|---|
| 153 | !-- Left/right |
|---|
| 154 | IF ( conserve_volume_flow .AND. ( outflow_l .OR. outflow_r ) ) THEN |
|---|
| 155 | |
|---|
| 156 | volume_flow(1) = 0.0 |
|---|
| 157 | volume_flow_l(1) = 0.0 |
|---|
| 158 | |
|---|
| 159 | IF ( outflow_l ) THEN |
|---|
| 160 | i = 0 |
|---|
| 161 | ELSEIF ( outflow_r ) THEN |
|---|
| 162 | i = nx+1 |
|---|
| 163 | ENDIF |
|---|
| 164 | |
|---|
| 165 | DO j = nys, nyn |
|---|
| 166 | ! |
|---|
| 167 | !-- Sum up the volume flow through the south/north boundary |
|---|
| 168 | DO k = nzb_2d(j,i)+1, nzt |
|---|
| 169 | volume_flow_l(1) = volume_flow_l(1) + u(k,j,i) * dzw(k) |
|---|
| 170 | ENDDO |
|---|
| 171 | ENDDO |
|---|
| 172 | |
|---|
| 173 | #if defined( __parallel ) |
|---|
| 174 | IF ( collective_wait ) CALL MPI_BARRIER( comm1dy, ierr ) |
|---|
| 175 | CALL MPI_ALLREDUCE( volume_flow_l(1), volume_flow(1), 1, MPI_REAL, & |
|---|
| 176 | MPI_SUM, comm1dy, ierr ) |
|---|
| 177 | #else |
|---|
| 178 | volume_flow = volume_flow_l |
|---|
| 179 | #endif |
|---|
| 180 | volume_flow_offset(1) = ( volume_flow_initial(1) - volume_flow(1) ) & |
|---|
| 181 | / volume_flow_area(1) |
|---|
| 182 | |
|---|
| 183 | DO j = nysg, nyng |
|---|
| 184 | DO k = nzb_2d(j,i)+1, nzt |
|---|
| 185 | u(k,j,i) = u(k,j,i) + volume_flow_offset(1) |
|---|
| 186 | ENDDO |
|---|
| 187 | ENDDO |
|---|
| 188 | |
|---|
| 189 | ENDIF |
|---|
| 190 | |
|---|
| 191 | ! |
|---|
| 192 | !-- South/north |
|---|
| 193 | IF ( conserve_volume_flow .AND. ( outflow_n .OR. outflow_s ) ) THEN |
|---|
| 194 | |
|---|
| 195 | volume_flow(2) = 0.0 |
|---|
| 196 | volume_flow_l(2) = 0.0 |
|---|
| 197 | |
|---|
| 198 | IF ( outflow_s ) THEN |
|---|
| 199 | j = 0 |
|---|
| 200 | ELSEIF ( outflow_n ) THEN |
|---|
| 201 | j = ny+1 |
|---|
| 202 | ENDIF |
|---|
| 203 | |
|---|
| 204 | DO i = nxl, nxr |
|---|
| 205 | ! |
|---|
| 206 | !-- Sum up the volume flow through the south/north boundary |
|---|
| 207 | DO k = nzb_2d(j,i)+1, nzt |
|---|
| 208 | volume_flow_l(2) = volume_flow_l(2) + v(k,j,i) * dzw(k) |
|---|
| 209 | ENDDO |
|---|
| 210 | ENDDO |
|---|
| 211 | |
|---|
| 212 | #if defined( __parallel ) |
|---|
| 213 | IF ( collective_wait ) CALL MPI_BARRIER( comm1dx, ierr ) |
|---|
| 214 | CALL MPI_ALLREDUCE( volume_flow_l(2), volume_flow(2), 1, MPI_REAL, & |
|---|
| 215 | MPI_SUM, comm1dx, ierr ) |
|---|
| 216 | #else |
|---|
| 217 | volume_flow = volume_flow_l |
|---|
| 218 | #endif |
|---|
| 219 | volume_flow_offset(2) = ( volume_flow_initial(2) - volume_flow(2) ) & |
|---|
| 220 | / volume_flow_area(2) |
|---|
| 221 | |
|---|
| 222 | DO i = nxlg, nxrg |
|---|
| 223 | DO k = nzb_v_inner(j,i)+1, nzt |
|---|
| 224 | v(k,j,i) = v(k,j,i) + volume_flow_offset(2) |
|---|
| 225 | ENDDO |
|---|
| 226 | ENDDO |
|---|
| 227 | |
|---|
| 228 | ENDIF |
|---|
| 229 | |
|---|
| 230 | ! |
|---|
| 231 | !-- Remove mean vertical velocity |
|---|
| 232 | IF ( ibc_p_b == 1 .AND. ibc_p_t == 1 ) THEN |
|---|
| 233 | IF ( simulated_time > 0.0 ) THEN ! otherwise nzb_w_inner not yet known |
|---|
| 234 | w_l = 0.0; w_l_l = 0.0 |
|---|
| 235 | DO i = nxl, nxr |
|---|
| 236 | DO j = nys, nyn |
|---|
| 237 | DO k = nzb_w_inner(j,i)+1, nzt |
|---|
| 238 | w_l_l(k) = w_l_l(k) + w(k,j,i) |
|---|
| 239 | ENDDO |
|---|
| 240 | ENDDO |
|---|
| 241 | ENDDO |
|---|
| 242 | #if defined( __parallel ) |
|---|
| 243 | IF ( collective_wait ) CALL MPI_BARRIER( comm2d, ierr ) |
|---|
| 244 | CALL MPI_ALLREDUCE( w_l_l(1), w_l(1), nzt, MPI_REAL, MPI_SUM, & |
|---|
| 245 | comm2d, ierr ) |
|---|
| 246 | #else |
|---|
| 247 | w_l = w_l_l |
|---|
| 248 | #endif |
|---|
| 249 | DO k = 1, nzt |
|---|
| 250 | w_l(k) = w_l(k) / ngp_2dh_outer(k,0) |
|---|
| 251 | ENDDO |
|---|
| 252 | DO i = nxlg, nxrg |
|---|
| 253 | DO j = nysg, nyng |
|---|
| 254 | DO k = nzb_w_inner(j,i)+1, nzt |
|---|
| 255 | w(k,j,i) = w(k,j,i) - w_l(k) |
|---|
| 256 | ENDDO |
|---|
| 257 | ENDDO |
|---|
| 258 | ENDDO |
|---|
| 259 | ENDIF |
|---|
| 260 | ENDIF |
|---|
| 261 | |
|---|
| 262 | ! |
|---|
| 263 | !-- Compute the divergence of the provisional velocity field. |
|---|
| 264 | CALL cpu_log( log_point_s(1), 'divergence', 'start' ) |
|---|
| 265 | |
|---|
| 266 | IF ( psolver == 'multigrid' ) THEN |
|---|
| 267 | !$OMP PARALLEL DO SCHEDULE( STATIC ) |
|---|
| 268 | DO i = nxl-1, nxr+1 |
|---|
| 269 | DO j = nys-1, nyn+1 |
|---|
| 270 | DO k = nzb, nzt+1 |
|---|
| 271 | d(k,j,i) = 0.0 |
|---|
| 272 | ENDDO |
|---|
| 273 | ENDDO |
|---|
| 274 | ENDDO |
|---|
| 275 | ELSE |
|---|
| 276 | !$OMP PARALLEL DO SCHEDULE( STATIC ) |
|---|
| 277 | DO i = nxl, nxra |
|---|
| 278 | DO j = nys, nyna |
|---|
| 279 | DO k = nzb+1, nzta |
|---|
| 280 | d(k,j,i) = 0.0 |
|---|
| 281 | ENDDO |
|---|
| 282 | ENDDO |
|---|
| 283 | ENDDO |
|---|
| 284 | ENDIF |
|---|
| 285 | |
|---|
| 286 | localsum = 0.0 |
|---|
| 287 | threadsum = 0.0 |
|---|
| 288 | |
|---|
| 289 | #if defined( __ibm ) |
|---|
| 290 | !$OMP PARALLEL PRIVATE (i,j,k) FIRSTPRIVATE(threadsum) REDUCTION(+:localsum) |
|---|
| 291 | !$OMP DO SCHEDULE( STATIC ) |
|---|
| 292 | DO i = nxl, nxr |
|---|
| 293 | DO j = nys, nyn |
|---|
| 294 | DO k = nzb_s_inner(j,i)+1, nzt |
|---|
| 295 | d(k,j,i) = ( ( u(k,j,i+1) - u(k,j,i) ) * ddx + & |
|---|
| 296 | ( v(k,j+1,i) - v(k,j,i) ) * ddy + & |
|---|
| 297 | ( w(k,j,i) - w(k-1,j,i) ) * ddzw(k) ) * ddt_3d & |
|---|
| 298 | * d_weight_pres |
|---|
| 299 | ENDDO |
|---|
| 300 | ! |
|---|
| 301 | !-- Additional pressure boundary condition at the bottom boundary for |
|---|
| 302 | !-- inhomogeneous Prandtl layer heat fluxes and temperatures, respectively |
|---|
| 303 | !-- dp/dz = -(dtau13/dx + dtau23/dy) + g*pt'/pt0. |
|---|
| 304 | !-- This condition must not be applied at the start of a run, because then |
|---|
| 305 | !-- flow_statistics has not yet been called and thus sums = 0. |
|---|
| 306 | IF ( ibc_p_b == 2 .AND. sums(nzb+1,4) /= 0.0 ) THEN |
|---|
| 307 | k = nzb_s_inner(j,i) |
|---|
| 308 | d(k+1,j,i) = d(k+1,j,i) + ( & |
|---|
| 309 | ( usws(j,i+1) - usws(j,i) ) * ddx & |
|---|
| 310 | + ( vsws(j+1,i) - vsws(j,i) ) * ddy & |
|---|
| 311 | - g * ( pt(k+1,j,i) - sums(k+1,4) ) / & |
|---|
| 312 | sums(k+1,4) & |
|---|
| 313 | ) * ddzw(k+1) * ddt_3d * d_weight_pres |
|---|
| 314 | ENDIF |
|---|
| 315 | |
|---|
| 316 | ! |
|---|
| 317 | !-- Compute possible PE-sum of divergences for flow_statistics |
|---|
| 318 | DO k = nzb_s_inner(j,i)+1, nzt |
|---|
| 319 | threadsum = threadsum + ABS( d(k,j,i) ) |
|---|
| 320 | ENDDO |
|---|
| 321 | |
|---|
| 322 | ENDDO |
|---|
| 323 | ENDDO |
|---|
| 324 | |
|---|
| 325 | localsum = localsum + threadsum * dt_3d * & |
|---|
| 326 | weight_pres(intermediate_timestep_count) |
|---|
| 327 | |
|---|
| 328 | !$OMP END PARALLEL |
|---|
| 329 | #else |
|---|
| 330 | IF ( ibc_p_b == 2 .AND. sums(nzb+1,4) /= 0.0 ) THEN |
|---|
| 331 | !$OMP PARALLEL PRIVATE (i,j,k) |
|---|
| 332 | !$OMP DO SCHEDULE( STATIC ) |
|---|
| 333 | DO i = nxl, nxr |
|---|
| 334 | DO j = nys, nyn |
|---|
| 335 | DO k = nzb_s_inner(j,i)+1, nzt |
|---|
| 336 | d(k,j,i) = ( ( u(k,j,i+1) - u(k,j,i) ) * ddx + & |
|---|
| 337 | ( v(k,j+1,i) - v(k,j,i) ) * ddy + & |
|---|
| 338 | ( w(k,j,i) - w(k-1,j,i) ) * ddzw(k) ) * ddt_3d & |
|---|
| 339 | * d_weight_pres |
|---|
| 340 | ENDDO |
|---|
| 341 | ENDDO |
|---|
| 342 | ! |
|---|
| 343 | !-- Additional pressure boundary condition at the bottom boundary for |
|---|
| 344 | !-- inhomogeneous Prandtl layer heat fluxes and temperatures, respectively |
|---|
| 345 | !-- dp/dz = -(dtau13/dx + dtau23/dy) + g*pt'/pt0. |
|---|
| 346 | !-- This condition must not be applied at the start of a run, because then |
|---|
| 347 | !-- flow_statistics has not yet been called and thus sums = 0. |
|---|
| 348 | DO j = nys, nyn |
|---|
| 349 | k = nzb_s_inner(j,i) |
|---|
| 350 | d(k+1,j,i) = d(k+1,j,i) + ( & |
|---|
| 351 | ( usws(j,i+1) - usws(j,i) ) * ddx & |
|---|
| 352 | + ( vsws(j+1,i) - vsws(j,i) ) * ddy & |
|---|
| 353 | - g * ( pt(k+1,j,i) - sums(k+1,4) ) / & |
|---|
| 354 | sums(k+1,4) & |
|---|
| 355 | ) * ddzw(k+1) * ddt_3d & |
|---|
| 356 | * d_weight_pres |
|---|
| 357 | ENDDO |
|---|
| 358 | ENDDO |
|---|
| 359 | !$OMP END PARALLEL |
|---|
| 360 | |
|---|
| 361 | ELSE |
|---|
| 362 | |
|---|
| 363 | !$OMP PARALLEL PRIVATE (i,j,k) |
|---|
| 364 | !$OMP DO SCHEDULE( STATIC ) |
|---|
| 365 | DO i = nxl, nxr |
|---|
| 366 | DO j = nys, nyn |
|---|
| 367 | DO k = nzb_s_inner(j,i)+1, nzt |
|---|
| 368 | d(k,j,i) = ( ( u(k,j,i+1) - u(k,j,i) ) * ddx + & |
|---|
| 369 | ( v(k,j+1,i) - v(k,j,i) ) * ddy + & |
|---|
| 370 | ( w(k,j,i) - w(k-1,j,i) ) * ddzw(k) ) * ddt_3d & |
|---|
| 371 | * d_weight_pres |
|---|
| 372 | ENDDO |
|---|
| 373 | ENDDO |
|---|
| 374 | ENDDO |
|---|
| 375 | !$OMP END PARALLEL |
|---|
| 376 | |
|---|
| 377 | ENDIF |
|---|
| 378 | |
|---|
| 379 | ! |
|---|
| 380 | !-- Compute possible PE-sum of divergences for flow_statistics |
|---|
| 381 | !$OMP PARALLEL PRIVATE (i,j,k) FIRSTPRIVATE(threadsum) REDUCTION(+:localsum) |
|---|
| 382 | !$OMP DO SCHEDULE( STATIC ) |
|---|
| 383 | DO i = nxl, nxr |
|---|
| 384 | DO j = nys, nyn |
|---|
| 385 | DO k = nzb+1, nzt |
|---|
| 386 | threadsum = threadsum + ABS( d(k,j,i) ) |
|---|
| 387 | ENDDO |
|---|
| 388 | ENDDO |
|---|
| 389 | ENDDO |
|---|
| 390 | localsum = localsum + threadsum * dt_3d * & |
|---|
| 391 | weight_pres(intermediate_timestep_count) |
|---|
| 392 | !$OMP END PARALLEL |
|---|
| 393 | #endif |
|---|
| 394 | |
|---|
| 395 | ! |
|---|
| 396 | !-- For completeness, set the divergence sum of all statistic regions to those |
|---|
| 397 | !-- of the total domain |
|---|
| 398 | sums_divold_l(0:statistic_regions) = localsum |
|---|
| 399 | |
|---|
| 400 | CALL cpu_log( log_point_s(1), 'divergence', 'stop' ) |
|---|
| 401 | |
|---|
| 402 | ! |
|---|
| 403 | !-- Compute the pressure perturbation solving the Poisson equation |
|---|
| 404 | IF ( psolver(1:7) == 'poisfft' ) THEN |
|---|
| 405 | |
|---|
| 406 | ! |
|---|
| 407 | !-- Enlarge the size of tend, used as a working array for the transpositions |
|---|
| 408 | IF ( nxra > nxr .OR. nyna > nyn .OR. nza > nz ) THEN |
|---|
| 409 | DEALLOCATE( tend ) |
|---|
| 410 | ALLOCATE( tend(1:nza,nys:nyna,nxl:nxra) ) |
|---|
| 411 | ENDIF |
|---|
| 412 | |
|---|
| 413 | ! |
|---|
| 414 | !-- Solve Poisson equation via FFT and solution of tridiagonal matrices |
|---|
| 415 | IF ( psolver == 'poisfft' ) THEN |
|---|
| 416 | ! |
|---|
| 417 | !-- Solver for 2d-decomposition |
|---|
| 418 | CALL poisfft( d, tend ) |
|---|
| 419 | ELSEIF ( psolver == 'poisfft_hybrid' ) THEN |
|---|
| 420 | ! |
|---|
| 421 | !-- Solver for 1d-decomposition (using MPI and OpenMP). |
|---|
| 422 | !-- The old hybrid-solver is still included here, as long as there |
|---|
| 423 | !-- are some optimization problems in poisfft |
|---|
| 424 | CALL poisfft_hybrid( d ) |
|---|
| 425 | ENDIF |
|---|
| 426 | |
|---|
| 427 | ! |
|---|
| 428 | !-- Resize tend to its normal size |
|---|
| 429 | IF ( nxra > nxr .OR. nyna > nyn .OR. nza > nz ) THEN |
|---|
| 430 | DEALLOCATE( tend ) |
|---|
| 431 | ALLOCATE( tend(nzb:nzt+1,nysg:nyng,nxlg:nxrg) ) |
|---|
| 432 | ENDIF |
|---|
| 433 | |
|---|
| 434 | ! |
|---|
| 435 | !-- Store computed perturbation pressure and set boundary condition in |
|---|
| 436 | !-- z-direction |
|---|
| 437 | !$OMP PARALLEL DO |
|---|
| 438 | DO i = nxl, nxr |
|---|
| 439 | DO j = nys, nyn |
|---|
| 440 | DO k = nzb+1, nzt |
|---|
| 441 | tend(k,j,i) = d(k,j,i) |
|---|
| 442 | ENDDO |
|---|
| 443 | ENDDO |
|---|
| 444 | ENDDO |
|---|
| 445 | |
|---|
| 446 | ! |
|---|
| 447 | !-- Bottom boundary: |
|---|
| 448 | !-- This condition is only required for internal output. The pressure |
|---|
| 449 | !-- gradient (dp(nzb+1)-dp(nzb))/dz is not used anywhere else. |
|---|
| 450 | IF ( ibc_p_b == 1 ) THEN |
|---|
| 451 | ! |
|---|
| 452 | !-- Neumann (dp/dz = 0) |
|---|
| 453 | !$OMP PARALLEL DO |
|---|
| 454 | DO i = nxlg, nxrg |
|---|
| 455 | DO j = nysg, nyng |
|---|
| 456 | tend(nzb_s_inner(j,i),j,i) = tend(nzb_s_inner(j,i)+1,j,i) |
|---|
| 457 | ENDDO |
|---|
| 458 | ENDDO |
|---|
| 459 | |
|---|
| 460 | ELSEIF ( ibc_p_b == 2 ) THEN |
|---|
| 461 | ! |
|---|
| 462 | !-- Neumann condition for inhomogeneous surfaces, |
|---|
| 463 | !-- here currently still in the form of a zero gradient. Actually |
|---|
| 464 | !-- dp/dz = -(dtau13/dx + dtau23/dy) + g*pt'/pt0 would have to be used for |
|---|
| 465 | !-- the computation (cf. above: computation of divergences). |
|---|
| 466 | !$OMP PARALLEL DO |
|---|
| 467 | DO i = nxlg, nxrg |
|---|
| 468 | DO j = nysg, nyng |
|---|
| 469 | tend(nzb_s_inner(j,i),j,i) = tend(nzb_s_inner(j,i)+1,j,i) |
|---|
| 470 | ENDDO |
|---|
| 471 | ENDDO |
|---|
| 472 | |
|---|
| 473 | ELSE |
|---|
| 474 | ! |
|---|
| 475 | !-- Dirichlet |
|---|
| 476 | !$OMP PARALLEL DO |
|---|
| 477 | DO i = nxlg, nxrg |
|---|
| 478 | DO j = nysg, nyng |
|---|
| 479 | tend(nzb_s_inner(j,i),j,i) = 0.0 |
|---|
| 480 | ENDDO |
|---|
| 481 | ENDDO |
|---|
| 482 | |
|---|
| 483 | ENDIF |
|---|
| 484 | |
|---|
| 485 | ! |
|---|
| 486 | !-- Top boundary |
|---|
| 487 | IF ( ibc_p_t == 1 ) THEN |
|---|
| 488 | ! |
|---|
| 489 | !-- Neumann |
|---|
| 490 | !$OMP PARALLEL DO |
|---|
| 491 | DO i = nxlg, nxrg |
|---|
| 492 | DO j = nysg, nyng |
|---|
| 493 | tend(nzt+1,j,i) = tend(nzt,j,i) |
|---|
| 494 | ENDDO |
|---|
| 495 | ENDDO |
|---|
| 496 | |
|---|
| 497 | ELSE |
|---|
| 498 | ! |
|---|
| 499 | !-- Dirichlet |
|---|
| 500 | !$OMP PARALLEL DO |
|---|
| 501 | DO i = nxlg, nxrg |
|---|
| 502 | DO j = nysg, nyng |
|---|
| 503 | tend(nzt+1,j,i) = 0.0 |
|---|
| 504 | ENDDO |
|---|
| 505 | ENDDO |
|---|
| 506 | |
|---|
| 507 | ENDIF |
|---|
| 508 | |
|---|
| 509 | ! |
|---|
| 510 | !-- Exchange boundaries for p |
|---|
| 511 | CALL exchange_horiz( tend, nbgp ) |
|---|
| 512 | |
|---|
| 513 | ELSEIF ( psolver == 'sor' ) THEN |
|---|
| 514 | |
|---|
| 515 | ! |
|---|
| 516 | !-- Solve Poisson equation for perturbation pressure using SOR-Red/Black |
|---|
| 517 | !-- scheme |
|---|
| 518 | CALL sor( d, ddzu_pres, ddzw, p_loc ) |
|---|
| 519 | tend = p_loc |
|---|
| 520 | |
|---|
| 521 | ELSEIF ( psolver == 'multigrid' ) THEN |
|---|
| 522 | |
|---|
| 523 | ! |
|---|
| 524 | !-- Solve Poisson equation for perturbation pressure using Multigrid scheme, |
|---|
| 525 | !-- array tend is used to store the residuals, logical exchange_mg is used |
|---|
| 526 | !-- to discern data exchange in multigrid ( 1 ghostpoint ) and normal grid |
|---|
| 527 | !-- ( nbgp ghost points ). |
|---|
| 528 | CALL poismg( tend ) |
|---|
| 529 | |
|---|
| 530 | ! |
|---|
| 531 | !-- Restore perturbation pressure on tend because this array is used |
|---|
| 532 | !-- further below to correct the velocity fields |
|---|
| 533 | DO i = nxl-1, nxr+1 |
|---|
| 534 | DO j = nys-1, nyn+1 |
|---|
| 535 | DO k = nzb, nzt+1 |
|---|
| 536 | tend(k,j,i) = p_loc(k,j,i) |
|---|
| 537 | ENDDO |
|---|
| 538 | ENDDO |
|---|
| 539 | ENDDO |
|---|
| 540 | |
|---|
| 541 | ENDIF |
|---|
| 542 | |
|---|
| 543 | ! |
|---|
| 544 | !-- Store perturbation pressure on array p, used for pressure data output. |
|---|
| 545 | !-- Ghost layers are added in the output routines (except sor-method: see below) |
|---|
| 546 | IF ( intermediate_timestep_count == 1 ) THEN |
|---|
| 547 | !$OMP PARALLEL PRIVATE (i,j,k) |
|---|
| 548 | !$OMP DO |
|---|
| 549 | DO i = nxl-1, nxr+1 |
|---|
| 550 | DO j = nys-1, nyn+1 |
|---|
| 551 | DO k = nzb, nzt+1 |
|---|
| 552 | p(k,j,i) = tend(k,j,i) * & |
|---|
| 553 | weight_substep(intermediate_timestep_count) |
|---|
| 554 | ENDDO |
|---|
| 555 | ENDDO |
|---|
| 556 | ENDDO |
|---|
| 557 | !$OMP END PARALLEL |
|---|
| 558 | |
|---|
| 559 | ELSE |
|---|
| 560 | !$OMP PARALLEL PRIVATE (i,j,k) |
|---|
| 561 | !$OMP DO |
|---|
| 562 | DO i = nxl-1, nxr+1 |
|---|
| 563 | DO j = nys-1, nyn+1 |
|---|
| 564 | DO k = nzb, nzt+1 |
|---|
| 565 | p(k,j,i) = p(k,j,i) + tend(k,j,i) * & |
|---|
| 566 | weight_substep(intermediate_timestep_count) |
|---|
| 567 | ENDDO |
|---|
| 568 | ENDDO |
|---|
| 569 | ENDDO |
|---|
| 570 | !$OMP END PARALLEL |
|---|
| 571 | |
|---|
| 572 | ENDIF |
|---|
| 573 | |
|---|
| 574 | ! |
|---|
| 575 | !-- SOR-method needs ghost layers for the next timestep |
|---|
| 576 | IF ( psolver == 'sor' ) CALL exchange_horiz( p, nbgp ) |
|---|
| 577 | |
|---|
| 578 | ! |
|---|
| 579 | !-- Correction of the provisional velocities with the current perturbation |
|---|
| 580 | !-- pressure just computed |
|---|
| 581 | IF ( conserve_volume_flow .AND. ( bc_lr_cyc .OR. bc_ns_cyc ) ) THEN |
|---|
| 582 | volume_flow_l(1) = 0.0 |
|---|
| 583 | volume_flow_l(2) = 0.0 |
|---|
| 584 | ENDIF |
|---|
| 585 | |
|---|
| 586 | !$OMP PARALLEL PRIVATE (i,j,k) |
|---|
| 587 | !$OMP DO |
|---|
| 588 | DO i = nxl, nxr |
|---|
| 589 | DO j = nys, nyn |
|---|
| 590 | DO k = nzb_w_inner(j,i)+1, nzt |
|---|
| 591 | w(k,j,i) = w(k,j,i) - dt_3d * & |
|---|
| 592 | ( tend(k+1,j,i) - tend(k,j,i) ) * ddzu(k+1) * & |
|---|
| 593 | weight_pres(intermediate_timestep_count) |
|---|
| 594 | ENDDO |
|---|
| 595 | DO k = nzb_u_inner(j,i)+1, nzt |
|---|
| 596 | u(k,j,i) = u(k,j,i) - dt_3d * & |
|---|
| 597 | ( tend(k,j,i) - tend(k,j,i-1) ) * ddx * & |
|---|
| 598 | weight_pres(intermediate_timestep_count) |
|---|
| 599 | ENDDO |
|---|
| 600 | DO k = nzb_v_inner(j,i)+1, nzt |
|---|
| 601 | v(k,j,i) = v(k,j,i) - dt_3d * & |
|---|
| 602 | ( tend(k,j,i) - tend(k,j-1,i) ) * ddy * & |
|---|
| 603 | weight_pres(intermediate_timestep_count) |
|---|
| 604 | ENDDO |
|---|
| 605 | ! |
|---|
| 606 | !-- Sum up the volume flow through the right and north boundary |
|---|
| 607 | IF ( conserve_volume_flow .AND. bc_lr_cyc .AND. bc_ns_cyc .AND. & |
|---|
| 608 | i == nx ) THEN |
|---|
| 609 | !$OMP CRITICAL |
|---|
| 610 | DO k = nzb_2d(j,i) + 1, nzt |
|---|
| 611 | volume_flow_l(1) = volume_flow_l(1) + u(k,j,i) * dzw(k) |
|---|
| 612 | ENDDO |
|---|
| 613 | !$OMP END CRITICAL |
|---|
| 614 | ENDIF |
|---|
| 615 | IF ( conserve_volume_flow .AND. bc_ns_cyc .AND. bc_lr_cyc .AND. & |
|---|
| 616 | j == ny ) THEN |
|---|
| 617 | !$OMP CRITICAL |
|---|
| 618 | DO k = nzb_2d(j,i) + 1, nzt |
|---|
| 619 | volume_flow_l(2) = volume_flow_l(2) + v(k,j,i) * dzw(k) |
|---|
| 620 | ENDDO |
|---|
| 621 | !$OMP END CRITICAL |
|---|
| 622 | ENDIF |
|---|
| 623 | |
|---|
| 624 | ENDDO |
|---|
| 625 | ENDDO |
|---|
| 626 | !$OMP END PARALLEL |
|---|
| 627 | |
|---|
| 628 | ! |
|---|
| 629 | !-- Conserve the volume flow |
|---|
| 630 | IF ( conserve_volume_flow .AND. ( bc_lr_cyc .AND. bc_ns_cyc ) ) THEN |
|---|
| 631 | |
|---|
| 632 | #if defined( __parallel ) |
|---|
| 633 | IF ( collective_wait ) CALL MPI_BARRIER( comm2d, ierr ) |
|---|
| 634 | CALL MPI_ALLREDUCE( volume_flow_l(1), volume_flow(1), 2, MPI_REAL, & |
|---|
| 635 | MPI_SUM, comm2d, ierr ) |
|---|
| 636 | #else |
|---|
| 637 | volume_flow = volume_flow_l |
|---|
| 638 | #endif |
|---|
| 639 | |
|---|
| 640 | volume_flow_offset = ( volume_flow_initial - volume_flow ) / & |
|---|
| 641 | volume_flow_area |
|---|
| 642 | |
|---|
| 643 | !$OMP PARALLEL PRIVATE (i,j,k) |
|---|
| 644 | !$OMP DO |
|---|
| 645 | DO i = nxl, nxr |
|---|
| 646 | DO j = nys, nyn |
|---|
| 647 | DO k = nzb_u_inner(j,i) + 1, nzt |
|---|
| 648 | u(k,j,i) = u(k,j,i) + volume_flow_offset(1) |
|---|
| 649 | ENDDO |
|---|
| 650 | DO k = nzb_v_inner(j,i) + 1, nzt |
|---|
| 651 | v(k,j,i) = v(k,j,i) + volume_flow_offset(2) |
|---|
| 652 | ENDDO |
|---|
| 653 | ENDDO |
|---|
| 654 | ENDDO |
|---|
| 655 | |
|---|
| 656 | !$OMP END PARALLEL |
|---|
| 657 | |
|---|
| 658 | ENDIF |
|---|
| 659 | |
|---|
| 660 | ! |
|---|
| 661 | !-- Exchange of boundaries for the velocities |
|---|
| 662 | CALL exchange_horiz( u, nbgp ) |
|---|
| 663 | CALL exchange_horiz( v, nbgp ) |
|---|
| 664 | CALL exchange_horiz( w, nbgp ) |
|---|
| 665 | |
|---|
| 666 | ! |
|---|
| 667 | !-- Compute the divergence of the corrected velocity field, |
|---|
| 668 | !-- a possible PE-sum is computed in flow_statistics |
|---|
| 669 | CALL cpu_log( log_point_s(1), 'divergence', 'start' ) |
|---|
| 670 | sums_divnew_l = 0.0 |
|---|
| 671 | |
|---|
| 672 | ! |
|---|
| 673 | !-- d must be reset to zero because it can contain nonzero values below the |
|---|
| 674 | !-- topography |
|---|
| 675 | IF ( topography /= 'flat' ) d = 0.0 |
|---|
| 676 | |
|---|
| 677 | localsum = 0.0 |
|---|
| 678 | threadsum = 0.0 |
|---|
| 679 | |
|---|
| 680 | !$OMP PARALLEL PRIVATE (i,j,k) FIRSTPRIVATE(threadsum) REDUCTION(+:localsum) |
|---|
| 681 | !$OMP DO SCHEDULE( STATIC ) |
|---|
| 682 | #if defined( __ibm ) |
|---|
| 683 | DO i = nxl, nxr |
|---|
| 684 | DO j = nys, nyn |
|---|
| 685 | DO k = nzb_s_inner(j,i)+1, nzt |
|---|
| 686 | d(k,j,i) = ( u(k,j,i+1) - u(k,j,i) ) * ddx + & |
|---|
| 687 | ( v(k,j+1,i) - v(k,j,i) ) * ddy + & |
|---|
| 688 | ( w(k,j,i) - w(k-1,j,i) ) * ddzw(k) |
|---|
| 689 | ENDDO |
|---|
| 690 | DO k = nzb+1, nzt |
|---|
| 691 | threadsum = threadsum + ABS( d(k,j,i) ) |
|---|
| 692 | ENDDO |
|---|
| 693 | ENDDO |
|---|
| 694 | ENDDO |
|---|
| 695 | #else |
|---|
| 696 | DO i = nxl, nxr |
|---|
| 697 | DO j = nys, nyn |
|---|
| 698 | DO k = nzb_s_inner(j,i)+1, nzt |
|---|
| 699 | d(k,j,i) = ( u(k,j,i+1) - u(k,j,i) ) * ddx + & |
|---|
| 700 | ( v(k,j+1,i) - v(k,j,i) ) * ddy + & |
|---|
| 701 | ( w(k,j,i) - w(k-1,j,i) ) * ddzw(k) |
|---|
| 702 | threadsum = threadsum + ABS( d(k,j,i) ) |
|---|
| 703 | ENDDO |
|---|
| 704 | ENDDO |
|---|
| 705 | ENDDO |
|---|
| 706 | #endif |
|---|
| 707 | |
|---|
| 708 | localsum = localsum + threadsum |
|---|
| 709 | !$OMP END PARALLEL |
|---|
| 710 | |
|---|
| 711 | ! |
|---|
| 712 | !-- For completeness, set the divergence sum of all statistic regions to those |
|---|
| 713 | !-- of the total domain |
|---|
| 714 | sums_divnew_l(0:statistic_regions) = localsum |
|---|
| 715 | |
|---|
| 716 | CALL cpu_log( log_point_s(1), 'divergence', 'stop' ) |
|---|
| 717 | |
|---|
| 718 | CALL cpu_log( log_point(8), 'pres', 'stop' ) |
|---|
| 719 | |
|---|
| 720 | |
|---|
| 721 | |
|---|
| 722 | END SUBROUTINE pres |
|---|
