Home

Context Navigation

source: palm/trunk/SOURCE/pres.f90 @ 1918

Last change on this file since 1918 was 1918, checked in by raasch, 8 years ago
bugfixes for calculating run control quantities, bugfix for calculating pressure with fft-method in case of Neumann conditions both at bottom and top, steering of pres modified, ocean mode now uses initial density profile as reference in the buoyancy term
Property svn:keywords set to `Id`
File size: 26.5 KB

Rev	Line
[1682]	1	!> @file pres.f90
[1036]	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	!
[1818]	16	! Copyright 1997-2016 Leibniz Universitaet Hannover
[1036]	17	!--------------------------------------------------------------------------------!
	18	!
[484]	19	! Current revisions:
[1212]	20	! ------------------
[1918]	21	! sum of divergence is also calculated when pres is called before the initial
	22	! first time step,
	23	! stearing is modified by using intermediate_timestep_count = 0 in order to
	24	! determine if pres is called before the first initial timestep,
	25	! bugfix: in case of Neumann conditions for pressure both at bottom and top,
	26	! mean vertical velocity is also removed for the first time step
	27	! bugfix for calculating divergences
[1683]	28	!
[1321]	29	! Former revisions:
	30	! -----------------
	31	! $Id: pres.f90 1918 2016-05-27 14:35:57Z raasch $
	32	!
[1909]	33	! 1908 2016-05-25 17:22:32Z suehring
	34	! New divergence for output into RUN_CONTROL file is calculated only at last
	35	! Runge-Kutta step
	36	!
[1846]	37	! 1845 2016-04-08 08:29:13Z raasch
	38	! nzb_2d replace by nzb_u\|v_inner
	39	!
[1800]	40	! 1799 2016-04-05 08:35:55Z gronemeier
	41	! Bugfix: excluded third dimension from horizontal volume flow calculation
	42	!
[1763]	43	! 1762 2016-02-25 12:31:13Z hellstea
	44	! Introduction of nested domain feature
	45	!
[1683]	46	! 1682 2015-10-07 23:56:08Z knoop
	47	! Code annotations made doxygen readable
	48	!
[1576]	49	! 1575 2015-03-27 09:56:27Z raasch
	50	! poismg_fast + respective module added, adjustments for psolver-queries
	51	!
[1343]	52	! 1342 2014-03-26 17:04:47Z kanani
	53	! REAL constants defined as wp-kind
	54	!
[1321]	55	! 1320 2014-03-20 08:40:49Z raasch
[1320]	56	! ONLY-attribute added to USE-statements,
	57	! kind-parameters added to all INTEGER and REAL declaration statements,
	58	! kinds are defined in new module kinds,
	59	! old module precision_kind is removed,
	60	! revision history before 2012 removed,
	61	! comment fields (!:) to be used for variable explanations added to
	62	! all variable declaration statements
[708]	63	!
[1319]	64	! 1318 2014-03-17 13:35:16Z raasch
	65	! module interfaces removed
	66	!
[1307]	67	! 1306 2014-03-13 14:30:59Z raasch
	68	! second argument removed from routine poisfft
	69	!
[1258]	70	! 1257 2013-11-08 15:18:40Z raasch
	71	! openacc loop and loop vector clauses removed, independent clauses added,
	72	! end parallel replaced by end parallel loop
	73	!
[1222]	74	! 1221 2013-09-10 08:59:13Z raasch
	75	! openACC porting of reduction operations, loops for calculating d are
	76	! using the rflags_s_inner multiply flag instead of the nzb_s_inner loop index
	77	!
[1213]	78	! 1212 2013-08-15 08:46:27Z raasch
	79	! call of poisfft_hybrid removed
	80	!
[1118]	81	! 1117 2013-03-27 11:15:36Z suehring
	82	! Bugfix in OpenMP parallelization.
	83	!
[1114]	84	! 1113 2013-03-10 02:48:14Z raasch
	85	! GPU-porting of several loops, some loops rearranged
	86	!
[1112]	87	! 1111 2013-03-08 23:54:10Z
	88	! openACC statements added,
	89	! ibc_p_b = 2 removed
	90	!
[1093]	91	! 1092 2013-02-02 11:24:22Z raasch
	92	! unused variables removed
	93	!
[1037]	94	! 1036 2012-10-22 13:43:42Z raasch
	95	! code put under GPL (PALM 3.9)
	96	!
[1004]	97	! 1003 2012-09-14 14:35:53Z raasch
	98	! adjustment of array tend for cases with unequal subdomain sizes removed
	99	!
[1]	100	! Revision 1.1 1997/07/24 11:24:44 raasch
	101	! Initial revision
	102	!
	103	!
	104	! Description:
	105	! ------------
[1682]	106	!> Compute the divergence of the provisional velocity field. Solve the Poisson
	107	!> equation for the perturbation pressure. Compute the final velocities using
	108	!> this perturbation pressure. Compute the remaining divergence.
[1]	109	!------------------------------------------------------------------------------!
[1682]	110	SUBROUTINE pres
	111
[1]	112
[1320]	113	USE arrays_3d, &
	114	ONLY: d, ddzu, ddzu_pres, ddzw, dzw, p, p_loc, tend, u, v, w
	115
	116	USE control_parameters, &
	117	ONLY: bc_lr_cyc, bc_ns_cyc, conserve_volume_flow, dt_3d, &
	118	gathered_size, ibc_p_b, ibc_p_t, intermediate_timestep_count, &
[1908]	119	intermediate_timestep_count_max, mg_switch_to_pe0_level, &
	120	nest_domain, nest_bound_l, nest_bound_n, nest_bound_r, &
	121	nest_bound_s, on_device, outflow_l, outflow_n, outflow_r, &
[1918]	122	outflow_s, psolver, subdomain_size, topography, volume_flow, &
	123	volume_flow_area, volume_flow_initial
[1320]	124
	125	USE cpulog, &
	126	ONLY: cpu_log, log_point, log_point_s
	127
	128	USE grid_variables, &
	129	ONLY: ddx, ddy
	130
	131	USE indices, &
	132	ONLY: nbgp, ngp_2dh_outer, nx, nxl, nxlg, nxl_mg, nxr, nxrg, nxr_mg, &
	133	ny, nys, nysg, nys_mg, nyn, nyng, nyn_mg, nzb, nzb_s_inner, &
[1845]	134	nzb_u_inner, nzb_v_inner, nzb_w_inner, nzt, nzt_mg, &
[1320]	135	rflags_s_inner
	136
	137	USE kinds
	138
[1]	139	USE pegrid
	140
[1320]	141	USE poisfft_mod, &
	142	ONLY: poisfft
	143
[1575]	144	USE poismg_mod
	145
[1320]	146	USE statistics, &
	147	ONLY: statistic_regions, sums_divnew_l, sums_divold_l, weight_pres, &
	148	weight_substep
	149
[1]	150	IMPLICIT NONE
	151
[1682]	152	INTEGER(iwp) :: i !<
	153	INTEGER(iwp) :: j !<
	154	INTEGER(iwp) :: k !<
[1]	155
[1682]	156	REAL(wp) :: ddt_3d !<
[1918]	157	REAL(wp) :: d_weight_pres !<
[1682]	158	REAL(wp) :: localsum !<
	159	REAL(wp) :: threadsum !<
[1918]	160	REAL(wp) :: weight_pres_l !<
[1]	161
[1762]	162	REAL(wp), DIMENSION(1:3) :: volume_flow_l !<
	163	REAL(wp), DIMENSION(1:3) :: volume_flow_offset !<
[1682]	164	REAL(wp), DIMENSION(1:nzt) :: w_l !<
	165	REAL(wp), DIMENSION(1:nzt) :: w_l_l !<
[1]	166
	167
	168	CALL cpu_log( log_point(8), 'pres', 'start' )
	169
[85]	170
[1918]	171	!
	172	!-- Calculate quantities to be used locally
[1342]	173	ddt_3d = 1.0_wp / dt_3d
[1918]	174	IF ( intermediate_timestep_count == 0 ) THEN
	175	!
	176	!-- If pres is called before initial time step
	177	weight_pres_l = 1.0_wp
	178	d_weight_pres = 1.0_wp
	179	ELSE
	180	weight_pres_l = weight_pres(intermediate_timestep_count)
	181	d_weight_pres = 1.0_wp / weight_pres(intermediate_timestep_count)
	182	ENDIF
[85]	183
[1]	184	!
[707]	185	!-- Multigrid method expects array d to have one ghost layer.
	186	!--
[1575]	187	IF ( psolver(1:9) == 'multigrid' ) THEN
[667]	188
[1]	189	DEALLOCATE( d )
[667]	190	ALLOCATE( d(nzb:nzt+1,nys-1:nyn+1,nxl-1:nxr+1) )
[707]	191
	192	!
	193	!-- Since p is later used to hold the weighted average of the substeps, it
	194	!-- cannot be used in the iterative solver. Therefore, its initial value is
	195	!-- stored on p_loc, which is then iteratively advanced in every substep.
[1762]	196	IF ( intermediate_timestep_count <= 1 ) THEN
[707]	197	DO i = nxl-1, nxr+1
	198	DO j = nys-1, nyn+1
	199	DO k = nzb, nzt+1
	200	p_loc(k,j,i) = p(k,j,i)
	201	ENDDO
	202	ENDDO
	203	ENDDO
[667]	204	ENDIF
	205
[1918]	206	ELSEIF ( psolver == 'sor' .AND. intermediate_timestep_count <= 1 ) THEN
[707]	207
	208	!
	209	!-- Since p is later used to hold the weighted average of the substeps, it
	210	!-- cannot be used in the iterative solver. Therefore, its initial value is
	211	!-- stored on p_loc, which is then iteratively advanced in every substep.
	212	p_loc = p
	213
[1]	214	ENDIF
	215
	216	!
[75]	217	!-- Conserve the volume flow at the outflow in case of non-cyclic lateral
	218	!-- boundary conditions
[106]	219	!-- WARNING: so far, this conservation does not work at the left/south
	220	!-- boundary if the topography at the inflow differs from that at the
	221	!-- outflow! For this case, volume_flow_area needs adjustment!
	222	!
	223	!-- Left/right
[709]	224	IF ( conserve_volume_flow .AND. ( outflow_l .OR. outflow_r ) ) THEN
[680]	225
[1342]	226	volume_flow(1) = 0.0_wp
	227	volume_flow_l(1) = 0.0_wp
[106]	228
	229	IF ( outflow_l ) THEN
	230	i = 0
	231	ELSEIF ( outflow_r ) THEN
	232	i = nx+1
	233	ENDIF
	234
	235	DO j = nys, nyn
	236	!
	237	!-- Sum up the volume flow through the south/north boundary
[1845]	238	DO k = nzb_u_inner(j,i)+1, nzt
[667]	239	volume_flow_l(1) = volume_flow_l(1) + u(k,j,i) * dzw(k)
[106]	240	ENDDO
	241	ENDDO
	242
	243	#if defined( __parallel )
[680]	244	IF ( collective_wait ) CALL MPI_BARRIER( comm1dy, ierr )
[106]	245	CALL MPI_ALLREDUCE( volume_flow_l(1), volume_flow(1), 1, MPI_REAL, &
	246	MPI_SUM, comm1dy, ierr )
	247	#else
	248	volume_flow = volume_flow_l
	249	#endif
[709]	250	volume_flow_offset(1) = ( volume_flow_initial(1) - volume_flow(1) ) &
[106]	251	/ volume_flow_area(1)
	252
[667]	253	DO j = nysg, nyng
[1845]	254	DO k = nzb_u_inner(j,i)+1, nzt
[106]	255	u(k,j,i) = u(k,j,i) + volume_flow_offset(1)
	256	ENDDO
	257	ENDDO
	258
	259	ENDIF
	260
	261	!
	262	!-- South/north
[709]	263	IF ( conserve_volume_flow .AND. ( outflow_n .OR. outflow_s ) ) THEN
[106]	264
[1342]	265	volume_flow(2) = 0.0_wp
	266	volume_flow_l(2) = 0.0_wp
[75]	267
[106]	268	IF ( outflow_s ) THEN
	269	j = 0
	270	ELSEIF ( outflow_n ) THEN
[75]	271	j = ny+1
[106]	272	ENDIF
	273
	274	DO i = nxl, nxr
[75]	275	!
[106]	276	!-- Sum up the volume flow through the south/north boundary
[1845]	277	DO k = nzb_v_inner(j,i)+1, nzt
[667]	278	volume_flow_l(2) = volume_flow_l(2) + v(k,j,i) * dzw(k)
[75]	279	ENDDO
[106]	280	ENDDO
	281
[75]	282	#if defined( __parallel )
[680]	283	IF ( collective_wait ) CALL MPI_BARRIER( comm1dx, ierr )
[75]	284	CALL MPI_ALLREDUCE( volume_flow_l(2), volume_flow(2), 1, MPI_REAL, &
	285	MPI_SUM, comm1dx, ierr )
	286	#else
	287	volume_flow = volume_flow_l
	288	#endif
	289	volume_flow_offset(2) = ( volume_flow_initial(2) - volume_flow(2) ) &
[106]	290	/ volume_flow_area(2)
[75]	291
[667]	292	DO i = nxlg, nxrg
[709]	293	DO k = nzb_v_inner(j,i)+1, nzt
[106]	294	v(k,j,i) = v(k,j,i) + volume_flow_offset(2)
[75]	295	ENDDO
[106]	296	ENDDO
[75]	297
	298	ENDIF
	299
[76]	300	!
[1762]	301	!-- Remove mean vertical velocity in case that Neumann conditions are
[1918]	302	!-- used both at bottom and top boundary, and if not a nested domain.
	303	!-- This cannot be done before the first initial time step because ngp_2dh_outer
	304	!-- is not yet known then.
	305	IF ( ibc_p_b == 1 .AND. ibc_p_t == 1 .AND. .NOT. nest_domain .AND. &
	306	intermediate_timestep_count /= 0 ) &
	307	THEN
	308	w_l = 0.0_wp; w_l_l = 0.0_wp
	309	DO i = nxl, nxr
	310	DO j = nys, nyn
	311	DO k = nzb_w_inner(j,i)+1, nzt
	312	w_l_l(k) = w_l_l(k) + w(k,j,i)
[76]	313	ENDDO
	314	ENDDO
[1918]	315	ENDDO
[76]	316	#if defined( __parallel )
[1918]	317	IF ( collective_wait ) CALL MPI_BARRIER( comm2d, ierr )
	318	CALL MPI_ALLREDUCE( w_l_l(1), w_l(1), nzt, MPI_REAL, MPI_SUM, &
	319	comm2d, ierr )
[76]	320	#else
[1918]	321	w_l = w_l_l
[76]	322	#endif
[1918]	323	DO k = 1, nzt
	324	w_l(k) = w_l(k) / ngp_2dh_outer(k,0)
	325	ENDDO
	326	DO i = nxlg, nxrg
	327	DO j = nysg, nyng
	328	DO k = nzb_w_inner(j,i)+1, nzt
	329	w(k,j,i) = w(k,j,i) - w_l(k)
[76]	330	ENDDO
	331	ENDDO
[1918]	332	ENDDO
[76]	333	ENDIF
[75]	334
	335	!
[1]	336	!-- Compute the divergence of the provisional velocity field.
	337	CALL cpu_log( log_point_s(1), 'divergence', 'start' )
	338
[1575]	339	IF ( psolver(1:9) == 'multigrid' ) THEN
[1]	340	!$OMP PARALLEL DO SCHEDULE( STATIC )
	341	DO i = nxl-1, nxr+1
	342	DO j = nys-1, nyn+1
	343	DO k = nzb, nzt+1
[1342]	344	d(k,j,i) = 0.0_wp
[1]	345	ENDDO
	346	ENDDO
	347	ENDDO
	348	ELSE
	349	!$OMP PARALLEL DO SCHEDULE( STATIC )
[1111]	350	!$acc kernels present( d )
[1003]	351	DO i = nxl, nxr
	352	DO j = nys, nyn
	353	DO k = nzb+1, nzt
[1342]	354	d(k,j,i) = 0.0_wp
[1]	355	ENDDO
	356	ENDDO
	357	ENDDO
[1111]	358	!$acc end kernels
[1]	359	ENDIF
	360
[1342]	361	localsum = 0.0_wp
	362	threadsum = 0.0_wp
[1]	363
	364	#if defined( __ibm )
	365	!$OMP PARALLEL PRIVATE (i,j,k) FIRSTPRIVATE(threadsum) REDUCTION(+:localsum)
	366	!$OMP DO SCHEDULE( STATIC )
	367	DO i = nxl, nxr
	368	DO j = nys, nyn
	369	DO k = nzb_s_inner(j,i)+1, nzt
[85]	370	d(k,j,i) = ( ( u(k,j,i+1) - u(k,j,i) ) * ddx + &
	371	( v(k,j+1,i) - v(k,j,i) ) * ddy + &
[673]	372	( w(k,j,i) - w(k-1,j,i) ) * ddzw(k) ) * ddt_3d &
	373	* d_weight_pres
[1]	374	ENDDO
	375	!
	376	!-- Compute possible PE-sum of divergences for flow_statistics
	377	DO k = nzb_s_inner(j,i)+1, nzt
	378	threadsum = threadsum + ABS( d(k,j,i) )
	379	ENDDO
	380
	381	ENDDO
	382	ENDDO
	383
[1918]	384	IF ( intermediate_timestep_count == intermediate_timestep_count_max .OR. &
	385	intermediate_timestep_count == 0 ) THEN
	386	localsum = localsum + threadsum * dt_3d * weight_pres_l
[1908]	387	ENDIF
[1]	388	!$OMP END PARALLEL
	389	#else
	390
[1111]	391	!$OMP PARALLEL PRIVATE (i,j,k)
	392	!$OMP DO SCHEDULE( STATIC )
[1221]	393	!$acc kernels present( d, ddzw, rflags_s_inner, u, v, w )
	394	!$acc loop collapse( 3 )
[1111]	395	DO i = nxl, nxr
	396	DO j = nys, nyn
	397	DO k = 1, nzt
[1221]	398	d(k,j,i) = ( ( u(k,j,i+1) - u(k,j,i) ) * ddx + &
	399	( v(k,j+1,i) - v(k,j,i) ) * ddy + &
	400	( w(k,j,i) - w(k-1,j,i) ) * ddzw(k) ) * ddt_3d &
	401	* d_weight_pres * rflags_s_inner(k,j,i)
[1]	402	ENDDO
	403	ENDDO
[1111]	404	ENDDO
	405	!$acc end kernels
	406	!$OMP END PARALLEL
[1]	407
	408	!
[1908]	409	!-- Compute possible PE-sum of divergences for flow_statistics. Carry out
	410	!-- computation only at last Runge-Kutta substep.
[1918]	411	IF ( intermediate_timestep_count == intermediate_timestep_count_max .OR. &
	412	intermediate_timestep_count == 0 ) THEN
[1908]	413	!$OMP PARALLEL PRIVATE (i,j,k) FIRSTPRIVATE(threadsum) REDUCTION(+:localsum)
	414	!$OMP DO SCHEDULE( STATIC )
	415	!$acc parallel loop collapse(3) present( d ) reduction(+:threadsum)
	416	DO i = nxl, nxr
	417	DO j = nys, nyn
	418	DO k = nzb+1, nzt
	419	threadsum = threadsum + ABS( d(k,j,i) )
	420	ENDDO
[1]	421	ENDDO
	422	ENDDO
[1908]	423	!$acc end parallel loop
[1918]	424	localsum = localsum + threadsum * dt_3d * weight_pres_l
[1908]	425	!$OMP END PARALLEL
	426	ENDIF
[1]	427	#endif
	428
	429	!
	430	!-- For completeness, set the divergence sum of all statistic regions to those
	431	!-- of the total domain
[1918]	432	IF ( intermediate_timestep_count == intermediate_timestep_count_max .OR. &
	433	intermediate_timestep_count == 0 ) THEN
[1908]	434	sums_divold_l(0:statistic_regions) = localsum
[1918]	435	ENDIF
[1]	436
	437	CALL cpu_log( log_point_s(1), 'divergence', 'stop' )
	438
	439	!
	440	!-- Compute the pressure perturbation solving the Poisson equation
[1575]	441	IF ( psolver == 'poisfft' ) THEN
[1]	442
	443	!
	444	!-- Solve Poisson equation via FFT and solution of tridiagonal matrices
[1575]	445	CALL poisfft( d )
[1212]	446
[1]	447	!
	448	!-- Store computed perturbation pressure and set boundary condition in
	449	!-- z-direction
	450	!$OMP PARALLEL DO
[1113]	451	!$acc kernels present( d, tend )
[1]	452	DO i = nxl, nxr
	453	DO j = nys, nyn
	454	DO k = nzb+1, nzt
	455	tend(k,j,i) = d(k,j,i)
	456	ENDDO
	457	ENDDO
	458	ENDDO
[1113]	459	!$acc end kernels
[1]	460
	461	!
	462	!-- Bottom boundary:
	463	!-- This condition is only required for internal output. The pressure
	464	!-- gradient (dp(nzb+1)-dp(nzb))/dz is not used anywhere else.
	465	IF ( ibc_p_b == 1 ) THEN
	466	!
	467	!-- Neumann (dp/dz = 0)
	468	!$OMP PARALLEL DO
[1113]	469	!$acc kernels present( nzb_s_inner, tend )
[667]	470	DO i = nxlg, nxrg
	471	DO j = nysg, nyng
[1]	472	tend(nzb_s_inner(j,i),j,i) = tend(nzb_s_inner(j,i)+1,j,i)
	473	ENDDO
	474	ENDDO
[1113]	475	!$acc end kernels
[1]	476
	477	ELSE
	478	!
	479	!-- Dirichlet
	480	!$OMP PARALLEL DO
[1113]	481	!$acc kernels present( tend )
[667]	482	DO i = nxlg, nxrg
	483	DO j = nysg, nyng
[1342]	484	tend(nzb_s_inner(j,i),j,i) = 0.0_wp
[1]	485	ENDDO
	486	ENDDO
[1113]	487	!$acc end kernels
[1]	488
	489	ENDIF
	490
	491	!
	492	!-- Top boundary
	493	IF ( ibc_p_t == 1 ) THEN
	494	!
	495	!-- Neumann
	496	!$OMP PARALLEL DO
[1113]	497	!$acc kernels present( tend )
[667]	498	DO i = nxlg, nxrg
	499	DO j = nysg, nyng
[1]	500	tend(nzt+1,j,i) = tend(nzt,j,i)
	501	ENDDO
	502	ENDDO
[1113]	503	!$acc end kernels
[1]	504
	505	ELSE
	506	!
	507	!-- Dirichlet
	508	!$OMP PARALLEL DO
[1113]	509	!$acc kernels present( tend )
[667]	510	DO i = nxlg, nxrg
	511	DO j = nysg, nyng
[1342]	512	tend(nzt+1,j,i) = 0.0_wp
[1]	513	ENDDO
	514	ENDDO
[1113]	515	!$acc end kernels
[1]	516
	517	ENDIF
	518
	519	!
	520	!-- Exchange boundaries for p
[1113]	521	IF ( numprocs == 1 ) THEN ! workaround for single-core GPU runs
	522	on_device = .TRUE. ! to be removed after complete porting
	523	ELSE ! of ghost point exchange
	524	!$acc update host( tend )
	525	ENDIF
[667]	526	CALL exchange_horiz( tend, nbgp )
[1113]	527	IF ( numprocs == 1 ) THEN ! workaround for single-core GPU runs
	528	on_device = .FALSE. ! to be removed after complete porting
	529	ELSE ! of ghost point exchange
	530	!$acc update device( tend )
	531	ENDIF
[1]	532
	533	ELSEIF ( psolver == 'sor' ) THEN
	534
	535	!
	536	!-- Solve Poisson equation for perturbation pressure using SOR-Red/Black
	537	!-- scheme
[707]	538	CALL sor( d, ddzu_pres, ddzw, p_loc )
	539	tend = p_loc
[1]	540
[1575]	541	ELSEIF ( psolver(1:9) == 'multigrid' ) THEN
[1]	542
	543	!
	544	!-- Solve Poisson equation for perturbation pressure using Multigrid scheme,
[667]	545	!-- array tend is used to store the residuals, logical exchange_mg is used
	546	!-- to discern data exchange in multigrid ( 1 ghostpoint ) and normal grid
	547	!-- ( nbgp ghost points ).
[778]	548
	549	!-- If the number of grid points of the gathered grid, which is collected
	550	!-- on PE0, is larger than the number of grid points of an PE, than array
	551	!-- tend will be enlarged.
	552	IF ( gathered_size > subdomain_size ) THEN
	553	DEALLOCATE( tend )
	554	ALLOCATE( tend(nzb:nzt_mg(mg_switch_to_pe0_level)+1,nys_mg( &
	555	mg_switch_to_pe0_level)-1:nyn_mg(mg_switch_to_pe0_level)+1,&
	556	nxl_mg(mg_switch_to_pe0_level)-1:nxr_mg( &
	557	mg_switch_to_pe0_level)+1) )
	558	ENDIF
	559
[1575]	560	IF ( psolver == 'multigrid' ) THEN
	561	CALL poismg( tend )
	562	ELSE
	563	CALL poismg_fast( tend )
	564	ENDIF
[707]	565
[778]	566	IF ( gathered_size > subdomain_size ) THEN
	567	DEALLOCATE( tend )
	568	ALLOCATE( tend(nzb:nzt+1,nysg:nyng,nxlg:nxrg) )
	569	ENDIF
	570
[1]	571	!
	572	!-- Restore perturbation pressure on tend because this array is used
	573	!-- further below to correct the velocity fields
[707]	574	DO i = nxl-1, nxr+1
	575	DO j = nys-1, nyn+1
	576	DO k = nzb, nzt+1
	577	tend(k,j,i) = p_loc(k,j,i)
	578	ENDDO
	579	ENDDO
	580	ENDDO
[667]	581
[1]	582	ENDIF
	583
	584	!
[707]	585	!-- Store perturbation pressure on array p, used for pressure data output.
	586	!-- Ghost layers are added in the output routines (except sor-method: see below)
[1918]	587	IF ( intermediate_timestep_count <= 1 ) THEN
[707]	588	!$OMP PARALLEL PRIVATE (i,j,k)
	589	!$OMP DO
[1113]	590	!$acc kernels present( p, tend, weight_substep )
[1257]	591	!$acc loop independent
[707]	592	DO i = nxl-1, nxr+1
[1257]	593	!$acc loop independent
[707]	594	DO j = nys-1, nyn+1
[1257]	595	!$acc loop independent
[707]	596	DO k = nzb, nzt+1
	597	p(k,j,i) = tend(k,j,i) * &
	598	weight_substep(intermediate_timestep_count)
[673]	599	ENDDO
[1]	600	ENDDO
[707]	601	ENDDO
[1113]	602	!$acc end kernels
[707]	603	!$OMP END PARALLEL
	604
[1762]	605	ELSEIF ( intermediate_timestep_count > 1 ) THEN
[707]	606	!$OMP PARALLEL PRIVATE (i,j,k)
	607	!$OMP DO
[1113]	608	!$acc kernels present( p, tend, weight_substep )
[1257]	609	!$acc loop independent
[707]	610	DO i = nxl-1, nxr+1
[1257]	611	!$acc loop independent
[707]	612	DO j = nys-1, nyn+1
[1257]	613	!$acc loop independent
[707]	614	DO k = nzb, nzt+1
	615	p(k,j,i) = p(k,j,i) + tend(k,j,i) * &
	616	weight_substep(intermediate_timestep_count)
[673]	617	ENDDO
	618	ENDDO
[707]	619	ENDDO
[1113]	620	!$acc end kernels
[707]	621	!$OMP END PARALLEL
	622
	623	ENDIF
[673]	624
[707]	625	!
	626	!-- SOR-method needs ghost layers for the next timestep
	627	IF ( psolver == 'sor' ) CALL exchange_horiz( p, nbgp )
[682]	628
[1]	629	!
	630	!-- Correction of the provisional velocities with the current perturbation
	631	!-- pressure just computed
[709]	632	IF ( conserve_volume_flow .AND. ( bc_lr_cyc .OR. bc_ns_cyc ) ) THEN
[1342]	633	volume_flow_l(1) = 0.0_wp
	634	volume_flow_l(2) = 0.0_wp
[1]	635	ENDIF
[707]	636
[1]	637	!$OMP PARALLEL PRIVATE (i,j,k)
	638	!$OMP DO
[1918]	639	!$acc kernels present( ddzu, nzb_u_inner, nzb_v_inner, nzb_w_inner, tend, u, v, w )
[1257]	640	!$acc loop independent
[673]	641	DO i = nxl, nxr
[1257]	642	!$acc loop independent
[1]	643	DO j = nys, nyn
[1257]	644	!$acc loop independent
[1113]	645	DO k = 1, nzt
	646	IF ( k > nzb_w_inner(j,i) ) THEN
	647	w(k,j,i) = w(k,j,i) - dt_3d * &
	648	( tend(k+1,j,i) - tend(k,j,i) ) * ddzu(k+1) * &
[1918]	649	weight_pres_l
[1113]	650	ENDIF
[1]	651	ENDDO
[1257]	652	!$acc loop independent
[1113]	653	DO k = 1, nzt
	654	IF ( k > nzb_u_inner(j,i) ) THEN
	655	u(k,j,i) = u(k,j,i) - dt_3d * &
	656	( tend(k,j,i) - tend(k,j,i-1) ) * ddx * &
[1918]	657	weight_pres_l
[1113]	658	ENDIF
[1]	659	ENDDO
[1257]	660	!$acc loop independent
[1113]	661	DO k = 1, nzt
	662	IF ( k > nzb_v_inner(j,i) ) THEN
	663	v(k,j,i) = v(k,j,i) - dt_3d * &
	664	( tend(k,j,i) - tend(k,j-1,i) ) * ddy * &
[1918]	665	weight_pres_l
[1113]	666	ENDIF
[673]	667	ENDDO
[1]	668
	669	ENDDO
	670	ENDDO
[1113]	671	!$acc end kernels
[1]	672	!$OMP END PARALLEL
[1113]	673
	674	!
	675	!-- Sum up the volume flow through the right and north boundary
	676	IF ( conserve_volume_flow .AND. bc_lr_cyc .AND. bc_ns_cyc .AND. &
	677	nxr == nx ) THEN
	678
	679	!$OMP PARALLEL PRIVATE (j,k)
	680	!$OMP DO
	681	DO j = nys, nyn
	682	!$OMP CRITICAL
[1845]	683	DO k = nzb_u_inner(j,nx) + 1, nzt
[1113]	684	volume_flow_l(1) = volume_flow_l(1) + u(k,j,nx) * dzw(k)
	685	ENDDO
	686	!$OMP END CRITICAL
	687	ENDDO
	688	!$OMP END PARALLEL
	689
	690	ENDIF
	691
	692	IF ( conserve_volume_flow .AND. bc_ns_cyc .AND. bc_lr_cyc .AND. &
	693	nyn == ny ) THEN
	694
	695	!$OMP PARALLEL PRIVATE (i,k)
	696	!$OMP DO
	697	DO i = nxl, nxr
	698	!$OMP CRITICAL
[1845]	699	DO k = nzb_v_inner(ny,i) + 1, nzt
[1113]	700	volume_flow_l(2) = volume_flow_l(2) + v(k,ny,i) * dzw(k)
	701	ENDDO
	702	!$OMP END CRITICAL
	703	ENDDO
	704	!$OMP END PARALLEL
	705
	706	ENDIF
[673]	707
[1]	708	!
	709	!-- Conserve the volume flow
[707]	710	IF ( conserve_volume_flow .AND. ( bc_lr_cyc .AND. bc_ns_cyc ) ) THEN
[1]	711
	712	#if defined( __parallel )
[622]	713	IF ( collective_wait ) CALL MPI_BARRIER( comm2d, ierr )
[1]	714	CALL MPI_ALLREDUCE( volume_flow_l(1), volume_flow(1), 2, MPI_REAL, &
	715	MPI_SUM, comm2d, ierr )
	716	#else
	717	volume_flow = volume_flow_l
	718	#endif
	719
[1799]	720	volume_flow_offset(1:2) = ( volume_flow_initial(1:2) - volume_flow(1:2) ) / &
	721	volume_flow_area(1:2)
[1]	722
	723	!$OMP PARALLEL PRIVATE (i,j,k)
	724	!$OMP DO
	725	DO i = nxl, nxr
	726	DO j = nys, nyn
[667]	727	DO k = nzb_u_inner(j,i) + 1, nzt
	728	u(k,j,i) = u(k,j,i) + volume_flow_offset(1)
[719]	729	ENDDO
[1845]	730	DO k = nzb_v_inner(j,i) + 1, nzt
[667]	731	v(k,j,i) = v(k,j,i) + volume_flow_offset(2)
	732	ENDDO
[1]	733	ENDDO
	734	ENDDO
[667]	735
[1]	736	!$OMP END PARALLEL
	737
	738	ENDIF
	739
	740	!
	741	!-- Exchange of boundaries for the velocities
[1113]	742	IF ( numprocs == 1 ) THEN ! workaround for single-core GPU runs
	743	on_device = .TRUE. ! to be removed after complete porting
	744	ELSE ! of ghost point exchange
	745	!$acc update host( u, v, w )
	746	ENDIF
[667]	747	CALL exchange_horiz( u, nbgp )
	748	CALL exchange_horiz( v, nbgp )
	749	CALL exchange_horiz( w, nbgp )
[1113]	750	IF ( numprocs == 1 ) THEN ! workaround for single-core GPU runs
	751	on_device = .FALSE. ! to be removed after complete porting
	752	ELSE ! of ghost point exchange
	753	!$acc update device( u, v, w )
	754	ENDIF
[1]	755
	756	!
	757	!-- Compute the divergence of the corrected velocity field,
[1908]	758	!-- A possible PE-sum is computed in flow_statistics. Carry out computation
	759	!-- only at last Runge-Kutta step.
[1918]	760	IF ( intermediate_timestep_count == intermediate_timestep_count_max .OR. &
	761	intermediate_timestep_count == 0 ) THEN
[1908]	762	CALL cpu_log( log_point_s(1), 'divergence', 'start' )
	763	sums_divnew_l = 0.0_wp
[1]	764
	765	!
[1908]	766	!-- d must be reset to zero because it can contain nonzero values below the
	767	!-- topography
	768	IF ( topography /= 'flat' ) d = 0.0_wp
[1]	769
[1908]	770	localsum = 0.0_wp
	771	threadsum = 0.0_wp
[1]	772
[1908]	773	!$OMP PARALLEL PRIVATE (i,j,k) FIRSTPRIVATE(threadsum) REDUCTION(+:localsum)
	774	!$OMP DO SCHEDULE( STATIC )
[1]	775	#if defined( __ibm )
[1908]	776	DO i = nxl, nxr
	777	DO j = nys, nyn
	778	DO k = nzb_s_inner(j,i)+1, nzt
	779	d(k,j,i) = ( u(k,j,i+1) - u(k,j,i) ) * ddx + &
	780	( v(k,j+1,i) - v(k,j,i) ) * ddy + &
	781	( w(k,j,i) - w(k-1,j,i) ) * ddzw(k)
	782	ENDDO
	783	DO k = nzb+1, nzt
	784	threadsum = threadsum + ABS( d(k,j,i) )
	785	ENDDO
[1]	786	ENDDO
	787	ENDDO
	788	#else
[1908]	789	!$acc kernels present( d, ddzw, rflags_s_inner, u, v, w )
	790	!$acc loop collapse( 3 )
	791	DO i = nxl, nxr
	792	DO j = nys, nyn
	793	DO k = 1, nzt
	794	d(k,j,i) = ( ( u(k,j,i+1) - u(k,j,i) ) * ddx + &
	795	( v(k,j+1,i) - v(k,j,i) ) * ddy + &
	796	( w(k,j,i) - w(k-1,j,i) ) * ddzw(k) &
	797	) * rflags_s_inner(k,j,i)
	798	ENDDO
[1113]	799	ENDDO
	800	ENDDO
[1908]	801	!$acc end kernels
[1113]	802	!
[1908]	803	!-- Compute possible PE-sum of divergences for flow_statistics
	804	!$acc parallel loop collapse(3) present( d ) reduction(+:threadsum)
	805	DO i = nxl, nxr
	806	DO j = nys, nyn
	807	DO k = nzb+1, nzt
	808	threadsum = threadsum + ABS( d(k,j,i) )
	809	ENDDO
[1]	810	ENDDO
	811	ENDDO
[1908]	812	!$acc end parallel loop
[1]	813	#endif
[667]	814
[1908]	815	localsum = localsum + threadsum
	816	!$OMP END PARALLEL
[1]	817
	818	!
[1908]	819	!-- For completeness, set the divergence sum of all statistic regions to those
	820	!-- of the total domain
	821	sums_divnew_l(0:statistic_regions) = localsum
[1]	822
[1908]	823	CALL cpu_log( log_point_s(1), 'divergence', 'stop' )
[1]	824
[1908]	825	ENDIF
	826
[1]	827	CALL cpu_log( log_point(8), 'pres', 'stop' )
	828
	829
	830	END SUBROUTINE pres

Note: See TracBrowser for help on using the repository browser.

Download in other formats:

| Impressum | ©Leibniz Universität Hannover |