BASEMENT 2.7 Crash when using Restart (sediment transport)

dazzle · 2018-01-31 07:27:58

Hi there,

I'm trying to set up a bedload transport simulation using a restart file.
The original calculation was aborted after approx. 940 seconds due to instabilities. Because I have set the restart_time_step to 60 seconds, I wanted to restart the simulation from timestep 900s. The command window does not show an error, but everytime I click on RUN Basement crashes shortly after. The generated ERR-file is empty, and the LOG-file is missing the timestep and output module (those are the last two modules in my input file and basement probably crashes before it has finished writing the LOG-file).

Because it is a time consuming simulation, I want to avoid starting from 0s again.
This is the first time I have used the restart function. Here is the input for the calculation:

--------------------------------------------------------------------------------------------------------------------
//---------------------------------------------------
// BASEMENT file
// generated by BASEview
//---------------------------------------------------
PROJECT {
title = TEST
author = DPo
date = 2018/01/29
}
DOMAIN {
PARALLEL {
number_threads = 7
}
PHYSICAL_PROPERTIES {
gravity = 9.81
viscosity = 1e-006
rho_fluid = 1000
}
BASEPLANE_2D {
region_name = test
GEOMETRY {
type = 2dm
file = mesh.2dm
STRINGDEF {
name = Inflow_1_strg
node_ids = (10 15 21 28 36)
upstream_direction = right
}
STRINGDEF {
name = Outflow_strg
node_ids = (1011 1010 1009 1091 1175 1174 1090 1089 1088 1078 1000 927 926 1002 1003 1082 1004 921 849 848 847 780 779 720 719 785 718 717 716 666 665 663 737 802 793 726 652 583 519 476 461 501 555 619 618 629 628 564 562 560 504 463 450 451 482 532 533 489 452 454 429 408 410 409 390 389 374 373 359 358 345 332 331)
upstream_direction = right
}
STRINGDEF {
name = Inflow_2_Strg
node_ids = (5584 5618 5648 5673 5698 5722 5744 5763 5784 5810 5837 5865 5891 5914 5937 5954 5969 5982 5996 6007)
upstream_direction = right
}
STRINGDEF {
name = Control_3
node_ids = (1956 2039 2120 2198 2274 2347 2418 2484 2555 2630 2695)
upstream_direction = right
}
STRINGDEF {
name = Control_4
node_ids = (1187 1270 1272 1188 1265 1264 1263 1262 1348 1426 1505 1591 1590 1680 1676 1677 1587 1586 1585 1588 1502 1512 1514 1602 1516 1437 1452 1533 1532 1619 1707 1625 1536 1535 1534 1541 1459 1380 1292 1209 1124 1123 1051 1137 1138 1224 1222 1221 1220 1303 1390 1389 1469 1553 1470 1400 1480 1479 1481 1403 1321)
upstream_direction = right
}
}
HYDRAULICS {
PARAMETER {
riemann_solver = exact
simulation_scheme = exp
minimum_water_depth = 0.01
}
FRICTION {
type = strickler
default_friction = 25
wall_friction = off
input_type = index_table
index = (1 2 3 10)
friction = (25 55 35 55)
}
BOUNDARY {
name = Inflow_1_bdy
string_name = Inflow_1_strg
type = hydrograph
slope = 10
file = inflow_45cms.dat
weighting_type = area
}
BOUNDARY {
name = Outflow_bdy
type = zero_gradient
string_name = Outflow_strg
}
INITIAL {
type = continue
file = CALC_0-900s.cgns
restart_solution_time = 900.000
}
BOUNDARY {
name = Inflow_2_bdy
string_name = Inflow_2_Strg
type = hydrograph
slope = 1
file = inflow_2.dat
weighting_type = conveyance
}
}
MORPHOLOGY {
PARAMETER {
porosity = 35
density = 2650
control_volume_type = constant
control_volume_thickness = 0.15
}
BEDMATERIAL {
GRAIN_CLASS {
diameters = (0.06 0.2 0.63 2 6.3 20 63 100)
}
MIXTURE {
name = Creekbed
volume_fraction = (6 3 4 8 14 21 26 18)
}
MIXTURE {
name = other
volume_fraction = (0 0 0 0 0 0 0 100)
}
SOIL_DEF {
name = Creekbed_soil
LAYER {
mixture = Creekbed
bottom_elevation = -15
}
}
SOIL_DEF {
name = other_soil
tau_erosion_start = 999999999
LAYER {
mixture = other
bottom_elevation = -0.000001
}
}
SOIL_ASSIGNMENT {
index = (1 2 3 10)
soil = (Creekbed_soil other_soil other_soil other_soil)
type = index_table
}
}
INITIAL {
type = continue
file = CALC_0-900s.cgns
}
BEDLOAD {
FORMULA {
bedload_formula = mpm_multi
bedload_factor = 0.5
bedforms_type = mpm_bedforms
}
PARAMETER {
limit_bedload_wetted = off
use_cell_averaged_bedload_flux = off
}
DIRECTION {
lateral_transport_type = lateral_bed_slope
lateral_index = (1)
lateral_transport_factor = 2.05
}
BOUNDARY {
type = IODown
string_name = Outflow_strg
}
BOUNDARY {
name = Sed_Inflow_2
type = sediment_discharge
string_name = Inflow_2_Strg
mixture = other
file = sed_inflow_2.dat
}
}
GRAVITATIONAL_TRANSPORT {
index = (1)
angle_failure_dry = (40)
angle_failure_wetted = (35)
angle_failure_deposited = (35)
gravity_transport_on_cells = fully wetted
angle_wetted_criterion = fully_wetted
min_changed_volume = 0.025
max_iterations = 7
cycle_step = 5
max_delta_z = 0.001
}
}
TIMESTEP {
CFL = 0.95
total_run_time = 6300
start_time = 900
}
OUTPUT {
console_time_step = 60
restart_time_step = 60
SPECIAL_OUTPUT {
type = BASEviz
output_time_step = 5
}
SPECIAL_OUTPUT {
type = node_centered
output_time_step = 60
values = (depth velocity wse friction deltaz)
format = sms
}
SPECIAL_OUTPUT {
stringdef_values = (Q wse u Qsed)
type = stringdef_history
output_time_step = 60
stringdefs = (Outflow_strg Control_3 Control_4)
}
}
}
}

--------------------------------------------------------------------------------------------------------------------

I hope maybe someone can locate the error.

Best
David

Lukas Vonwiller · 2018-01-31 09:55:49

Hi David

This is a known issue when restarting from both hydraulics and morphology (mainly on Windows). We tried to fix this bug already several times, but no success. There seems to be a problem with the external third party library CGNS, which is used to read and write cgns-files for restart. Maybe this problem will be solved in the next release, as we will update all 3rd party libraries and provide a 64-bit version also for Windows.

However, a possible workaround for now could be:

Sometimes it works after trying a couple of times (easier when starting Basement from the console)
Copy and rename the restart file, e.g. for restarting the morphology, mainly to avoid using exactly the same file
Use a different restart_solution_time for morphology and hydraulics. You could produce another hydraulic restart file with a short restart of a purely hydraulic simulation (e.g. simulation of 1 second)

I hope this helps and would appreciate a feedback of a successful workaround!

Best, Lukas

dazzle · 2018-01-31 11:13:28

Hi Lukas,

thank you for your reply and your suggestions.
I'm afraid BASEMENT still won't start the calculation.

You mention that this is a known issue mainly on windows - do you think I could use files from the old/aborted simulation - run on a Windows 10 machine - and restart the simulation on a Linux/Ubuntu PC?

Best
David

Lukas Vonwiller · 2018-01-31 11:15:39

Hi David, yes sure this should work!
Best, Lukas

User Forum of Software BASEMENT

#1 2018-01-31 07:27:58

BASEMENT 2.7 Crash when using Restart (sediment transport)

#2 2018-01-31 09:55:49

Re: BASEMENT 2.7 Crash when using Restart (sediment transport)

#3 2018-01-31 11:13:28

Re: BASEMENT 2.7 Crash when using Restart (sediment transport)

#4 2018-01-31 11:15:39

Re: BASEMENT 2.7 Crash when using Restart (sediment transport)

Board footer