order of recoating, dwell and simulation

[franzbuL982R]

Hi I have been trying to find an answer in the documentation, but was so far unsuccessful. 

Situation:

I would like to compare my simulation to some real world process. However, in the experiment where I got the data from they first melted a layer, then waited for 13 sec, measured the temperature of the new layer and recoated the new layer afterwards. 

1. I am aware that I wont be able to have exactly the same process, since the temperature is simulated after the recoating and not before. However I was wondering if there is a way to come as close as possible to this process. 

An option I looked into was the dwell time. However, I was not able to find more information on the dwell time. whether the temperature is simulated before or after the dwell time.  

2. Furthermore, I've found the description of the powder bed dwell time (*PBDL) rather confusing. the description states that its only parameter "r1:" is supposed to be the "nr. of parts on the build plate".  while I was expecting to define a fixed dwell time. 

To put it in graphics: 

1. ■■■■■ recoat time ■■■■■ | ■■■■■ dwell time ■■■■■ | ■■■■■ simulation ■■■■■ 

2. ■■■■■ recoat time ■■■■■ | ■■■■■ simulation ■■■■■ | ■■■■■ dwell time  ■■■■■

3. ■■■■■ dwell time  ■■■■■ | ■■■■■ simulation ■■■■■ | ■■■■■ recoat time ■■■■■

what's the order of the recoating, dwell and simulation? 

Best regards, 

Franz

[franzbuL982R]

Hi I have been trying to find an answer in the documentation, but was so far unsuccessful. 

Situation:

I would like to compare my simulation to some real world process. However, in the experiment where I got the data from they first melted a layer, then waited for 13 sec, measured the temperature of the new layer and recoated the new layer afterwards. 

1. I am aware that I wont be able to have exactly the same process, since the temperature is simulated after the recoating and not before. However I was wondering if there is a way to come as close as possible to this process. 

An option I looked into was the dwell time. However, I was not able to find more information on the dwell time. whether the temperature is simulated before or after the dwell time.  

2. Furthermore, I've found the description of the powder bed dwell time (*PBDL) rather confusing. the description states that its only parameter "r1:" is supposed to be the "nr. of parts on the build plate".  while I was expecting to define a fixed dwell time. 

To put it in graphics: 

1. ■■■■■ recoat time ■■■■■ | ■■■■■ dwell time ■■■■■ | ■■■■■ simulation ■■■■■ 

2. ■■■■■ recoat time ■■■■■ | ■■■■■ simulation ■■■■■ | ■■■■■ dwell time  ■■■■■

3. ■■■■■ dwell time  ■■■■■ | ■■■■■ simulation ■■■■■ | ■■■■■ recoat time ■■■■■

what's the order of the recoating, dwell and simulation? 

Best regards, 

Franz

[jeff.irwin76CLG · Accepted solution]

Hi @franzbuL982R,

There is a simulation time step at the beginning and end of each layer.  The end of one layer is only an instant before the start of the next layer.  Depending on your *TRAN settings, you may sometimes get 3 or more time steps per simulation layer group, but the usual case is 2 time steps per layer group.

The beginning of the layer occurs as soon as the laser turns on.  The end of the layer is after the laser has scanned the whole layer, after the recoater blade (or roller) has swept across, and just before the laser turns on for the next layer.

Maybe it will help to look at a log file snippet, e.g. for example 02:

Starting refine
Solving layer 2 of 17
Number of refined nodes = 3979
Number of refined elements = 2280
Adding dwell (processing) time : 40.6054687500000
Meshed cross sectional xy area = 72.1874999999993
Number of equations = 2754
Number of constrained eqns = 1225
Direct sparse solver = PARDISO
4-byte integer version

Starting auxspar
Number of no zeros symmetric = 30,432
Sparse preprocessing complete
results file name =results/02_thermal_00002.bin
inc = 4 time = 220.60547 iter = 1 eps = 0.20673E+03
inc = 4 time = 220.60547 iter = 2 eps = 0.39478E-12
Writing record: 1, time: 220.605468750000
Increment end
CPU wall for increment 4 = 00:00:00.11, since start = 00:00:12.82
inc = 5 time = 441.11094 iter = 1 eps = 0.17221E+03
inc = 5 time = 441.11094 iter = 2 eps = 0.31458E-12
Finished writing file results\02_thermal_5.case
Writing record: 2, time: 441.110937500000
Increment end
CPU wall for increment 5 = 00:00:00.03, since start = 00:00:12.86
Layer end
Total RAM = 33,245,048 kB
Total RAM used = 24,316,660 kB, 73.14 %
RAM used for this process = 59,228 kB, 0.18 %

Available disk space = 344.137 GB

Starting refine
Solving layer 3 of 17

Number of refined nodes = 4812
Number of refined elements = 2952
fmesh = results\02_thermal_3.geo
Adding dwell (processing) time : 40.6054687500000
Meshed cross sectional xy area = 72.1874999999993
Number of equations = 3587
Number of constrained eqns = 1225
Direct sparse solver = PARDISO
4-byte integer version

Starting auxspar
Number of no zeros symmetric = 40,624
Sparse preprocessing complete
results file name =results/02_thermal_00003.bin
inc = 6 time = 441.21094 iter = 1 eps = 0.20641E+03
...

The second layer group starts at time increment (step) 4 at 220.6 seconds and ends at 441.1 seconds.  The next layer group (layer group three) starts an instant later at 441.2 seconds.  In between 220.6 s and 441 s (a difference of 220.6 s), there are 40.6 s of processing time when the laser is on.  This processing time is a simple function of the cross-sectional area and the processing parameters.  The remaining 180 s are comprised of recoating time, 10 s per layer for 18 physical layers per simulation layer group.

If you need to add more types of dwell other than laser-on time and recoating time, e.g. work stoppage time for machine maintenance, you can use the *ZDWL card.

If you want finer time resolution than two time steps per layer group, you could try a moving source model instead of a part-scale powder bed model.

[jeff.irwin76CLG · Accepted solution]

Hi @franzbuL982R,

There is a simulation time step at the beginning and end of each layer.  The end of one layer is only an instant before the start of the next layer.  Depending on your *TRAN settings, you may sometimes get 3 or more time steps per simulation layer group, but the usual case is 2 time steps per layer group.

The beginning of the layer occurs as soon as the laser turns on.  The end of the layer is after the laser has scanned the whole layer, after the recoater blade (or roller) has swept across, and just before the laser turns on for the next layer.

Maybe it will help to look at a log file snippet, e.g. for example 02:

Starting refine
Solving layer 2 of 17
Number of refined nodes = 3979
Number of refined elements = 2280
Adding dwell (processing) time : 40.6054687500000
Meshed cross sectional xy area = 72.1874999999993
Number of equations = 2754
Number of constrained eqns = 1225
Direct sparse solver = PARDISO
4-byte integer version

Starting auxspar
Number of no zeros symmetric = 30,432
Sparse preprocessing complete
results file name =results/02_thermal_00002.bin
inc = 4 time = 220.60547 iter = 1 eps = 0.20673E+03
inc = 4 time = 220.60547 iter = 2 eps = 0.39478E-12
Writing record: 1, time: 220.605468750000
Increment end
CPU wall for increment 4 = 00:00:00.11, since start = 00:00:12.82
inc = 5 time = 441.11094 iter = 1 eps = 0.17221E+03
inc = 5 time = 441.11094 iter = 2 eps = 0.31458E-12
Finished writing file results\02_thermal_5.case
Writing record: 2, time: 441.110937500000
Increment end
CPU wall for increment 5 = 00:00:00.03, since start = 00:00:12.86
Layer end
Total RAM = 33,245,048 kB
Total RAM used = 24,316,660 kB, 73.14 %
RAM used for this process = 59,228 kB, 0.18 %

Available disk space = 344.137 GB

Starting refine
Solving layer 3 of 17

Number of refined nodes = 4812
Number of refined elements = 2952
fmesh = results\02_thermal_3.geo
Adding dwell (processing) time : 40.6054687500000
Meshed cross sectional xy area = 72.1874999999993
Number of equations = 3587
Number of constrained eqns = 1225
Direct sparse solver = PARDISO
4-byte integer version

Starting auxspar
Number of no zeros symmetric = 40,624
Sparse preprocessing complete
results file name =results/02_thermal_00003.bin
inc = 6 time = 441.21094 iter = 1 eps = 0.20641E+03
...

The second layer group starts at time increment (step) 4 at 220.6 seconds and ends at 441.1 seconds.  The next layer group (layer group three) starts an instant later at 441.2 seconds.  In between 220.6 s and 441 s (a difference of 220.6 s), there are 40.6 s of processing time when the laser is on.  This processing time is a simple function of the cross-sectional area and the processing parameters.  The remaining 180 s are comprised of recoating time, 10 s per layer for 18 physical layers per simulation layer group.

If you need to add more types of dwell other than laser-on time and recoating time, e.g. work stoppage time for machine maintenance, you can use the *ZDWL card.

If you want finer time resolution than two time steps per layer group, you could try a moving source model instead of a part-scale powder bed model.

[franzbuL982R]

thanks alot for your reply! 
I think "simulation" might have not been the best choice when I stated my question. 
as far as I understood the part scale simulation, the temperature output in the results is given after the last layer of the layer group has been simulated and the powder freshly recoated. (right?) However, is it somehow possible to have the measurement before the recoating? like right after the laser stopped?

or something that comes close to following timeline: 

Laser

Wait

Thermal Snapshot

Recoat

[franzbuL982R]

thanks alot for your reply! 
I think "simulation" might have not been the best choice when I stated my question. 
as far as I understood the part scale simulation, the temperature output in the results is given after the last layer of the layer group has been simulated and the powder freshly recoated. (right?) However, is it somehow possible to have the measurement before the recoating? like right after the laser stopped?

or something that comes close to following timeline: 

Laser

Wait

Thermal Snapshot

Recoat