stefanix1972
Aerospace
Dear all,
I'm trying to analyze a spacecraft manouevre around its CoG by Nastran 2019 SOL400 transient dynamic analysis.
Preliminarly, I checked with a small model what the best solution could be, and I found other SOLs giving problems:
SOL109 direct transient gives "inflated" deformed shape (its dimensions increase in the plane norma to the axis of rotation);
SOL129 direct non linear transient does not accept SPCD imposed rotation (around CoG);
SOL112 modal transient does not give corrected shape (it inflates);
Therefore, with small model, SOL400 is fine for me, it gives right answers.
Running full S/C model (some hundreds of thousands nodes) convergence problems occur.
I tried different options, also SOL400 with fixed timestep (not correct, I know), but also in this case deformed geometry is not real (it inflates).
Therefore I used ADAPT option as below, trying to tune different parameters and forcing the solution to continue by GENERAL -10 (negative sign).
Analysis parameres are the following ones:
$------1$------2$------3$------4$------5$------6$------7$------8$------9
$ NONLINEAR TIME HISTORY DEFINITION
$NLSTEP ID TOTTIME CTRLDEF
NLSTEP 555 140.0 SEVERELY
$ GENERAL MAXITER MINITER MAXBIS CREEP
GENERAL -10
$ ADAPT DTINITF DTMINF DTMAXF NDESIR SFACT INTOUT NSMAX
$ IDAMP DAMP CRITTID IPHYS LIMTAR RSMALL RBIG
$ ADJUST MSTEP RB UTOL
ADAPT 0.0017851.0-7 0.001785 0
1
$ MECH CONV EPSU EPSP EPSW KMETHOD KSTEP MRCONV
$ MAXQN MAXLS LSTOL FSTRESS
MECH A PFNT
$------1$------2$------3$------4$------5$------6$------7$------8$------9
Now, the problem is that I need to set a maximum timestep (DTMAXF) to properly capture some appendices dynamics. At the same time convergences
criteria make the timestep shortening and the analysis is becoming long, as you can see from the picture.
Just consider that I need to arrive to 140 seconds and now I'm stuck around 21 seconds after 3 days run.
I'm running on a company server therefore hardware is not a problem with normal models, unfortunately here I'm using full S/C model.
I also tried to cut some large portions of the model, but the part of interest is anyway very large in terms of DoFs.
Has anyone a suggestion about the analysis or an alternative approach? I know that multibody is an alternative but I'm not familiar with it.
Regards,
Stefano.
I'm trying to analyze a spacecraft manouevre around its CoG by Nastran 2019 SOL400 transient dynamic analysis.
Preliminarly, I checked with a small model what the best solution could be, and I found other SOLs giving problems:
SOL109 direct transient gives "inflated" deformed shape (its dimensions increase in the plane norma to the axis of rotation);
SOL129 direct non linear transient does not accept SPCD imposed rotation (around CoG);
SOL112 modal transient does not give corrected shape (it inflates);
Therefore, with small model, SOL400 is fine for me, it gives right answers.
Running full S/C model (some hundreds of thousands nodes) convergence problems occur.
I tried different options, also SOL400 with fixed timestep (not correct, I know), but also in this case deformed geometry is not real (it inflates).
Therefore I used ADAPT option as below, trying to tune different parameters and forcing the solution to continue by GENERAL -10 (negative sign).
Analysis parameres are the following ones:
$------1$------2$------3$------4$------5$------6$------7$------8$------9
$ NONLINEAR TIME HISTORY DEFINITION
$NLSTEP ID TOTTIME CTRLDEF
NLSTEP 555 140.0 SEVERELY
$ GENERAL MAXITER MINITER MAXBIS CREEP
GENERAL -10
$ ADAPT DTINITF DTMINF DTMAXF NDESIR SFACT INTOUT NSMAX
$ IDAMP DAMP CRITTID IPHYS LIMTAR RSMALL RBIG
$ ADJUST MSTEP RB UTOL
ADAPT 0.0017851.0-7 0.001785 0
1
$ MECH CONV EPSU EPSP EPSW KMETHOD KSTEP MRCONV
$ MAXQN MAXLS LSTOL FSTRESS
MECH A PFNT
$------1$------2$------3$------4$------5$------6$------7$------8$------9
Now, the problem is that I need to set a maximum timestep (DTMAXF) to properly capture some appendices dynamics. At the same time convergences
criteria make the timestep shortening and the analysis is becoming long, as you can see from the picture.
Just consider that I need to arrive to 140 seconds and now I'm stuck around 21 seconds after 3 days run.
I'm running on a company server therefore hardware is not a problem with normal models, unfortunately here I'm using full S/C model.
I also tried to cut some large portions of the model, but the part of interest is anyway very large in terms of DoFs.
Has anyone a suggestion about the analysis or an alternative approach? I know that multibody is an alternative but I'm not familiar with it.
Regards,
Stefano.
