Week - 8 Mass Scaling

 OBJECTIVE:

 1) To reduce the runtime of the model using the mass scaling and ensuring the stability of the mass scaling i.e, the model with mass scaling and without mass scaling should behave the same.

 2) To observe the effect of the DT2MS on the CPU time and also limiting the mass scaling to only 8%.

 3) To observe the effect of the TSSFACE(Time scale factor) on the CPU time for running the simulation.

 4) To run the same model using the implicit solver with necessary changes and compare the explicit and implicit run time.

 Fig 1 Given model

CASE SET UP  AND EXECUTION :

CASE 1 : STUDYING THE EFFECT OF DT2MS ON CPU RUNTIME

1. Determining the baseline timestep for the given model

        The given model is run without any mass scaling to determine the baseline timestep . Since, there is no mass scaling during this run, the percentage mass increase would be zero.

        The baseline timestep is found to be  4.89e-5 sec.

2. Mass scaling using the control timstep

       The mass scaling is done by using the Control timestep card and changing the DT2MS value in the card. Since, the baseline timestep has been found to be 4.89 e-5 in the first run without mass scaling, the DT2MS value should be higher than the baseline timestep value i.e, 4.89 e-5. 

      The DT2MS in the second run has been entered as 1.5e-4 sec  and the model is run using the explicit solver. The percentage increase in the mass due to the mass scaling is also noted down along with the CPU runtime.

Fig 2 Changing the DT2MS value

3. Optimizing the DT2MS for the percentage increase in the mass

       The Percentage increase in the mass during the second run has been found to be 30.517% which is not accepteable as it is higher than 8% mass increase limit.

        The percentage increase in the mass can be controlled by reducing the value of the DT2MS and at the same time ensuring that the value stays higher than the baseline timestep . This way with each run the DT2MS value is reduced utill the percentage increase in the mass gets below the target value of the 8% and the CPU runtime is reduced as compared to the CPU runtime of the model without mass scaling.

CASE 2 : STUDYING THE EFFECT OF TIME STEP SCALE FACTOR (TSSFAC)

1. Changing the Time step scale factor(TSSFAC) keeping the DT2MS constant

        The DT2MS value for which the mass scaling was optimized for 8% is kept constant and the TSSFAC value is changed in each run in order to study the effect of the Time step scale factor on the CPU runtime.

CASE 3 : Running the model using the Implicit solver 

1. Adding the Implicit_Genral card

      In addition to the already existing cards, the Implicit_general control card is added in order to carry out the Implicit analysis. The IMFLAG value is kept at "1" in order for turning ON the flag for the Implcit analysis.

Fig 3 Creating the Implcit general card for turning the implicit flag ON

  2.Creating the IMLICIT_AUTO card for automatic adjustment of timestep

                     The flag for the automatic adjustment of the timestep is turned ON by entering the value for AUTO in the IMPLICIT_AUTO card.

Fig 4 Creating the card Implicit_auto for automatic adjustment of the timestep

 3. Creating the IMPLICIT_SOLUTION card for setting the limit for the no of iterations and the tolerance limit for the displacement relative convergence .

                                 The IMPLICIT_SOLUTION card is created for setting the limit for the no iterations at each timestep for convergance. In this card, the displacement realative convergance tolerance limit can also be set.

Fig 5 Creating the IMPLICIT_AUTO card for limiting the no of Iterations for each timestep and setting the tolerance limit for the displacement convergance

4. Creating the IMPLICIT_SOLVER card 

                 This an optional card which  applies to implicit calculations. The linear equation solver performs the CPU-intensive stiffness matrix inversion.

Fig 6 Creating the IMPLICIT_SOLVER card

5. Creating the boundary conditions

     i. Constraining the lower pin

          The lower pin is constrained by using the Single point constraint set card . The already created  Node set of the lower pin is  used for constraining the lower pin. The lower pin is constrained in all the directions. 

Fig 7 Constraining the lower pin in all degrees of freedom

      ii.Constraining the Upper pin 

          The upper pin is constrained in all the degrees of freedom except the translational in  z-direction . The Nodeset of the upper pin is selected in Single point constraint set card.

Fig 8 Constraining the upper pin

     iii. Prescribed motion set for the Upper Pin

            The Upper pin is subjected to prescribed motion set which is the displacement in the z-direction. The prescribed motion Rigid card is created in which the Part id of the upper pin is made use of. The degree of freedom which should be attributed to the Upper pin is mentioned in the card. Also the type of the motion that is being prescribed is mentioned in the card. The Curve which described the motion with respect to the time is also assigned in the card.

Fig 9 Applying the prescribed motion to the upper pin

  6. All the other cards such as *contact, *control termination, * Material, *section , *Part already exist and are kept unchanged.

RESULTS:

1. CPU runtime vs DT2MS

Fig 10 CPU runtime vs Time step

INFERENCE :

 1.The CPU runtime intially increases with the increase in the Time step but then it starts to decrease with the increase in the time step(DT2MS). Thus, it can be concluded that total runtime can be controlled by increasing the time step      with the addition of the Mass also known as Mass scaling .

 2. However, the addition of the Mass can lead to inaccuracies hence, the increase in time step has to be optimized for 8% Mass increase.  Thus , the time step at which the mass scaling is only 8% is 1.02E-4.

 3.  Any further increase in the time step value pf 1.02E-4 may lead to Mass scaling of above 8% which may not be accepteable in this case. 

 4. The CPU runtime for the time step of 1.02E-4 has been found to be 14.75hrs as compared to the CPU runtime of 61.94 hrs in case of the model run without mass scaling. 

2. CPU runtime Vs Time step scale factor (TSSFAC)

Fig 11 CPU TIME VS TIME STEP SCALE FACTOR(TSSFAC)

INFERENCE : 

1.For the constant DT2MS value, the CPU runtime increases with the decrease in the Time step scale factor(TSSFAC) as the time step applied at each run is the product of DT2MS and TSSFAC. Hence, for the run where the DT2MS =           1.02E-4 sec and TSSFAC = 0.9 ,

                        Time step = (1.02E-4)* 0.9 = 9.18E-5 sec.

    Thus, constant DT2MS value as the TSSFAC value is decreased after every run , the time step value at each run is getting decreased and this leading to an overall increase in the CPU runtime.

2. The Optimum TSSFAC for the analysis is 0.9

3. Implicit Analysis

 Note :  *The Simulation for the Implicit ananlysis could be found in the Powerpoint presentation that has been attached with the report.

Fig 12 Failure near the Notch

 INFERENCE :

 1.It the CPU runtime in the Implicit solution is just 4m and 42 sec which is very fast as  compared to the explicit analysis.

 2. It can be observed from the simulation that the plate is most likely to fail near U-Notch due to the presence of the huge stress concentration near the u-notch.

 3.The area of huge stress concentration is due to the discontinuity.

CONCLUSION:
1. The simulation was Optimized for the Time step(DT2MS) and the Time step scale factor (TSSFAC) of 1.02E-4 and 0.9 respectively.

2. The mass scaling was achieved for the targeted value of 8%.

3. The effect of the Time step(DT2MS) on the CPU run time was established successfully.

4.The effect of the Time step scale factor (TSSFAC)  on the CPU runtime was established successfully.

5. The Implicit analysis for the given model was run successfully and the comparison of the runtime between the Implicit analysis and the explicit analysis was carried out successfully.