Problem with units: Job too light

[ashu28]

Hello all,
I am simulating a job which has a dimension in microns. So these are the units I am using:
Length: micro meter (1e-6m), Mass: Kg, Time: secSo my E is in MPa, Forces are in micro-Newtons and density is in kg/(micro-meter)^3

Now this means that for my analysis (polycarbonate):
E=2000 MPa
Density = 1300e-18 (kg/cubic micro meters)

Thus the density becomes too low and hence I have to use arbitrarily high mass scaling (1e17). This means that I increase density artificially a bit too much, which is to my knowledge not appropriate. Not applying mass scaling makes simulation go on and on for days.
But I need to have length in microns only. Can someone advice on what should I do ?

Regards,
Ashu

[Jorgen]

Your set of units seem consistent to me. In your units the density is expressed in terms of 1e18 kg/m^3. That is certainly a large number, but why do you say that is not appropriate? Does it help to run your simulation in double precision?

An alternative might be to derive a different set of units that still use micro-meters and seconds, but perhaps use nano Newtons for force.

- Jorgen

[ashu28]

Thank you for the reply. My concern is that I use the following mass scaling for machining simulations:
*variable mass scaling, type=below min, dt=1e-4, freq = (something)

This gives me the percentage change in the mass of the system in orders of 1e15 which is way large. This indicates that ABAQUS is artificially increasing the density which in turn might mean that the density supplied is very low. In fact, if I dont touch anything else and increase the density, I dont need to have such high mass scaling (I have noted this through simulations)

Thus I feel there is something wrong about density or units in here. The job size is 2 x 2 sq micro meters.

Am I thinking correct or I am mistaken somewhere. I am curious as to why do you or others simulate without such a high mass scaling.

Please help me out

Ashu

[Jorgen]

If the Poisson's ratio is 0.4 (just as an example), then the stable time increment size is approximate as specified in the equation below.

Hence, if you use:
[L] = micro meter
[M] = kg
[t] = s

and the material specific values:
L = 1
rho = 1300e-18
E = 2000

then you get a stable time increment of about 6e-10

In order to get a stable time increment of 1e-4 (as you specified), then you would have to multiply the density by a factor of about 3e10, which is a very large value.

If possible, I suggest a less aggressive mass scaling approach.

- Jorgen