Have you ever seen a shaker table used for creating Random Vibration? You'll notice that the system being shaken is bolted down to the table. The bolts and fixture have to be bulky and big, otherwise they too can contribute to the modal frequencies of the system being shook.
In Simulation, the system is therefore to be given a fixed geometry at the interface to the destined shaker table. Basically, at the mounting points. You don't model the shaker table. These fixed geometry points are then excited to the PSD map you provide.
Global excitation means that all fixed points are excited in the direction you specify. Base excitation means you specify which fixed geometric surfaces are excited. In the case above, if the system is only anchored to the shaker table, then you can use Global excitation as there is no other fixed geometry. However, say that one end of the system is bolted against a wall, e.g., an actuator. Here you can't excite that additional fixed spot, so Base Excitation is needed.
The other reason to use Base Excitation is if you have different PSDs to apply at each fixed geometry, as if you had accelerometers physically measuring the accelerations at those points.
I have addressed an anomaly in Simulation 2009 v2.1 that Base Excitations always invoke the Sparse Solver. You cannot flag the FFEPlus solver. This leads to hopelessly chewed up memory allocations for ~ 900,000 DOF systems with ~ 8 GB of memory. Always use Global Excitation unless you absolutely need it.
I have a questiong regarding Uniform Base Excitation. I have been given the "Y Displacement data" (Displacement Vs Hz data) to be applied on an enclosure fixed at 3 ends (i.e. its base) to perform the Linear Dynamic analysis and study the response. What value should I specify under my Displacements (with respect to directions in Y direction) after I add the given curve?
I usually just use "1" to specify direction, then define the actual value with my time variation curve, but am not 100% sure that is the best way to do it. Maybe someone more experienced can verify this.
Thanks for your reply. I thought that the value specified under direction is the maximum load value and the time curve specified represents the fraction of the load and 1 being the max. Again, I am not sure and I need to confirm this.
It turns out that it doesn't matter which way you do it. I figured that because the units for the frequency dependent curve were Hz vs.dimensionless that this might be the case, but I just ran a study to verify it. It seems that the initial value put in the direction field is just multiplied by the value in the PSD curve. I ran one study (Test1) with a value of 1 mm^2/Hz, and a PSD curve from 5-500 Hz with break points as follows:
5 Hz, 1e-5
100 Hz, 2e-5
200 Hz, 3e-5
300 Hz, 3e-5
400 Hz, 2e-5
500 Hz, 2e-5
The second study that I ran (Test 2) used a value of 1e-4 mm^2/Hz, and a PSD curve as follows:
5 Hz, 0.1
100 Hz, 0.2
200 Hz, 0.3
300 Hz, 0.3
400 Hz, 0.2
500 Hz, 0.2
The results are identical -- I've attached the .jpeg for each one. Hope this clears things up.
That clears my doubt. So it does consider the multiplier with the value under directions. I really appreciate your help.
Yes, I too have found that it acts as multiplier. It's useful if you have Maximum Predicted Environment (MPE) and need to, say, jump up 6 dB on the PSD by multiplying everything by 2.