Hello everybody Calculating mechanical natural frequencies of a structure, (e.g. reciprocating compressor) by a finite element software, I think I have to bring the moving parts and process forces out of the calculations and only the integrated structure (i.e. frame, cylinders, etc.) be analyzed. Is it right? In other words the moving parts weight (e.g. pistons) and process forces (e.g. gas compression force) effect on system response and not on whole structure MNFs. Please put me in the picture ,if possible with a reliable reference. Thanks

Well if you have fully elastic liaisons btw each parts you won't get any coupled vibrations (what I assume you are looking for). Best way to do here is to look parts after parts. Then you do a rigid model of the whole assembly and see how it behave. Further actions might be necessary depending on the result and what you do want to achieve. But make sure your result are not meshes dependant (no singularities). What Software are you working on ? SRDFMC

Natural frequencies have to do with eigenvalues of the compliance and mass matrices. It doesn't have anything to do with process forces. For a structure like an air compressor, the frequencies will be really high (think hitting it with a hammer and what sound you would get)

This i really a stupid awnser Maniac... There is nothing like the books for engineers. Parts will interact sometime and that's what we are talking abt. And by the way, none engineer shld be maniac. That's the worst way way to go in our professin. Open minded, curious, modest...

I would be careful whose answer I called stupid, if I were not able to spell "awmser" correctly; further, criticizing someone's professionalism is best left for those that can spell "professin". As to the actual substance: the question regarded natural frequency of a structure. Before cheap FEA codes and powerful computers this was determined by sticking accelerometers on the structure and whacking it with an (instrumented) hammer. Looking at the frequency and phase shift of the output of the various accelerometers gave natural frequencies and modal shapes. If this is not what he is looking for then the question needs to be rephrased. The only thing that affects natural frequencies of a structure are masses and stiffnesses. Natural frequencies will be expressed in the form w=sqrt(k/m) with w being a frequency, k is a generalized stiffness, and m is a generalized mass. The mass of the components in the compressor matter. If he is asking about the natural frequency of a linear oscillating compressor then 1) this is not a structural frequency and 2) the process forces DO matter because the pressure on the piston varies with position, which can be approximated by a spring rate. For STRUCTURES with multiple degree of freedom systems the equations become a matrix equation that i can't remember the exact form of off the top of my head. Look it up. As far as 'maniacal' - its a deliberate malapropism of 'mechanical' - you know - a play on words, a pun of sorts, a quibble. In the same way that engineering is applied science, "maniacal engineering" is applied mad science. I am an eclecto-maniacal engineer, and if you have a problem with applied mad scientists then I pity your incurious closed mind and the small world you must live in. A rational man adapts to his surrounding, an irrational man expects his surroundings to adapt to him, therefore all progress is made by irrational men. ---- or maniacs.

srdfmc - do not insult other members. Forum rules: http://www.mechanicaldesignforum.com/showthread.php?332-MECHANICAL-DESIGN-FORUM-RULES-please-read&#