Input Reciprocating Machine Motion into Dynamic Analysis?

[Harry.Braywood]

Hi Everyone,

I am relatively new to Inventor Nastran, particularly with Dynamic Analyses. I work for a company that manufactures units for the refrigeration and HVAC industry. We use compressors which have crankshafts and pistons (like car engines), and I would just like some advice on how best to induce the loads into the framework. 

I am aware that enforced motion would be the best type of load to induce here. I am also aware that constraints may need to be applied about the same direction as the enforced motion. There are two main motions at play here. The crankshaft and the pistons. The compressor model is imported and from another manufacturer and given how complex its geometry is, I have used a concentrated mass with the centre point being the centre of gravity for the machine.

What would be the best way to induce the enforced rotational motion from the crankshaft?

What would be the best way to induce the reciprocating motion from the pistons? 

I would like to do a Frequency Response Analysis, because the machines can run at different speeds, and I would like to see how the structure responds at these different speeds.

Any advice will be appreciated.

[John_Holtz]

Hi @Harry.Braywood . Welcome to the Inventor Nastran forum.

Frequency response applies sinusoidal loads to the model. The loads operate at the same frequency (which I assume would be true for the rotating shaft moving the reciprocating pistons) and can be offset from each other (such as 4 cylinders at different stages). Here's the key question: are the forces from the rotating shaft and reciprocating piston sinusoidal? (or close enough to assume sinusoidal). I would think the crankshaft reaction force does create a sinusoidal load on the structure. For the shaft, this article describes how to apply the load. (The load is a force, not an enforced motion. You may know how far the shaft/piston moves relative to the body of the compressor, but you do not know how far it moves relative to the ground which is what an enforced motion does. Enforced motion is not the correct load type in this situation.) See Analyze a rotating component with an imbalance load in Inventor Nastran.

Note that a rotating load is not analyzed as rotating. You analyze the structure once with the "rotating" load in the Y direction to see what happens in the vertical direction. Then you analyze the structure a second time with the "rotating" load in the Z direction to see what happens in the horizontal direction. The assumption is what happens in the vertical and horizontal directions are independent, so you do not need to "physically" rotate the load. You're just looking at the max that happens as the load "passes through" the horizontal or vertical directions.

If the piston is not compressing anything, then I agree that the force to move the piston is sinusoidal; that is, reaction force between the shaft and piston F = Fmax*sin(frequency*time). Depending on how the cycle behaves, the force may be higher at some points in the cycle than a sine curve and lower (much lower?) at other points in the cycle than a sine curve.

• If you are willing to assume the piston loads are sinusoidal, then Frequency Response would work, and the piston load would be another force at the concentrated mass in the appropriate direction. (X? Z? Some other angle?)
• If the load cycle is not close enough to sinusoidal to use the Frequency Response analysis, then you better have a lot of free time to let the model run. A non-sinusoidal load curve will require a Transient Response analysis. Operating at N rpm and running the analysis for X hundreds of cycles to get from the initial speed (0 is the easiest to enter) to the final speed where the pattern repeats on each cycle will take some time to run. You will want 20 to 50 time steps per cycle, so the total number of calculation steps is large. (Do a small test model first to work out the setup.)

Does this make sense? Let us know.

John