Creating a larger master and slave system

The really significant reductions in CPU time from using Discrete Partitioning come from systems with at least 100 state variables. However, for learning the technique it is convenient to work with a much smaller system and some modestly-sized systems can give very respectable speedups. The system below is much more complex than the previous example, but it still has only 37 state variables. It has a very detailed model of a three cylinder hydraulic pump, a pressure relief valve and a hydraulic motor.

With this level of detail, phenomena such as pump ripple (about 100 Hz) and the dynamics of the relief valve are modeled, and the system will always be very "busy". The longest hydraulic pipe is 1 m and the expected lowest frequency in this pipe is about 300 Hz. Therefore it is reasonable to include some pipe wave dynamics. However, since the pump ripple is about 100 Hz we only need a small number of frequency modes in the pipe.

Note

This system has been chosen because it demonstrates the important features of discrete partitioning, and runs can be performed in a reasonable period of time. This system is not in any way intended to be a model of good hydraulic design. On most machines, runs can be completed in 15 minutes. Simcenter Amesim library manuals normally do not contain examples including components from other optional libraries.

This example, however, contains components from the Hydraulic Component Design library. The reason for this is that industrial usage of hydraulic Discrete Partitioning shows that the examples that achieve the greatest CPU time improvement are mixed Hydraulic and Hydraulic Component Design systems.

Figure 19: Model ready for Discrete Partitioning

The discrete hydraulic component submodels are as follows:

Figure 20: Center piece component submodels

The parameters concerning the number of nodes of these are as follows:

Parameter title	HLLWPIPE0-1	HLLWPIPE0-2
Selection of number of nodes	Maximum user def.	Maximum user def.
number of nodes	3	9
diameter of pipe	10 mm	10 mm
length of pipe	0.3 m	1 m

This makes the discrete cells in the shorter pipe roughly the same length as those in the longer pipe. There are 9 frequency modes in the long pipe and 3 in the short. Running the system on one workstation for 1 second gives a CPU time around 2100 seconds. We will take two reference results.

• The flow rate from the pump: Figure 21: Pump flow rate

• The pressure at the pump outlet: Figure 22: Pump outlet pressure