Comparison with Hammer¶
During the development process, we have consistently compared TSNet results with Bentley Hammer [HAMMER] using different networks and many different transient events. In this section, we present the comparison between TSNet and Hammer results. The Hammer models used to generate the events is also included in the GitHub example directory (https://github.com/glorialulu/TSNet/tree/master/examples/networks) for the user’s reference.
Tnet 0¶
We first show the comparison for a simple network, consisting of one reservoir, two pipes, and one valve, as illustrated in Figure 23. The wave speed for both pipes is \(1200m/s\), and lengths and diameters are given in the figure. The transient event is generated by closing the end-valve at the beginning of the simulation during 2s; thus, the flow rate at the end valve decreases linearly from \(0.05m^3/s\) at \(t=0s\) to \(0m^3/s\) at \(t=2s\) and remains zero thereafter. Figure 24 (a) shows the flow rate through the valve, and Figure 24 (b) presents the pressure transients generated at node N-1 during 60s simulation period. The solid line represents TSNet results and the dashed-dotted line shows Hammer results. These results indicate a perfect match between TSNet and Hammer simulation results for this simple network.
Tnet 3¶
We then show the comparison between TSNet and Hammer results for a more complicated network, Tnet3, for three different transient events:
- Shut down of PUMP-1,
- Burst at JUNCTION-73, and
- Closure of VALVE-1.
The results for the three transient-generating events are shown in Figure 25, Figure 26, and Figure 27, respectively.
- We would like to note beforehand that we do not expect to obtain the exact same results from TSNet as Hammer since different numerical schemes were implemented, such as wave speed adjustment, pressure dependent demand, and boundary conditions.
Pump shut-down¶
Both TSNet and Hammer are utilized to simulate the shut down of PUMP-1. The time step is specified as 0.002s in both software. Figure 25 reports the pressure change with respect to the nominal pressure at multiple junctions, where the solid lines represent TSNet results and the dashed lines show Hammer results. TSNet and Hammer results are very similar to each other in terms of attenuation and phase shift throughout the 20s simulation period, despite slight discrepancies, which can be explained by the different wave speed adjustment schemes and boundary condition configurations adopted by the two software.
Burst event¶
Aburst event was simulated at Junction-73 using both TSNet and Hammer. Figure 26 reports the pressure change with respect to the nominal pressure at multiple junctions, where the solid lines represent TSNet results, and the dashed lines show Hammer results. It can be observed that during the first transient cycle, i.e., around 0-8s, TSNet and Hammer results exhibit very good agreement with each other. Although the discrepancies increase a bit in terms of attenuation and phase shift during the latter period, the overall match is satisfactory considering that different time step and wave speed adjustment schemes are adopted in the two software.
Valve closure¶
Both TSNet and Hammer are utilized to simulate the closure of VALVE-1. The comparison of the results is presented in Figure 27. Again, adequate resemblance can be observed between the TSNet (a) and Hammer results (b). Considering that pressure transients are of smaller amplitude and more chaotic, the results are presented in two separate plots with same scale for clarity.