Question of Lesson2_WaterGEMS. wtg (II)

Question

Hi 
 Please have a look at P. 28 of your 'Quick Start Lessons' (HAMMER). The graph is attached here in more details. 
 I have some problems understanding blue graph:

1- Two maximum pressures have occurred between 15-23 seconds. These are because collapsing vapor pockets at high points (J1 & J3). Vapor pockets create while first decreased pressure wave moves upward and collapse while pressure wave return backward from reservoir to pump. Total length of pipes is 1855 m, wave speed is 1250 m/s. Then T will be 1.484 sec (=1855/1250). 
 It means that less than 3 sec (10-13 sec after running the model) is required in which wave returns to the vapor pockets to collapse them. Why both maximum pressures were created between 15-23 not 10-13 sec? 
 
 2- Pump begins to start after 25 sec. It have to have flow (even a little) after starting. But as Graph shows, there is not any flow until 28.1 sec after running the model. Why does it occur? 
 
 Thank you very much.

Jesse Dringoli · Answer

The hydraulics of a transient event can be quite complex. I recommend animating the Profile path ("path 1" in this case) to get the best visual of what is happening during the transient. At 5.0 seconds, the pump shuts down, causing a downsurge wave to travel downstream. That reaches the downstream reservoir at about 6.5 seconds based on the wave speed, reflects and returns back to the pump at about 8.0 seconds. The pump speed is still decreasing at this point which has an influence. At about 8.0 seconds, the pressure has dropped low enough at high point J-1 that a vapor pocket starts to form. 
 As the separated water column reaches a flow of zero at around 13.0 s (see time history of flow at P2:J1), the vapor pocket stops increasing and starts to collapse (decrease in volume). At 15.5 s, the vapor pocket is fully collapsed, the adjacent water columns collide, causing an upsurge. The vapor pocket collapse occurs right nex to the pump and quickly reflects off of it (the pump check valve is closed at this point). As that upsurge wave travels toward the downstream reservoir, reflection from the other branch of the network (connected to J2) reaches J2 and a decrease in pressure occurs (16.5 s). By 17.0 s, the first upsurge wave from the vapor pocket collapse reaches the downstream reservoir, then reflects back. As that wave is traveling back, it interacts with the downsurge wave that came through P8 to J2 and combines, causing a high magnitude decrease in pressure (see animated profile between 17.2 and 17.6 s). This travels in opposite directions, reflects off pipe ends and combines with other waves, eventually dropping the pressure low enough at the J3 location to form another vapor pocket. As you can see, the hydraulics can be quite complex. 
 As a result of the above, the second vapor pocket at J3 eventually collapses, causing another upsurge that travels in both directions, reflects and interacts with other waves, and so on. 
 Regarding why the flow does not start to increase until a few seconds after the pump speed starts to increase - the reason is because as the speed is starting to increase, the pump at first cannot overcome the discharge head already presnet in the system (essentially operating at its shutoff head - think of the pump curve scaled down at lower speeds). As higher speeds, the pump is able to overcome the shutoff head and starts to be able to pass a positive flow. See related wiki article: 
 Flow from pump is delayed after pump startup

Jesse Dringoli · Answer

1) In the Time History tab of the Transient Results Viewer, you can view a graph of hydraulic grade and flow on either side of the pump (pipe endpoints adjacent to the pump). Make sure these points have been added as report points in the calculation options. If you enter a number in the Report Period of the pump, you can also view a table of flow, speed, upstream and downstream HGL through the pump under Reports > Transient Analysis Reports > Transient Analysis Detailed Report. See more here: 
 
 communities.bentley.com/.../11111.how-to-graph-extended-transient-results-such-as-gas-volume-for-hydropneumatic-tanks-pump-or-turbine-speed-air-valve-extended-data-etc 
 
2) Currently the colors are identified by the Settings button > Profile Options > Colors tab. On my end, I am indeed seeing an issue when attempting to change the colors in version 08.11.06.113 on Windows 10. Are you seeing black colored cells, with a crash occurring when hovering over the color cells? (with the crash starting with "System.InvalidCastException..."?) I've logged this for our developers to look into for the next release. Please send me a private message with your contact information and company name (or "site ID" seen under Help > About) and I'll log a Service Request linking to the defect report, so that you can receive an alert when a new version is released with a fix. 
 
In the meantime, use this as your legend, for "Hydraulic Grade and Air/Vapor Volume"): 
 
Green line - physical elevations 
Black line - Initial conditions hydraulic grade 
Red line - Maximum transient hydraulic grade 
Blue line - Minimum transient hydraulic grade 
Red line at top - Max vapor/air volume

Jesse Dringoli · Answer

1) HAMMER saves results at pipe endpoints, so "P2:J1" shows the results at the end of pipe P2 adjacent to junction J1 and "P1:J1" shows the results at the end of pipe P1 adjacent to junction J1. 
 2) Yes, a negative flow means the flow is going opposite to the direction of the pipe orientation. So, if the flow is toward the "start node", it will display as negative. The same applies to the initial conditions (steady state/EPS) 
 3) P2:J1 as at the vapor pocket location, so as long as the vapor pocket is present, the pressure and hydraulic grade result will show as being equal to the vapor pressure limit - vapor pockets can only be present when this condition occurs. The pressure increase that causes them to collapse (decrease in volume) basically applies on the other side of the pocket, which in this example case is not exactly at P2:J1, so the results to not reflect that change in pressure. 
 4) Try animating profiles "path 1" and "path 2" to understand this better. Path 2 covers the path that goes down P-8. The upsurge wave that travels down P-8 reflects off the partially closed valve VLV1 (which has an operating rule that starts to close at t=10 s). It reflects back as a decrease in head, which then travels up to J-2 and into the other path seen in "path 1", which is why you see a decrease in head around =16.5 s. It may help to decrease the run duration and change the report times calculation option to report at all times, so you can see a more smooth/detailed animation. 
 5) I'm not sure I understand your question - see previous response for item 4 - animate path 2 to see what's happening. 
 6) Transient waves can indeed move without flow, but flow is generally required to dampen the waves (from headloss). Read more on wave propogation in the help and in chapter 13 of our AWDM book.

Jesse Dringoli · Answer

1) There are times when the results at either side of a node may not be equal. For example this can happen if a vapor pocket or air pocket forms exactly at the node location (such as the case with J1 in the lesson2 model) - when the pocket changes volume, the change on one side of the node may be different from the other, and the resulting head and flow may be different. 
 In the case of the base scenario in the lesson2 finished model, as the vapor pocket at J1 starts to increase in volume, flow moves away from it in a downstream direction based on the hydraulics of the system, so a positive flow is seen at P2:J1, whereas zero flow is seen at P1:J1. Conversely when the pocket starts to decrease in volume, the downstream water column moves back toward the pocket, so a negative flow is seen in P2:J1 whereas the upstream side is still at rest (zero flow at P1:J1). 
 6) Yes - this is demonstrated in Sample1.wtg. Also see this article: 
 How can I model a frictionless pipe?

Jesse Dringoli · Answer

If you look at the headloss results field for those pipes, you'll see it's still a positive value even with only one decimal place considered. So, it looks like the small Darcy Weisbach e that you used is not small enough in this case. 
 
If I switch to the Hazen Williams method, I see that you already entered 99,999,999,999.0 as the friction coefficient. With that, the headloss is zero in the initial conditions (at one decimal place) and the D-W friction factors show as 0.000 in the Transient Analysis Output Log.

Jesse Dringoli · Answer

Yes, I agree that transient waves can be dampened over time due to their interaction with other waves, as demonstrated in your model. There's some information about this in the help topic "Wave Reflection and Transmission in Pipelines" and the section called "Wave Reflection and Transmission" in the AWDM book starting on page 589.

Jesse Dringoli · Answer

Regarding #1, after some further discussion with a colleague, it should be clarified that energy will not be lost in a frictionless system and the reason for the wave dampening in the case of Lesson2 is in part due to the exit losses that occur at demand points. Basically the junction demands in the model introduce a loss (between the pipeline pressure and atmospheric pressure) which can dissipate the transient energy. 
 Also, the assertion that waves will persist without any dampening assumes pipes flowing full, but the example model in question experiences cavitation. 
 I have documented this here: 
 http://communities.bentley.com/products/hydraulics___hydrology/w/hydraulics_and_hydrology__wiki/27956.what-can-cause-a-transient-wave-to-dampen

Jesse Dringoli · Answer

1) I'll have to think about this some more and possibly consult with some colleagues 
 
2) The latter is correct. My most recent response overrides my initial response. As seen in my most recent response, I discussed this with a colleague who confirmed that energy will not be lost in a frictionless system. See answer and wiki article in my previous post. 
 
3) After further investigation, we confirmed that cavitation will not cause energy loss in a frictionless system. I've added this to the previously mentioned article.