Hammer Column Separation Sewage Rising Main

Question

Hi All, 
 I'm modelling an existing DN450 AC Class15 foul rising main in Hammer. The total rising main length is approx 2.9km, however the highest gradient point is located at approx 2.1km. In order to simulate the highest gradient point in Hammer, I simulated it as a reservoir at that location. In accordance with as built drawings, there is only one air valve along the route at approx 1.1km chainage. Considering that we could not get any info regards the existing air valve we have assumed that it is a 2" combination air valve. We also had to add a new proposed air valve just before the connection to the existing rising main. The existing pumping station capacity will be upgraded from 100l/s to 162l/s. 
 What I was able to observe during the transient simulation is that a column separation will occur at the air valve location and that a subsequent column slam will cause significant upsurge. So my question would be (disclaimer - I'm quite new to the surge analysis) is there a chance that in reality it would not actually come to the column slam event as the column on the right hand side of the air valve would not have anything to bounce back against (it would just continue to discharge at the discharge manhole while the cavity expands-air valve intakes air)? Am I wrong in thinking this? 
 
 Thank you and Kind Regards, 
 Oliver

Jesse Dringoli · Answer

Matija Oliver said: is there a chance that in reality it would not actually come to the column slam event as the column on the right hand side of the air valve would not have anything to bounce back against (it would just continue to discharge at the discharge manhole while the cavity expands-air valve intakes air)? Am I wrong in thinking this? 
 I have read this many times and I think I may understand what you're asking, but I'm still a bit fuzzy. Are you expecting that the downstream water column will continue to move away from the air pocket, draining out part of the downstream pipe? The air and vapor pocket modeling in HAMMER is an estimate based on the assumptions and limitations documented in this article from our Wiki: Assumptions and limitations of tracking air or vapor pockets in HAMMER 
 My guess is that if you animate the profile path in the Transient Results Viewer, after the air pocket forms, the downsurge wave continues past the air valve, reflects off the downstream endpoint, and then travels back toward the air valve as a higher pressure wave. I see that the downstream boundary condition seems to be at an elevation that is roughly equal to or maybe slightly higher than the air valve location. By the end of the simulation (you may need to extend it if needed), do you find that some air is let back in once conditions settle down to the final steady state? 
 It may help to look at the Time History tab of the Transient results viewer - graph flow as well as air volume on either side of the air valve, which may give some further insight. 
 If this does not help, please elaborate on your question, provide further screenshots (or video) of the animated profile path and graphs, or provide a copy of the model with any steps needed: Sharing Hydraulic Model Files on the OpenFlows Forum

Jesse Dringoli · Answer

Thanks for providing the model file. Here is my interpretation of the transient simulation from the graphs and profile animation in the Transient Results Viewer: 
 1) When the pump shuts down the downsurge wave front reaches the air valve location, causing subatmospheric pressures, so the air valve opens and starts letting air in. 2) At around 60 seconds, things have settle down to the point where the model tries to start reaching the new steady state/equilibrium point. Since the downstream reservoir is at a slightly higher elevation than the air valve (87.57 m vs. 87.10 m), the downstream water columns starts to flow in reverse back toward the air valve (you can see this in a flow time history for S705:AV-1). The upstream reservoir is at a lower elevation but the water column on the upstream side of the air valve in question does not move in reverse toward the pump (only slight movement due to elasticity) because the pump has the check valve option enabled. 3) From the movement of water from the reservoir (87.57 m) to the air valve (87.10 m), the air pocket is eventually expelled completely and an upsurge results from the water columns colliding. The air valve's outflow orifice diameter is set fairly large (50 mm) which influences the rate that the air is expelled. Using a smaller diameter can help slow this down, reduce velocity and therefore the upsurge. (refer to the Joukowsky equation) 4) The upsurge waves reflect back off the endpoints, meet again, cause a low pressure "downsurge" wave, which introduces air at the high point again. 5) Conditions settle again, air is released, an upsurge occurs, and this repeats several times (conditions do not fully settle down until near the end of the simulation or even beyond)

Jesse Dringoli · Answer

Matija Oliver said: It just seems counterintuitive to me that that column of water is still present, I would think that a lot of the volume would discharge through the gravity system. Considering that there is approximately 2.2m3 of air volume intake in the pipeline(approx 14m of the pipe would be filled with air), and as you noted majority of this volume will go into the downstream side of the pipe. Do you think that anyway the 0.47m difference would be enough to return the column back to the air valve elevation and cause the collision? 
 Part of the problem is with the assumptions and limitations in HAMMER's tracking of air and vapor pockets, as described in this article previously shared. Since HAMMER does not track the air/liquid interface, the boundary water level (HGL) remains the same at both ends, which is why the water columns still travels back toward the air valve due to the 0.47 m difference in elevation between the reservoir (or D2A elevation) and the air valve elevation. Meaning, when air is introduced into the air valve, by default HAMMER does not track the air pocket traveling down the pipe and causing the water surface elevation to drop as it drains down the pipe, and when air is introduced into the D2A, it does not track that air going down the pipe and the resulting drop in HGL from the water surface elevation going down. So, the water column continues to move toward the air valve, eventually expelling the full air pocket and resulting in an upsurge from the velocity of that column. 
 In reality if the reservoir location is an open end, some of that downstream end may drain down but only enough to expel the air pocket out of the air valve in question. Depending on the timing of the waves and water column elasticity, the 0.47 m drop in water surface elevation may or may not be enough to expel the full air pocket. 
 The 2.2 m^3 volume of air that HAMMER reports is an estimate based on the aforementioned assumptions. In order to get a more accurate calculation of the volume of air introduced into the air valve, you would need to account for the air/liquid interface tracking along the pipe. This can be accomplish to some extent by using the Extended CAV calculation option. Note that this option only enables the tracking of air in the adjacent two pipes and only if the adjacent nodes are at lower elevations than the air valve. So, you may need to combine a few pipes (perhaps using Skelebrator, or manually) if the air/liquid interface needs to extend beyond that (the slope of the pipes seems fairly continuous for some distance so you might be OK doing this) 
 With this, you should be able to achieve a more accurate simulation of the volume of air that enters the air valve, and get a better sense of whether the drop in water level at the downstream end (D2A) would cause enough volume change to push out that entire air pocket. However, it still isn't an exact simulation because the air pocket at the D2A site does not work with the Extended CAV option as explained here . Additionally, I believe you said there is already uncertainty with the orifice size which will also skew the results. So, you may need to use a sensitivity analysis with the size. 
 
 Some engineering judgement may need to be employed here with these computer model assumptions in mind. Perhaps something conservative. Maybe you can consider that the results are telling you that there might be a problem here with air pocket slam in some conditions and air valve orifice sizes and that the pipeline may be at risk. You could then instead focus on mitigation strategies. 
 Matija Oliver said: When I replace the existing air valve with an air valve with surge protection (C80-SP) something strange happens. Even if I extend simulation time to over 4h, the air never gets expelled as per screengrab below. This would be more in line with my thinking that the downstream water column would not have enough momentum to expel that air and it would just stay in the pipe. What are your thoughts on this? 
 In order to troubleshoot this, I would need more information about how you configured this "C80-SP". What model element did you use? (A surge valve? What settings?) How exactly does this device operate? Can you provide a new copy of the model with steps to reproduce? 
 Transient simulations will be very sensitive to the way that elements are configured, so it is important that the configuration is correct to avoid skewing the results or instability (which appears to be occurring in this case). If you think the setup is OK here, I would try a smaller calculation timestep, review other user notifications and ensure that you're using the latest version of HAMMER (10.03.02.75).

Jesse Dringoli · Answer

Matija Oliver said: I simulated the Extended CAV for the air valve but wasn't sure how to read the report as it was explained in the article. There are two results for the Air Valve 2 with the head (96.7m) higher than the maximum pressure envelope observed in the transient result viewer (approx. 94m). Could you please help me understand the results below? It seems that the air volume has not been captured correctly as it shows 0. 
 The results you saw are for the pipe endpoints on either side of the air valve. When using the Extended CAV option, the air pocket volume can in part be attributed to the adjacent pipes and not just the endpoint next to the air valve. The following articles explain more about reviewing the results: 
 Modeling Reference - Air Valves (see section "tracking of air pockets" 
 Assumptions and limitations of tracking air or vapor pockets in HAMMER 
 There are two main options for viewing Extended CAV (air-liquid interface tracking) results: look at the per-timestep details in the "Output Log" text report or animate the profile path (consider a "Detail" profile view). Other general air valve reporting details are also available by adding a number in the "report period" field in the air valve properties then looking at the bottom of the Analysis Detailed Report. Lastly you can see the min and max volume in the "Vapor and air pockets" text report. 
 As a side note, in the new model revision you sent me in HAMMER 10.03.02.75 the results are showing that when positive pressure returns, the air pockets are not being released when using the new custom air flow curve that you specified. The free air outflow rate is very small for the air outflow curve; 39 L/S at a pressure of 2 m compared to 100 L/s at 2 m for the "sample 2in outflow" curve that you can import from the engineering library (2 in = ~51 mm). Additionally, the air outflow curve's pressure is not in ascending order with respect to the outflow rate. As the pressure increases, the air flow starts to increase, then it decreases, then increases again. This does not seem to match with the "air release" graph in the PDF you attached. 
 However, from extensive testing this morning, I have found that the custom air flow curves may not be working well in this model when combined with the Extended CAV option. If you turn the option off OR change to diameter-based sizes the air is expelled. I would suggest estimating the equivalent circular orifice size and using the diameter method if you need to use the Extended CAV option in this model. TO see the air-liquid interface drop at the endpoint of the system, you could consider place an air valve there as well (at the D2A location, but with the D2A still included at a slightly lower elevation, connected to the air valve via a short pipe). As mentioned in earlier replies, the Extended CAV option would not track the drop in liquid elevation at the end of the system with the D2A. 
 I would again suggest also taking a step back and considering what I had mentioned in my previous reply: 
 Jesse Dringoli said: Some engineering judgement may need to be employed here with these computer model assumptions in mind. Perhaps something conservative. Maybe you can consider that the results are telling you that there might be a problem here with air pocket slam in some conditions and air valve orifice sizes and that the pipeline may be at risk. You could then instead focus on mitigation strategies. 
 As for the "C80-SP" valve stability problem - that appears to be due to the custom air flow curve, possibly in combination with the Extended CAV option as mentioned earlier. Try using the equivalent diameter approach. Your description of this air valve and the literature provided seem to suggest that it would effectively operate as a triple-acting air valve, where the inflow orifice diameter would be set based on the capacity during air intake, the "large air outflow" size would be based on the capacity during initial outflow, and the "small air outflow" size would be based on the capacity when the float rises up just before all the air is exhausted (I assume that the float will cause some reduction, if not then this could be a double acting air valve).

Jesse Dringoli · Answer

Matija Oliver said: For some reason I don't have any additional info below " *** SNAPSHOT OF EVERY END POINT AT START OF TIME STEP 2 ** *" table. Please see attached report, actually in my report it refers to TIME STEP 66066. 
 I opened the text file you attached and I see the section of interest and the table below it. If you have Notepad++ it starts on line 471 and the details of interest start on line 548. I do see that the usual details are not visible - try entering a number in the report period field in the air valve properties, make sure they are added as report points in the transient calculation options, close the text file, re-compute the model and try again. If this does not help please send an updated version of the model and identify any special steps. Note that the profile animation approach tends to be a better way to visualize anyways, so you may not want to pursue this. 
 Matija Oliver said: I tried to mirror the "S" shape of the curve as close as possible.It appears like the curve is bent inwards between 0.1 and 0.3 bar. 
 Perhaps I misunderstood the literature but that "S" shaped curve seemed to be referring to a certain size inflow orifice, whereas the outflow rating curve appeared to be shown in a separate table to the right. 
 Matija Oliver said: Only concern I have is that the air inlfow curve is not the same as the full 80mm orifice size(red curve) as it can be seen on the data sheet. So I wonder how to assess the air inflow oriffice size? I might do several iterations and try to match the results when the Extended CAV is set to false. 
 Yes I recommend the sensitivity analysis approach, then settle with a conservative assumption if the results are sensitive. 
 Matija Oliver said: I guess that the extended CAV does not have as big impact on the results quality as air flow calculation method, is it correct to assume that the more "correct" and precise option would be the one with the air flow curves. Even though that it is less conservative. Would you agree with me? Or is it as always safer to go with the more conservative result? 
 This may depend on the situation. In some cases the air-liquid interface tracking may be more important than precise control over the air flow curve. I would suggest first trying a sensitivity analysis with the diameter vs. air flow curve with the Extended CAV option turned off, to help you decide if the air flow curve is significant to this analysis. If the "S" curve gives you trouble, test the sensitivity to a smoother/linear estimated curve. 
 Matija Oliver said: I definitely agree that the results are telling me that there might be an issue. My primary mitigation strategy will be to replace the existing air valve with the C80-SP which will limit the air outflow and subsequently mitigate the severity of the column slam. It has been noted that there will still be some vacuum pressure occurrence along the rising main and I will probably need to mitigate this considering that the rising main is 35y old AC main. 
 Maybe you could also look into the impact of installing a PSV on the downstream side to raise the initial HGL, which might help prevent negative pressure upon pump shutdown. This may result in inefficient pump operation though.