Pressure increasing in pipes due to water hammer

Question

Hello, 
 
 I want to modelize a water hammer in a facility : I modelize it with 2 hydropneumatic tank linked with pipes. I have 4 TCV valves with a transient operation (the closure time is about 0,2s) to create the watter hammer. The velocity is around 2 m/s in the pipe and the difference of pressure between the 2 tank is around 1 bar. When the watter hammer is created, the pressure increase to 18 bars. I have several questions : The problem is that it's not equivalent to the reality. The increasing of the pressure is around 5 bars. Is it an establishment of the pattern problem ? I have also a problem with the time of closure of the valve, because when I increase it, it doesn't affect the pressure, whereas it should be. In the software bentley hammer, is it the relative or absolute pressure in the time history ? Thank you in advance. 
 Regards

Scott Kampa · Accepted Answer

Is this an actual system, or are you basing the comparison on hand calculations? HAMMER doesn't use the Joukowski formula directly; it uses Method of Characteristics (details can be found in the Theory section in the Help documentation). So if you are using this model as a way of comparing results to hand calculations, it might not be a surprise that the results are different. The Joukowski formula can be used as an approximation, but it is not ideal for all modeling cases. 
 Even if you are using hand calculations, you may want to look at the pipe lengths in the HAMMER model. HAMMER will adjust the lengths of the pipes so that a pressure wave will move from one end of the pipe to another in even intervals with the time step. This might also account for some difference. You can see the length adjustment and the adjusted pipe lengths used by HAMMER in the pipe flextables. You may need to add these fields using the Edit button in the flextable. 
 Otherwise, there are some general things you can try in terms of the setup of the model. For instance, you have three hydropneumatic tanks, two of which appear to be basically acting as reservoir. You can try replacing these two with reservoirs to see if there is any stabilizing of the results. 
 Second, you have a check valve that inhibits flow from the other hydropneumatic tank. This will make it so that flow from the tank cannot help with transients that may occur. In addition, this tank is fairly small. A larger tank may allow for better protection. 
 Lastly, a valve closure of 0.3 seconds is not very different from 0.2 seconds. You may find better results if the valve closure was 2 seconds instead. 
 Regards, 
 Scott

Mal Sharkey · Answer

Hi, 
 I took a look at your model and have a couple of comments / observations: 
 1. It appears to me that you are using unrealistic valve discharge coefficients. They are currently set to 10 m^3/s/(m H2O)^0.5, which means for a headloss of 1m you could get 10,000 litres per second through these 52mm valves. This causes the problems that Jesse mentioned in an earlier response (please have a look at the link that Jesse posted for more info). 
 2. There is a significant transient at the start of the simulation, caused by flows rushing into HT-2. I doubt whether this occurs in your experimental system, so you may want to check the initial HGL in HT-2. 
 3. For your experimental results, the pressure fluctuations are occurring very rapidly - are you sure that the pressure transducer is capturing them all? 
 I was able to get sensible looking results by making two changes to your model. 
 - First, reduce the initial transients by changing the HT-2 initial HGL to, say, 7m. 
 - Second, change the valve discharge coefficients (initial and fully open) to, say, 5 l/s/kPa^0.5. 
 Try this and you will see that the max pressure in the system is about 5 bars (same as in your experiment). 
 Also, I recommend that you turn on the 'Generate Animation Data' option in the Calculation Options so that you can watch an animation of the profiles as the pressure changes over time. 
 Regards, 
 Mal

Scott Kampa · Answer

Hello, 
 For the first two questions, this may be related to the setup of the model. For instance, a very fast or instantaneous valve closure is more likely to cause a large spike in pressure. If you believe that the hydropneumatic tanks in the model should better mitigate the pressure increase occurring because of the valve closure, you may want to review your hydropneumatic tank settings and user notifications. For instance, if you are getting a user notification implying that the tank is empty, you may need to adjust the tank data. 
 Closing the valves more slowly may cause a smaller spike in pressure. Depending on the setup of the hydropneumatic tanks, you may see little in the way of transient behavior in the model results. 
 Here is a link to a TechNote on hydropneumatic tanks and how they are used in HAMMER: communities.bentley.com/.../modeling-reference-_2d00_-hydropneumatic-tanks.aspx 
 For the pressure reporting, HAMMER uses gauge pressure. 
 Please let us know if you have any further questions. If there are model-specific questions, it may be best to see a copy of the model. There are two options for sharing your model files on BE Communities. If you would like the files to be visible to other members, compress the files into a zip file and upload them as an attachment using the ‘Advanced Reply editor’ before posting. If your data is confidential, you can follow the instructions in the link below to send it to us via Bentley Sharefile. Files uploaded to Sharefile can only be viewed by Bentley. 
 communities.bentley.com/.../bentleysecurefilesupload.aspx 
 If you upload the model to Sharefile, please post here letting us know, so that we know that it is available. 
 Regards, 
 Scott Kampa 
 Bentley Technical Support

Jesse Dringoli · Answer

In addition to what Scott said, I would also check for vapor and/or air pockets. If this occurs and air is released too quickly (or a large vapor pocket collapses) it can lead to a severe "upsurge". Create a profile and animate it in the transient results viewer, paying close attention to what happens and how the system reacts. 
 Regarding not seeing a response as you change the closure delay; this could be realistic, but I would also check to make sure that you have the headloss coefficient (or discharge coefficient) entered in the valve properties. For the reason why, see the below support solution: 
 communities.bentley.com/.../12683.gradual-closure-not-occurring-when-using-valve-operating-rule-solution-500000058523.aspx 
 This could also explain why a valve closure may see a more severe transient response than expected. 
 As Scott said though, seeing the model will help us provide the most accurate answer.

Scott Kampa · Answer

Hello, 
 First, check the initial conditions results in HAMMER and compare them to the experimental setup. HAMMER uses the initial conditions as a starting point for the transient analysis. If the initial conditions are not the same, the transient results may be impacted as well. Results to check include hydraulic grade, flow, and velocity. The flows are fairly large, and a fast valve closure time can compound the results. 
 Second, is the experimental model really so short? The total length between the two hydropneumatic tanks that act as true boundaries is only about 8 meters. With the current setup, HAMMER is applying large length adjustments to pipes, increasing the smallest (about 0.4 meters) to a much larger relative length. In fact, all pipes but two are less than a meter is length. HAMMER tends to operate best with longer pipe lengths. This can impact the results as well. 
 Third, are both of the TCVs closing at the same time, as indicated in the model? This could compound the pressure spikes, so if only one should be closing, you may want to change this. 
 Regards, 
 Scott