In the example SNFAIL from the June 1994 News, would you use the same time history method for multiple events? For example, say the turbine tripped (you have the ramp down curve) at t=0 (took a reasonable time to ramp down), then the valve at node 30 closed some 10ms after the turbine trip (or started to close). How would you set this up?

That SNFAIL example had four overlapping events analyzed together, each with their own load vector and timing. You can include several events. There is no formal limit on the overall duration of the analysis but you may run into memory problems if you try to run a very long or a very detailed (short time step) analysis.

We are seeing more users collecting time history data from transient hydraulic analyis packages and pushing those results through CAESAR II. My suggestion is to start simple and build from there. Don't try to hit the home run on the first pass.

Ah - good point about the elbow (impact) events being separate (and demonstrated). I'd been thinking about the valve closing being the 'event' and wasn't sure what limit on 'events' there might be. (Not, apparently, good.)

Yes, that's what we're getting ready to do - push a transient hydraulic analysis on thru CAESAR. Relatively simple side first - pump, check valve, two actuated valves....I'll build it up slowly, adding one 'item' at a time.

Thanks.

Dear Coade Members,

This is my first dynamic analysis on Water Hammer, i have some queries related to it, can u please clarify....

I am following the water hammer procedure as in Caesar-II application guide.

In our case, Pipe size 10" Sch. 40, Velocity 2m/s, design pressure 50 barg and temperature is 85 Deg C.

The pipeline starts from source well and connects to the pump suction. The pipeline length is approximately 1 km between the connections.

There are many bends in our problem. The elbow pairs we are considering in our problem is the first 300 meter pipeline from the pump suction point and we considering 4 elbow pairs for the water hammer analysis.

In Caesar-II guide it is mentioned that two elbow pairs at node 45 to 75 and 90 to 110 are considered for water hammer. To calculate the time duration of the load between the elbow pair, the length L of 90 feet is taken for 45-75 elbow pair and 75 feet is taken for 90-110 elbow pair.

My query is why the reference length shall not be taken from the source of water hammer, in my case we are considering the source of water hammer as the pressure control valve.
(i.e.) let assuem L1 be the length for the first elbow pair from the source and the load duration may be t=L1/c.

To calculate the length for the second elbow pair, why not the length shall be taken from the source instead between the elbow pairs (ie) L2 = L1 + length of second elbow pair. Also the unbalanced force might have reduced some how due to the impact in the first elbow pair.

Also why we are going for a separate spectrum for each elbow pair. It seems that the force will be high when the pressure wave hit the first elbow pair and get diminishes due to friction and successive elbow pair when it travel along the pipeline.

Also in Caesar-II example, for generating the spectrum table the input seems to have some error (i.e) for force, the value is used as 1.0.

Instead of providing a big anchor support to limit the displacement, shall i provide water hammer suppressors...

Please clarify........

Hello Sillyman,

You are exactly pointing the difference between Time History and Water Hammer analysis. In the real world, water hammer loads apply at the same time on albow-elbow pair segments with time lags and in this situation you should use Time History analysis. In fact, if you perform Water Hammer analysis, you isolate albow-elbow pair segments and evaluate system responses one by one.
I recommend you to do SNFAIL training job. It will help you to understand the Time History Analysis.
TO eliminate pressure surge and water hammer loads in pipelines and piping systems, you can place surge tank or PSV at appropriate locations. BOSFluids and PIPENET Transient are good programs for these kind of problems.

Dear Richard Ay,

For which case we must give the priority for water hammer loads (i.e.)

Case-1 - During the pump trip off, there was a water hammer effect inside the pipe and the pipe before and after the elbow was displaced to approx. 125 to 150 mm in lateral direction. This is the data we received from the operation people.

Case-2 - Based on the Caesar-II application guide, the dynamic model was done, where the displacement at that paticular elbow is found to be only 40-50 mm and the stresses are seems to be ok.

But when i manually input the displacement of 150mm at the paticular elbow point as per operation data, and do the analysis by static equivalent approach, the pipe stress is more than the allowable limit....

My query is for water hammer analysis, the priority must be given to the operation data or to the theoritical calculations.....

Please clarify...

I suggest you review the model and the loading data for your "case 2". You should have been able to get close to the field observations defined in "case 1". Something is not right.

Sillyman,

Firstly, refine your model by adding more realistic supports. If reqired, add steel structures and connect supports to them. Existing supports' gaps must be used instead of those specified during design phase.

Secondly, In dynamic analysis module, go to "Control Parameteres" tab and in "Static Load Case for Nonlinear Restraint Status" setting select OPE load case to consider any support lift-off or closed gap. Then tweak "Stiffness Factor for Friction" setting to obtain a real response of piping system to applied waterhammer load, start with 1, 10, 100, .... for each value run analysis and check displacement results. Also be sure to Include Missing Mass correction.

If it doesn't work, send CAESAR II static and dynamic input files.

Dear Richard Ay,

In Water Hammer / Slug flow (Spectrum) - Control parameters

What is the difference between Lumped mass model and Consistent mass model?

The analysis time and memory space seems to be more for consistent mass model case...

Please clarify....

This should be discussed in the "Help text" for that item ...

The Lumped Mass approach only has values on the diagonal of the mass matrix. The Consistent Mass approach has a fully populated mass matrix.

Dear Richard Ay,

Is it is preferred to go for Lumped mass approach (Water hammer case) since to run the dynamic analyis using consistent mass approach it took 4 hrs.

The consistent mass approach is (typically) more accurate. However, if you have sufficient node points between restraints and bends, the lumped mass approach should be adequate.

Dear Richard Ay,

To reduce the effect of water hammer one of the way is to have long straight length of the pipeline from the water hammer source so that the hammer force shall be diminishes due to the friction force.

If i am wrong please clarif.

In my case, after some 4 elbow pairs, in the fifth elbow pair i inputted the hammer force in the DLF spectrum generator as (length between elbow pair is 200 meter, hammer force is 73313 N and Speed of sound is 1257 m/s)

0 ms - 0 N
5 ms - 73313 N
159.1 ms - 73313 N
164.1 ms - 0 N

In the DLF spectrum generator, it seems that the hammer force will be developed within 5 ms and it continues till 159.1 ms and after 5 ms the hammer load will come to zero force.

From the DLF spectrum it is seen that as the straight length between the elbow pairs increases, the effect of hammer load will be more or the displacement at the end of elbow pair will be high.

Please clarify...

Dear Richard Ay,

Since the water hammer analysis using time history analyis seems to be conservative than the force spectrum (water hammer analysis).

Why dont Coade shall have only a time history analysis for water hammer?

Also the input required for the force spectrum and time history analysis are almost the same.

Please clarify...

Force response spectrum analysis is not as exact as time history. But time history also requires more exact data. Use time history when you are sure of your timing, use force response spectrum when you can only estimate.

Dear Richard Ay and Dave Diehl,

Please clarify the following queries in Time History Analysis of SNFAIL,

1. What is the normal fluid flow direction (i.e.), whether the fluid flow is from node 125 to 5 or from node 5 to 125? From the newsletter June 1994, it is seen that the normal flow direction is from node 125 to node 5.

2. Why the force at node 30 (i.e.) at the start of Angle valve is considered? There seems to be no force acting at node 30.

3. If the normal flow direction is from node 125 to 5, then due to sudden closure of the angle valve, the water hammer fluid direction may be from node 30 towards node 125. Then the direction of water hammer force at node 100 shall be in (+X) Direction and at node 100 it shall be in (+Y) Direction and at node 120, it shall be in (-Z) direction.

4. Then why the forces are listed like below in newletter June 1994,

(-X) load @ node 30 - (Will there be any force acting at the angle valve).
(-Y) load @ node 100 - Why not the direction of force shall be in (+X) Direction.
(+Z) load @ node 110 - Why not the direction of force shall be in (+Y) Direction.
(+X) load @ node 120 - Why not the direction of force shall be in (-Z) Direction.

Please clarify....

Dear Richard Ay and Dave Diehl,

With reference to June 1994 Coade M.E.N., under "Establishing Time History Input - An Example", Pg. # 9, it is stated that

"The point of application and direction: At the start of each straight run of pipe upstream of the valve with the load pointing downstream.

(-X) load @ node 30
(-Y) load @ node 100
(+Z) load @ node 110
(+X) load @ node 120"

Why the forces are considered acting in the downstream of the valve or from the start point of the valve.

Since the water hammer direction is from the node 30 to node 125, the force also must be along the direction of the water hammer.

Please clarify....

Dear Richard Ay and Dave Diehl,

I am waiting for your reply....

1) Yes, the flow is from 125 to 5.

2) For this sort of analysis the actual location (which node) doesn't matter, as long as the load is applied somewhere in that straight run of pipe. What we're looking at is the unbalanced load in that straight segment.

3) This explained in the newsletter article. The downstream end of the run (in question) sees a pressure of "P + dP", while the upstream end of the run sees only "P". Hence the load, in this instance, is applied in the direction of the flow.

4) Please see #3 above.

Dear Richard Ay,

So from your above suggestion, it seems that this type of force application is only for time history analysis.

For force spectrum analysis, the applicatiion of force shall be as i queried earlier.

Thankyou...

Dear Richard Ay,

While running the time history analysis i am facing the problem of runtime error. Please find attached sketch for additonal information.

Please clarify...

What version and build are you running?

Can you run the RELIEF example, both Statics and Dynamics?

Can you send your job file in to TechSupport so we can run it here?

Mr. Richard, Mr. Dave and Mr. Farhad,

I am new user of pipenet. Please clarify the following query,

1. When i try to interface the pipenet (.FRC file) with the Caesar, it display an error message as "The file specified was not a pipenet file, was not in the correct format, or it contained no pipe segments". Please clarify how to rectify it.

2. Is it is a necessary requirement that model of pipenet shall exactly matches with the model of Caesar-II such as up and down of pipe elevations, location of bends / elbows, length of piping system, branch connections and locations etc. Please clarify it....

Dear Paldex,

Keep in mind that you should generate simple force in PIPENET (uni-directional force). Then, when imported to CAESAR II you can define the direction of force. CAESAR II automatically generates dynamic water hammer input file from FRC file. If you want to perform time history analysis, you should instruct CAESAR II to not generate dynamic input file, but you should create TH input file by yourself.
Remember, PIPENET model doesn't have geometry in XYZ coordinate system. There are only length and elevation change from point to point.
The location of force that you define in PIPENET, is a distance between 2 nodes along the direction of pipe.

Dear Farhad,

Thankyou for your reply.

From your information, i have the following doubts,

1. Can i allow caesar-II to generate dynamic input file, when i want to perform force spectrum method.
2. If i want to do analysis using time history, then the interface of pipenet with Caesar-II is not possible ?
3. My Caesar-II model is a little bit complex, and i want to analyse the system for water hammer case. We are using Pipenet transient module to calculate the unbalanced pressure effect and force at bends. In our model we have many bends from the source of water hammer (ESDV valve), we are concentrating on the bends near from the source (ESDV). The piping system contains ups and downs. But i dont find the option of drawing 3D in the pipenet such as ups and downs, so i am modeling in a single plane without elevation changes. Is this way is ok to calculate the unbalanced pressure.

Please correct me, if i am wrong.

Thankyou...

Dear Paldex

You can do time history analysis for water Hammer case in caesar-II. Please follow below points.

1. You must have force time data for all nodes in one file. Firstly you have to extract force time data to make saparate force time data files for each node and save the files in same folder where .CAESAR II file exists.
2. Then open Dnamic analysis window and choose Time history type analysis.


Please find the attached document for time history.

For Pipenet, i think you have to model in 3D to do accurate analysis.

For any other query, please feel free to ask.


Surjeet

Dear Surjeet/Dave/Farhad,

I am doing water hammer for a pump trip case using Caesar-II dynamic water hammer spectrum method. In which the stresses and the displacements are within the allowable limit. My query is whether the pump nozzle load must be checked to within the allowable limit as per API-610 or shall be ignored for dynamic transient case. Since the dynamic load at the pump nozzle point is too high.

Please clarify...

Hi Mr.Dave/Richard,

Please clarify the Time history - DLF spectrum generator doubts,

For details please refer the attachment - Time History...

Thankyou...

You can't mix frequency domain analysis with time domain analysis - they are completely different. When working in the time domain, you can specify your loading pulse (in seconds) directly - this will be used to define the loading on the system in a time history analysis. When working in the frequency domain, you use the "DLF Generator" to convert your time pulse to a DLF curve, as a function of frequency. This is a one-way conversion, there is no way to go back to your pulse from the DLF curve.

Arbitrarily changing the axis of the loading curve from frequency to time, or time to frequency is invalid. The curve is the curve and it has fixed characteristics.

Hi Richard Ay,

Thanks for your reply and it almost solved 95% of my doubt, so my final requisition for entering the time vs load for time history analysis. The following are the way i think i must enter for defining time vs load,

1. Actually to enter the time vs load for a spectrum (for ex. spectrum name "THF1330" as in my earlier post"), "Enter/edit spectra data" must be opened to enter directly, but it is not possible without the creation of "DLF-Spectrum generator" so for that, firstly open the DLF-Spectrum generator and enter the time vs load, where we find the a graph for Freq vs DLF. Save it and click ok.
2. In the dynamic input main spread sheet, under the "Range" option change the frequency to time.
3. Now open the "Enter/edit spectra data" where you find the spectrum with name for ex. "THF1330" what we defined through spectrum generator.
4. In which, the frequency is changed to time(ms) but the values will remain same for the frequency. So delete all the values and freshly enter the "time vs load" in that one. Finally click done.
5. This is how, i am defining the time vs load for time history analysis.

From my point of view the above way to define the time vs load for time history seems to be strange, if i am wrong please correct me or if there is any easy way to define it, please share it...

Thankyou.

Hi Richard,

I am waiting for your reply .......

1) On the "Spectrum Name" tab, define the name, axis types, and axis interpolations.
2) You should now be able to directly enter the pulse data points. Do this, don't use the DLF generator - that isn't what you want.
3) Define the Force Set(s)
4) Define the Load Case
5) Setup the Combination Case if you need it.
6) Address the Control Parameters
7) Run

Thankyou very much Mr.Richard....

Thankyou very much Mr.Richard....

Dear All,

Most of the examples I found for Spectrum/Time History analysis are for valve is suddenly closed. I try to find an example for liquid entering an empty pipe but not available. For valve closed suddenly case, we are calculating the speed of sound in order to know the t. But for pump start up case where there is no liquid inside the pipe, do we have to calculate the speed of sound too?

Appreciate your advice.

Thanks.