Method of Characteristics or Bond Graph

[Simba1978]

Hi all

I am looking to purchase a hydraulics surge analysis software for my company, and have narrowed it down to Flowmaster and Amesim. One uses Method of Characteristcs and the other Bond Graph method... and of course both say their method is better

Does anyone know the advantages and disadvantages of each? and any feedback on the software mentioned

Thank you in advance

Simba

 

[BigInch]

I think you would do better to not look into the black box. Look to the package that offers the most convenience in the construction of models and ease of interpretation of the results coming out of the black box for the particular types of systems you are interested in studying.

For example, Winsys, the process simulator can do pipeline analysis, but IMO does not offer ease and convenience to pipeline designers in entering data and building typical flow models for pipeline and the liquids handling facilities. So for work light on process and heavy on liquid handling facilities, I would most definately and totally go with Flow-*whatever*. For work involving little liquid handling relative to the process load, great use WinSim. I don't have experience with your alternate Amesim, but it looks heavy on the systems modeling capability and light on the convenience of use for specific liquid handling systems. Its not to say it can't do it and do it well, but that's not my point. My point is how easily and efficiently the package will allow you to do your job of analysis, interpretation and flexibility of presenting the results. I think its probably going to be Flowmaster.

http://virtualpipeline.spaces.msn.com

"What gets us into trouble is not what we don't know, its what we know for sure" - Mark Twain

 

[Simba1978]

Thanks BigInch

Flowmaster does seem the more user friendly. But if Bond graph method is a better way to modelling transients in systems then i think accuracy of results overides any easy use benefits.

Can you advise me on the main benefits of each method?

Thanks again

 

[insult2injury]

How do you plan to validate the results if you don't understand the math behind the program?

I2I

 

[Simba1978]

I understand the basic theory behind each method, but i don't have the in depth knowledge to make a judgement on which method is best for modelling fluid transient behaviour.

validation against experimental results.

what are the limitations of Method of characteristcs and similiarly Bond Graphs? things like variable timestep and modelling discontinuities are some i have researched but other issues would be greatly appreciated

Thank you

 

[BigInch]

I think I can verify the method of characteristics is very accurate, if modeled correctly. It is a well proven technique, with almost fifty years of use as adapted to computer code and many more years of manual calculation before that. I don't see the quetion of which one is "better" as very relavent, given that you're going to model similar problems. The ease of constructing a model, appropriately to the answers and level of detail you want to get from the model and without making modeling or numerical data mistakes and interpreting the results far overweighs any consideration of calculation accuracy coming out of the box. Even if one method were to be more accurate than the other, any difference I doubt would amount to a hill of beans anyway. As with all computer simulations, if modeled incorrectly, or if mistakes are easier to make in one GUI versus the other system, you're sunk before you ever put the equations into the CPU. GO with EASE OF USE AND INTERPRETATION OF RESULTS! I can't possibly overstress its importance. Believe me on this one, or I'm sure later you will never forgive yourself. Been there, done that.

http://virtualpipeline.spaces.msn.com

"What gets us into trouble is not what we don't know, its what we know for sure" - Mark Twain

 

[sailoday28]

insult2injury (Mechanical)
I agree with you. Even those with some experience misinterpret results. With compressible flow, I have found many individuals using moc programs in which the individual and program cannot calculuate correct results.
ie, sudden rupture of a high pressure pipe line with compressible flow. You'd be supprised at answers for the instantenous pressure at pipe exit.

Regards

 

[insult2injury]

I'm curious about the experimental results. Why do you need the program if you will conduct experiments to tell you the same answer?

I2I

 

[Simba1978]

Thanks for the advice,

The point of purchasing the software is to reduce the number of experiments in the future, but we need to trust the results and comparing them to our past experimental data for validation is one way of building our confidence in any software.

Cheers

 

[btrueblood]

"Why do you need the program if you will conduct experiments to tell you the same answer? "

Because the experimental data is much more expensive than a stack of computer runs. If you even get to run transient tests, you are lucky; more likely you will calibrate your current job's computer simulations based upon either lab-generated experiments (for which you have a lot of data but the case modelled is rather simple, and there aren't very many of these cases to model, so..), or field failures (for which you will have limited data, many assumptions, etc.).

FWIW, I have no idea what the "Bond Graph" method is, but would agree with BigInch that it probably works just fine, and the choice of software is more about how easily it can be used for your anticipated work situation. The Method of Characteristics is well known, and described in a number of gas- and fluid-dynamics texts. To quote from a book on my shelf, Fluid Transients in Systems, Wylie, B. and Streeter, V., Prentice-Hall, 1993:

"For rapid transients in liquid pipeline systems the mehod of characteristics is generally considered to be the numerical method by which others may be judged for accuracy and efficiency. Other methods are available and in certain situations there are valid reasons to consider their use."

Personally, my fluid dynamics software consists of several dozen spreadsheets, some ancient fortran code that I occasionally dust off, some compiled codes that I can dust off if I ever need to, and some 20+ years of experience that lets me (usually) predict on the back of an envelope what the computer eventually tells me. Then we go try it in the real world, and Ma Nature grades my work.

 

[stanier]

AFT's Impulse provide a guide to the use of the Method of Characteristics in the matter of vapour formation.

The MOC method of simulation, without cavitation, is a reliable method and accurate solution. However after cavitation occurs the reliability of predictions deceases significantly.

The discrete cavity model is offer by AFT's Impulse (WYlie et all pgs 66-67) 1993. This does have its drawbacks as significant non physical pressure spikes may occur as the bubbles collapse. Significant engineering judgement is then required to interpret results.

Geoff