[sohail1]

New to CFD as a field. Working to develop accurate hydraulic models for long Oil & Gas pipelines. We have following goals for these models: real-time computation, highest possible accuracy, maximize reusability from one pipeline to the next, modular approach and efficient model development. We will acquire real pipeline training data for parameter estimation to tune to models. Once good fit models are in place we will use them for a) data validation and anomaly detection by comparing sensor measurements against model simulation b) online optimization using model predictions. The online use is the reason that models must calculate in real-time. We will likely run the model at sub-second intervals (since things can change any time eg. valves opening and closing, pumps coming on etc.). The space discretization steps along the pipeline may be of the order of 5 meters to 250 meters depending on the total length of the pipeline. It is clear that we will need to respect the NS / mass momentum energy conservation equations. What is not clear is what discretization / calculation methods we should consider eg. FVM / FDM / MOC / Upwind / SIMPLE ... ? Also are there any public domain libraries may help? I understand that this is not a typical CFD use case.

[lcarasik]

Originally Posted by sohail1

New to CFD as a field. Working to develop accurate hydraulic models for long Oil & Gas pipelines. We have following goals for these models: real-time computation, highest possible accuracy, maximize reusability from one pipeline to the next, modular approach and efficient model development. We will acquire real pipeline training data for parameter estimation to tune to models. Once good fit models are in place we will use them for a) data validation and anomaly detection by comparing sensor measurements against model simulation b) online optimization using model predictions. The online use is the reason that models must calculate in real-time. We will likely run the model at sub-second intervals (since things can change any time eg. valves opening and closing, pumps coming on etc.). The space discretization steps along the pipeline may be of the order of 5 meters to 250 meters depending on the total length of the pipeline. It is clear that we will need to respect the NS / mass momentum energy conservation equations. What is not clear is what discretization / calculation methods we should consider eg. FVM / FDM / MOC / Upwind / SIMPLE ... ? Also are there any public domain libraries may help? I understand that this is not a typical CFD use case.

CFD (i.e 3-D N-S solvers) is not likely to help you accomplish your goal. Problems of this size should be investigated using "system" level codes that can simulate large systems like that very quickly. There is a lot of information out there that can be found with some good google searches.

[sohail1]

I understand the comment that 3 D NS will be an overkill. From what I understand the model must be able to support transients to be useful. Typically the term being used in the pipeline market is RTTM - Real Time Transient Model. May be we are talking about a 1D NS solution? There seems to be a lot of literature on the net around NS based pipeline transient models employing Methods of Characteristics for numerical solutions.