As computers have increased in speed and capacity, pipeline modeling software has kept pace. Modelers can simulate larger networks, and work on local personal computers (PCs) instead of Time share systems. Now, modelers can construct larger, more complex, more detailed, and precise models. Many modelers assume that building with more precision yields a more accurate model. Is this really true? Are these models any more accurate than the less-detailed models of previous decades? Does the added precision (detail) improve hydraulic calculations? Could additional precision make the modeling process more difficult? What is the necessary level of detail? Inexperienced modelers may find such questions difficult to answer. If the device is in the field or on the drawing, they reason, it must be in the model. In many cases, though, including detail on a device requires assumptions about hydraulic parameters that affect the modeling accuracy. They need to consider the purpose of the model, the computational significance of the device to the model, and how their hydraulic assumptions affect model accuracy and performance. Also, hydraulic models are being used in many different areas of pipeline companies. No longer just planning tools, these models serve as tools to operations and even marketing staff. How does the use of the model influence the level of precision needed? Can the same model apply to more than one application? These questions should be asked whether you are running steady state or transient models, simulating transmission or distribution, or gathering and analyzing gas or liquid fluids. Remember learning to solve story problems? The most difficult lesson to learn was how to ascertain what information was actually needed for the solution! The same applies here. Pipeline simulation models are extremely large math problems. As with any math problem, it is essential to consider the solution one seeks. To converge on the solution, use only the information required to arrive at the solution. None of the questions posed here have absolute answers. This paper provides insight and guidelines to aid in exploring the answers and to make model building an easier process. The real examples provided draw on the combined modeling experience of the authors.
Simulation has made some major advances in the pipeline industry. Prior to the 1960's, most pipeline simulation was done with hand calculations (using slide rules and log tables) and nomi graphs or lookup tables. Most of this work used empirical correlations such as Panhandle A, Panhandle B, or Weymouth, since there was no friction factor to calculate and no iterations to find the answers. In the 1960s, people began to recognize the capabilities of computers (which were in their infancy), and developed computer programs to run pipeline simulations. Because of the restrictions of the hardware available, the original programs were fairly simple and could handle a very limited problem size (normally with a maximum of 100 devices). Engineers would submit jobs and wait hours or, sometimes, days to receive results.
