The model has many unique pipe lengths

A model with many unique pipe lengths often require a large number of sections in order to correctly use the Method of Characteristics. A large number of pipe sections in a model results in long run times and typically offers little improvement in accuracy.

How are unique pipe lengths contributing to long run times?

To run a transient simulation, AFT Impulse sections pipes to apply the Method of Characteristics solution to the waterhammer equations. In any system there will be a single controlling pipe. The controlling pipe is the pipe with the fewest sections - generally one. AFT Impulse uses the length of a section in the controlling pipe to divide all pipes in the model into sections and determine a uniform time step for the transient simulation.

A model with many unique pipe lengths often requires the controlling pipe to be broken into multiple sections in order to divide all pipes into uniform sections and uniform time steps. The run time is proportional to the square of the number of sections, meaning many sections in a model will often cause extremely long run times.

What can the user do?

Consider adjusting the lengths of the pipes in the model such that short pipes share a common pipe length and that longer pipes can be perfectly divided by the controlling pipe. Using exact pipe lengths can enhance calculation accuracy in many applications. However, transient simulations have many inherent uncertainties, and precise modeling of pipe lengths may not offer a noticeable improvement. Rounding short pipe lengths such that the lengths are perfectly divisible by the controlling pipe length, introduces minimal inaccuracy while significantly reducing the number of sections in the model and resulting run time. Be sure to perform sensitivity studies and confirm the steady-state results with these modifications.

For instance, consider a model with five pipes of lengths 1.01, 2.31, 3.75, 6.00, and 1000 meters. All pipes have unique lengths and none are perfectly divisible into another. To section these pipes within the acceptable variance limits, the controlling pipes needs to be split into two sections, and over 2100 total sections are required to perform the calculations. Rounding the four shortest pipes to 5 meters increases the total pipe length of the model by only 0.7%. In return, the number of pipe sections is reduced by a factor of 10 and the run time is reduced by a factor of 100! This 0.7% change in total pipe length can be considered negligible for any reasonable transient simulation - there are greater uncertainties than this in almost all studies.