Flow test data of Hannah1 and others were used with Program MPTF to determine how well the results of a computed simulation would agree with the measurements of an actual flow test. Computed pressures and flow rates were obtained with applications of each of two methods for determining friction factors. An "experience" or efficiency factor E was used with each method to optimize the matching of computed and observed or logged data. The experimental pipeline was 15.95 miles in length, and four equal pipe segments were used for flow simulation. Measured data were recorded tor 5-min intervals, but substantial reduction in the number of iterations required per time step was achieved by using time steps of less than 5 min for treating relatively rapid parameter changes. For one hour of flow during which rates of flow doubled, rates were matched to 7.3 MMcfd and pressures to 1.5 psi. These results are in substantial agreement with results of verification tests made with other programs.
The pipeline was 19.56 in diameter and for purposes of computation was divided into four equal segments of 21054 ft each. There was an elevation rise of 100 ft. Molal average molecular weight of the natural gas was 20.3. A flow temperature of 60° F was assumed. Fig 1 is a diagram of the line. Flow is from left to right. Pressure P" 1 and output rate Q" 5 were regarded as parameters having the time-dependent values observed in the flow test. Input rate Q1 and output pressure P5 were computed with the purpose of matching logged values of the test. To provide pressures P2, P3, and P4 for the outset of computing, a steady state of no flow was assumed. The factors f are computed before the first time step to serve without change for all time steps. E2 is constant but if increased will increase Q in direct proportion. It ordinarily will have a value of the magnitude 1. By change of E2 the effect of factors f can be modified to permit a computed match of logged data for an actual pipeline now. Method 3 gives effect to pipe diameter, specific gravity of the gas, and rate of now. The friction factor and its origin is that reported by Flanigan and Wilson3. The Fannin friction factor is included in the general now equation, for all four of the verification tests in the time range 12:10 to 12:40 p.m. Halbert and Lotito4 investigated these differences and concluded that they could have been caused by a dead-end stub which was known to exist part way down the line but which was not included in the model. Logged velues of Q" 5 peaked at 12,25 p.m. with the value 129.7 MMcfd, compared to 119.1 MMcfd 10 min earlier and 124.4 MMcfd 10 min later. On reduction with time or logged pressures Pl and P5 the dead-end stub unloaded gas in addition to that entering the pipeline.
