Optimization of a large gas pipeline network results in reduced hydraulic analysis time, reduced fuel consumption, and generally improved pipeline operation. This paper evaluates the practical and technical application of a Genetic Algorithm-based optimization tool, which accurately identifies the optimum compressor, control valve, and block valve operation while constrained by various system limitations including linepack. This implies multi-objective optimization of a highly constrained network with a large number of decision variables. The paper also discusses the opportunities and challenges associated with potential integration into the business of pipeline system operation of the TransCanada Alberta sub-system. Comparison with traditional manual optimization results shows considerable improvement in fuel optimization. Finally, surrogate low-fidelity approximation methods have been employed to reduce the computational demand required by genetic algorithm high fidelity based methods.
A = constant matrix
fit = fitness parameter
g = fuel consumption
H = adiabatic head across a compressor unit m& = mass flow rate
N = compressorspeed
ND or R = number of delivery or receipt points
obj = objective variable
P = pressure
PL or U = lower or upper bound of pressure
Q = actual inlet flow rate to a compressor unit
r = penalty parameter
R = domain
S = source term in mass conservation
W = compressorpower
α = constant
η = compressor adiabatic efficiency
ϕ =pipeline hydraulic function
a =average
d = discharge
D = delivery
r = compressor unit r
R = receipt
s = suction
min = minimum
max = maximum
The TransCanada system may be described as a collection of measurement, pipeline, compressor station, and valve facilities of every size, type, vintage, design, and configuration imaginable. This is not much different than many large pipeline systems worldwide which serves to underline the significant challenge associated with optimization. The Alberta system is somewhat unique in terms of throughput (10-12 Bcfd) which cannot be influenced by TransCanada, linepack (13-15 Bcf) which can only be influenced in the extremes, climate (-50 to 120 F), soil conditions (some frozen areas), and diversity of MAOP (400 to 1440 psig), but the most significant challenge associated with hydraulic modeling and optimization stems from the network nature of the system and the number of multiple flow paths, many of which are bi-directional. In order to effectively model the Alberta system it can be divided into no more than four sub-systems, the largest of which is referred to as North and the smallest of which is referred to as West Path as shown in Fig.1.
