Abstract
A pipeline network that collects and transports animal wastes to a treatment plant (e.g., a digester) is an interesting alternative to vehicles (no traffic increase, -61.31% CO2 emissions, -42.46% energy consumption in the case study presented here). However, pipeline networks require careful design to minimise installation costs, costs that depend on pipe length more than diameter. The optimisation can therefore be modelled as a "Euclidian Steiner minimum tree" problem that can be solved by using Kruskal's and Simpson's algorithms respectively to delineate a preliminary minimum-spanning-tree path and to optimise the paths by introducing new bifurcation points. The presented case study involved 32 dairy farms belonging to a milk consortium in a 1000-km(2) area. The proposed procedure resulted in a network that extended 98.75 km. However, this was reduced to 97.47 km by introducing additional branching points (total length -1.30%, locally up to -8.54%). Another possibility for cost minimisation is to select farms based on their position. If only farms within 20 km from the centre were considered, the network length decreases to 69.25 km and was optimised to 67.97 km (-1.84%). Although the decreases achieved by optimisation were small, they fall within the estimated range (mu = 2.35%, sigma- = 1.24%) and correspond to 73 632 epsilon and 76 696 epsilon in investment savings. Economically speaking, selecting farms is more efficient with -30.15% of the investment and annual costs and -21.83% of the cost-per-mass-unit. The universality and versatility of the algorithms used makes the proposed approach suitable for designing piping networks in extended geographical areas.