Optimal Sectioning of Hydrocarbon Transport Pipeline By Volume Minimization and Environmental Vulnerability Assessment

Pipelines are usually used to transport liquid hydrocarbons due to the fact that they are considered one of the safest ways to achieve this task. Despite this fact the US DOT have reported 149 incidents until May 2014, totaling 6,296 net lost bbl. Some of these spills represent serious environmental damage such as the one at Michigan on August 2010, leaving 16 miles of Kalamazoo river polluted; and the one that took place at North Dakota on September 2013 affecting 30,000 m² of soil.

In countries characterized by a reach biodiversity, not just locally but globally important ecosystems can be lost or seriously damaged. The spill at Monagas Venezuela on February 2012 affecting more than 200 species, 4 of them endemic is a clear example.

In order to minimize the consequences and thus the risk, pipelines have shutdown systems involving sectionalizing valves for mitigation; making the number and location of equipment  imperative in the design and management of the pipeline; especially for areas where topography characteristics are a main variables due to mountain conditions and  ecological importance.

To do so some researches have proposed optimizations based on the minimization of spilled volume or the minimization of costs associated to environmental impact by optimizing the number or the location of valves. This poses an alternative way of sectioning, which supports the standards and recommended practices on industry; for example 16 km is the maximum distance between valves proposed by the National Association of Pipeline Safety Representatives on the United States.

This project has as output the optimal sectioning; both number and location of valves for liquid hydrocarbon transport pipelines, the optimization is divided on two stages: volume minimization and the incorporation of environmental vulnerability of the area. Both stages are programmed on Java generating a computational tool.

Stage one (volume minimization) include the operational variables of the pipeline and the characteristics of the topography, so the volume can be divided on: static and dynamic volume. The calculation associated to the static spilled volume is achieved by the hydrostatic forces on the pipeline associated to topography characteristics coupled with a source model. The selection of the number and location of valves is based on the minimal spilled volume after the location of the valves by a linear optimization algorithm using flow on networks. The algorithm consist on the division of the pipeline by i nodes, each one associated with a maximum spilled volume (max V_i), the algorithm founds the network composed by the nodes with the minimum sum of maximum volumes.

Stage two (environmental vulnerability) takes into account not just the spilled volume but the vulnerability of the area. To assess the last one, a number of factors divided on hydric, biotic and abiotic are selected and ranked on five levels, each one associated with the level of damage given a specific characteristic of the area describing an specific factor, where the highest level represent the highest damage and thus a highly vulnerable area. To include this vulnerability characterization on the optimization algorithm a weight a_i is assigned to each of the five levels and then included with the spilled volume on the objective function. The optimization algorithm includes for this case the weight on the characterization of the node by its multiplication with the spilled volume.

For both stages the maximum distance between valves is included as a restriction and it is defined by the Class location on CSA Z662 and an additional weight is summed at the end per valve installed by the algorithm.

The results obtained support the decision making process for the location of sectionalizing valves, already stablished on standards recommendations and representing the minimum volume and environmental damage configuration. Fundamental optimization variables are identified such as the pre stablished distance between nodes on the optimization algorithm, as smaller the distance less uncertainties, bigger resolution times and well distributed the valves.

 All the distances between nodes represent a minimization on spilled volume, but it also can be observed that configurations with higher number of valves not necessarily implies less spilled volume, indicating the importance of the location of the equipment. An advantage of the methodology and the tool is the use of factors, since its number can vary depending on the available information to characterize all the segments of the pipeline.

Regarding the weight per valve installed, as it increases, the number of valves and its location stays constant, meaning the algorithm at a certain point takes into account the weights associated to the network (composed by the volume and the environmental vulnerability) over the importance of the weight associated to each installed valve.

Finally through the results it can be concluded that the location of the valves is an important variable of decision and that the optimal configuration not only depends on the number of valves. The more detailed the information on the pipeline (less distance between nodes) better resolution, and higher resolution times. Including the cost of the valve on the network implies a reduction near to the 50% of the number of the valves for the higher costs, and reaches a constant value at this point.