Improving Natural Gas Network Performance by Quantifying The Effects of Braess' Paradox

Open Access
Phillips, Temitope
Graduate Program:
Energy and Mineral Engineering
Doctor of Philosophy
Document Type:
Date of Defense:
April 04, 2012
Committee Members:
  • Dr Luis Ayala, Dr Seth Blumsack, Dissertation Advisor
  • Luis Ayala, Committee Chair
  • Seth Adam Blumsack, Committee Chair
  • Turgay Ertekin, Committee Member
  • James Terry Engelder, Special Member
  • Improving Gas Network Performance
  • Braess' Paradox
In 1968, Dietrich Braess introduced the “Braess Paradox” which is a scenario where the construction of a new link in a traffic network causes or worsens congestion in the transportation network system, thus increasing the total travel time for users. This concept has now been extended to electrical network systems, some liquid pipeline networks, computer networks and general mathematical research. Researchers have been able to show that under certain conditions, the Braess Paradox occurs in all of these systems. The primary objective of this research is to observe the effects of Braess’ Paradox in natural gas pipeline networks, and possible methods of optimizing natural gas flow with this knowledge. Currently, no research has been carried out concerning the effects of the Braess Paradox in natural gas pipeline networks. Upon reviewing some of the research that has already been carried out in electrical network systems and some pipe network systems, we have now demonstrated that there are conditions under which the Braess Paradox occurs in natural gas systems. Being able to determine the conditions under which this might occur would definitely aid optimization of natural gas pipeline network systems in designing the structures and in selecting the pipe routes. We ran several simulations using small gas networks, ranging from single-pipe cases to Wheatstone structures and five-pipe structures, to much larger networks with as many as 100 pipes. Based on our research, we have concluded that there are conditions where we will observe the paradox in networks with four or more pipes, depending on the topology of the network, the conductivity of the pipes, and the ratio of flow in the inlet (supply) and outlet (demand) nodes. For the much larger network with 100 pipes or more, we also observe the occurrence of the iv paradox by reducing the networks to Ward equivalents of the smaller network structures, and observing the sensitivity of flow parts of the large network to slight changes in gas demand. Performing these tests has enabled us observe that indeed, our hypothesis about the occurrence of the paradox in Natural Gas networks, was true. The work also proposes an economic model that studies the costs involved with observing the paradox within an already-existing network, and how this affects the decision-maker’s net present value (NPV) for that particular network. Based on the NPV, the decision-maker can decide whether to shut down the pipeline causing the paradox, or to completely remove the pipeline from the network. Otherwise, doing nothing about the paradox will result in loss of money over the life of the project.