ABSTRACT
Infrastructure planning for networked infrastructure such as grid electrification (or piped supply of
water) has historically been a process of outward network expansion, either by utilities in response to
immediate economic opportunity, or in response to a government mandate or subsidy intended to
catalyze economic growth. While significant progress has been made in access to grid electricity in Asia,
where population densities are greater and rural areas tend to have nucleated settlements, access to
grid electricity in Sub-Saharan Africa remains low; a problem generally ascribed to differences in
settlement patterns. The discussion, however, has remained qualitative, and hence it has been difficult
for planners to understand the differing costs of carrying out grid expansion in one region as opposed to
another. This paper describes a methodology to estimate the cost of local-level distribution systems for
a least-cost network, and to compute additional information of interest to policymakers, such as the
marginal cost of connecting additional households to a grid as a function of the penetration rate. We
present several large datasets of household locations developed from satellite imagery, and examine
them with our methodology, providing insight into the relationship between settlement pattern and the
cost of rural electrification.
Download Article


A recent article published in Science and featured in the Economist describes a computational model for building a network based on the behavior of the slime mold. Researchers placed oat flakes representing cities around Tokyo and found that the slime model accurately recreated the Tokyo rail network. The authors then specify a mathematical model to simulate the slime mold’s behavior.
Amoeba-Inspired Network Design
Science 22 January 2010
http://www.sciencemag.org/cgi/content/summary/327/5964/419
Rules for Biologically Inspired Adaptive Network Design
Science 22 January 2010
http://www.sciencemag.org/cgi/content/abstract/327/5964/439
Railways and slime moulds: Network-engineering problems can be solved by surprisingly simple creatures
Economist 21 January 2010
http://www.economist.com/sciencetechnology/displaystory.cfm?story_id=15328524
In addition to our chemotaxic neuron growth model, I would like to add an implementation of Tero et al’s slime mold model as a network building algorithm in our application.