I've participated in the "Traveling Santa Problem"-contest over at Kaggle, and the goal of that contest was closely connected to the Traveling Salesman Problem (TSP). The idea behind the TSP is to help a salesman to find the shortest route through a number of cities, and the salesman can only visit each city once. You can use the solutions to the TSP in real-life as well, often connected to different logistics-areas, but also when manufacturing circuit boards.
This is a plot showing the 150,000 different cities (or chimneys in this case):
Euclidean distance. This is the plot when using that method, using only 5000 chimneys:
Hilbert curve combined with the least distance method. An Hilbert curve will "snake" its way through all the chimneys, and at each Hilbert-point you can calculate the shortest path through that box. The plot will then look like this using all of the 150,000 chimneys:
If you are interested in the Python-code used, you can find it here: github