I'm looking for a small TSP instance (no more than a few thousand cities) that draws out the worst case factorial performance in Concorde (https://neos-server.org/neos/solvers/co:concorde/TSP.html). So far, it seems to break at nothing - the results returned in seconds or minutes. The theoretical exponential worst case performance should be noticable, even on Concorde, and even if N is in the range of just hundreds or a couple thousands.

All the small datasets available online (eg. http://www.math.uwaterloo.ca/tsp/data/index.html, http://www.tsp.gatech.edu/data/index.html) seem to be too easy, already long solved to optimality. And while there are work that proposed methods to generates difficult instances (eg. https://link.springer.com/chapter/10.1007/978-3-642-20525-5_1), I wish to know if there is already a dataset of hard, unsolved (yet small) instances available, so that I can focus on my actual research and not have to implement and experiment with published instance generation methods.

A similar question has been asked before, back in 2013: Where to find a set of hard Traveling Salesman Problems (with known solutions/approximations)?. But I find the answers unsatisfactory; besides what I need is a relatively small TSP instance - one where the difficulty is due to the structure rather than the size.