The premise of Central Place Theory is that goods/services have a natural range. The natural range for low-value bulk goods is fairly small. They need to be converted to a higher-value good to get a larger range. Iron ore, for example, is a pretty low-value commodity that requires significant additional resources (primarily fuel like coal or charcoal) to do much with. Iron is a much higher-value good that you're likely to ship much farther and worked iron is a very high value good. Consider Damascus steel: it required special treatment of special iron that was sourced from specific mines in India. When the traded iron because largely inaccessible due to political changes, the supply of the steel stopped and the ability to produce it was lost. An example of how a trade technology can make a huge difference is Roman hydraulic concrete: it needs pozzulan (a type of volcanic ash usually sourced from Italy bask then) for its properties and the availability of bulk shipping networks allowed it to be used around the Mediterranean for the creation of harbors. Those harbors allowed new trade nodes and shifting of certain supply networks. Some goods are high enough value (silphium and Lebanon cedar being two examples of the ancient world) to have large ranges, but were hunted to extinction and their trade networks collapsed.

Some goods are medium value but gain a lot when transported to processing centers. Wool is a great example. You need sheep and you need someone to collect the wool from the sheep. Then you need to clean the wool, dye the wool, and spin it into yarn. From there it can be knitted into goods or further woven into cloth that can be processing into more goods. Cleaning the wool takes fuel to heat the water and soap to remove the grease, dyes can be very expensive and of very limited production range (e.g. Murex purple or saffron yellow), carding and spinning are traditionally very labor intensive, knitting can be a cottage industry, and looms require a great deal of time and effort to create and use. There are any number of towns based primarily around the production of textiles (wool, linen, cotton, silk, and so on) in the historical record and available technology has had a huge impact on those industries. Cotton production used to be a fairly low-volume operation due to picking the fluff from the seeds, but the invention of the cotton gin gave new life to that industry (and the slave trade in the US) and thread-making and thread-weaving systems powered by something other than humans led to substantial concentrations of industries.

Changes in the landscape (e.g. a natural disaster that permanently closes an important mountain pass, a river that changes course, construction of a Roman road, or construction of the Erie canal) will have huge impacts on the trade networks because the transport cost for certain goods along some graph edges becomes much lower, effectively increasing the range of goods along those graph edges. Those changes will further distort the assumed hexagonal grid of CPT.

I think that you may be looking at the problem backwards. A "smelting town" won't have higher demand for some goods because it's a "smelting town": a town will arise there because the cost of conversion of ore to metal is cheapest there. That's why boom towns arose and collapsed all over the western US: it's usually not worth it to transport the ore very far and it's usually not worth it to transport supplies very far. When the value of the produced good falls below some profit level, the town declines. That value can be affected by simple economic factors like cost of transport, but also by political factors such as cost of law enforcement in the area. For two fairly modern examples, look at the history of Cerro Gordo in California and the story of the Los Angeles aqueduct in the same area. It's stories of boom and bust and transport networks. There is a small distortion available in the case of industries that require significant expensive infrastructure because they will pull in lower-grade materials from a larger area for a little while, but they will eventually decline as well.