Most of the costs are associated with buildings, support structures and construction work; the nuclear reactor cost is not that significant. The savings associated with mass production of nuclear components is overstated.  Reactors in the 1970's were built on a production line basis, but over 90% of the net project cost was related to construction activities - this is based on my first hand knowledge of these costs.

The $/KW cost of a number of small reactors will be greater than that of single large unit; classic economies of scale as seen throughout the history of power plants in general. No reason to believe the small reactors will deviate from historical patterns.

My estimate is based on the likely $/KW build cost and includes Owners invested equity, profit and debt repayment. 

The maneuvering capabilities of water reactors are limited by thermal considerations primarily involving the fuel. Generally, daily load changes can be accommodated, but rapid changes not so much. Taking one module off-line and the bringing it back on-line does not happen very quickly when considering operational  and administrative procedural limitations. 

Refueling activities are generally not that long in duration - maybe a week with several  days to bring the plant back up to hot standby. Considering the actual seasonal nature of the grid, taking a plant down for refueling is not that big a deal from the grid's perspective.  

Running for as long as you can at full load is driven by financial considerations,; maneuvering capabilities are not that useful because of the relatively high build cost on a $/KW basis.  By contrast, gas turbine power plants have a relatively high fuel cost and low build cost. Reducing gas turbine output has nowhere near the financial impact that occurs for a reactor with a low fuel cost but high ($/kW) build cost.

A better reactor fit for Puerto Rico is probably a small passively safe boiling water reactor (BWRX-300, ~300 mWe) because it can be relatively quickly licensed (it's a simplified and smaller version of a large BWR) and has a decent output. The X-energy gas reactor might be a good fit, but I believe licensing the design will be problematic; the Nuclear Regulatory Commission is hell bent on making the licensing of advanced reactors extremely complex (as in fiendishly expensive and time consuming). That will probably sink the X-energy design even though technically the reactor is pretty good. The NUSCALE design would work but I think the troublesome economies-of-scale associated with a number of modules may prove to be a financial issue.

All of the reactor types will have difficulty overcoming competitiveness issues so it probably comes down to: is a premium price for emissions free nuclear worth it? My guess: nuclear is probably not worth the aggravation and financial risk from a business perspective.