Ing-forms which are apparently related to both noun and verb have been studied in great depth. According to Aarts, there have been three major attempts to explain different behaviours:
(a) lumping (most such ing-forms are placed in a general class, 'gerund-participles' as Huddleston and Pullum label them in CGEL, though they have tests to split 'noun' from 'verb')
(b) splitting (all such ing-forms are put in either the noun or verb word class, depending on prototypical examples they are considered to most resemble)
(c) a gradience model, where a given example is placed somewhere along a noun-verb continuum, and is not placed in a traditional word class. Quirk et al in ACGEL suggest this model.
Without classifying as definite noun or verb (which depends on the test chosen),
- Modelling of the dynamo proved to be resource-consuming.
is more 'nouny', corresponding to a POSS-ing structure (compare
- [The] / [Filco's] manufacture of the dynamo proved to be cost effective.)
while
- Modelling the dynamo proved to be resource-consuming.
is more verby (compare
- The company was modelling the dynamo.
with obvious past participle plus direct object. The comparison ('test' if you like) is of course far from perfect.)
Either works in your first example (note that the verby variant speaks more of the active process being carried out), but in the second, as you suspect, as in some other cases, the more nouny variant is preferable after 'developments in'.