The rails' preference was always to fill. If that didn't meet their needs practically and/or economically, they'd bridge. If two girders, or a combination with a central pylon could do it, they'd do that preferentially. If not, they'd pay through the nose for a complicated truss design to cover the span. Costs go up non-linearly as the span increases.
The various constructions of viaduct are always gee-whiz getters, whether you drive under one or see one well-executed on a layout. But that's where it gets tricky: you have to do some detailed planning and careful construction, not least of which would be the surrounds. A viaduct that's only three real-world inches high isn't going to give anyone pause except to remark on how odd it looks. But, one set in a chasm or over a gorge where it towers 12 real inches high is going to elicit oohs and aaahs. My point (yes, I have one) is that what you craft should look convincing and realistic. Trying to fit a viaduct where it doesn't look right is going to disappoint you. Is this extension going to afford you a realistic setting for a viaduct? Maybe a timber trestle would look better, or simply a girder span if your era is after 1950, maybe a deck truss? For example, I seem to recall that the Atlas Warren Truss can be installed 'upside down' as a deck truss...might be wrong about that.