The solenoids we are looking at are rated at 50/60 Hz so that's a major constraint.
We'd use a 12V solenoid but the manufacturer suggested that a 120V AC solenoid has a quicker actuation rate, and actuation rate is a driving factor for our design. I could likely swap it with a 24V DC one easily - a 12V one might be more difficult to find. At the end of the day we definitely would prefer lower actuation time over an over-complicated power system.
You'd be better off using a 12V solenoid and over-driving it with a brief, higher voltage pulse, to increase the actuation speed. It's much easier to convert 12VDC to a higher DC voltage, than to 110VAC.
A brief overvoltage will not damage the solenoid. If the manufacturer advises against it, it will be because they're covering their back. I've worked on a project for work which required some solenoid valves to be actuated as quickly as possible. The driver circuit powered 24V rated solenoids with a brief 120V pulse to actuate the solenoid very quickly, dropping the voltage down to 24V, once the valve was open.
The DC:DC converter which provides the higher voltage from 12V, doesn't have to be continuously rated for the solenoid's full current. It can charge a capacitor, which is discharged into the solenoid. Indeed the simplest solution is a capacitive voltage doubler, but it will need a large capacitor, which might be too bulky.
Attached is the simplest way to fire a 12V solenoid, with a brief 24V pulse, before dropping the voltage back to 12V. It uses a relay (L1) with two normally open and one normally closed contact. L1 is the solenoid. You didn't specify the series resistance and inductance, so I guessed. D2 and R1 is a snubber, which limits the back-EMF. A higher value resistor will make the solenoid switch off faster, but with a higher back-EMF voltage, no resistor will make it much slower, but with only 0.7V of back-EMF.
When the relay is off, C1 is charged from the power supply, via R2 which limits the current to protect the relay contact; not very efficient but it works. When the relay turns on, C1 is connected in series with the power supply, reverse biasing D1 and boosting the output voltage to double the input. C1 discharges through the load and the voltage falls back towards the supply voltage. When C1 has discharged, D1 conducts and powers the load directly from the power supply. The delay in the output being turned on and off is due to the relay. The model I used is for a small, high sensitivity relay, which is fast. A larger relay will be considerably slower.
Note that the models for the relay and solenoid are basic. In real life the inductance of the coil will change, depending on the position of the armature, which isn't modelled.