A few questions to fully understand your objectives first.
1. What exactly do you mean by manage your system?
2. Are you wanting to integrate real time control, or simply assess the hydraulic performance of the existing system and use a hydraulic model (ie EPANET) to determine an appropriate operational regime and forward works program to cater for changing demand?
3. Are you wanting to see the impact of changing valve status or valve settings (eg flow control or prv) on the system?
4. Do you really have recorded system data (eg customer demands, some idea of flows, pressures and reservoir/tank levels, pump on/off)against which to validate your model? Sometimes modelling results can be dubious, even misleading if you do not have sufficient data to validate (i won't say calibrate in this instance)
5. How good is your GIS data (ie do you have alot of missing data/fields)
6. Can I ask where you are and if EPANET, ARCGIS and CAD are the only tools at your fingertips (ie have you already purchase the GIS and CAD software?)
To give you my preliminary thoughts - if you are looking for a cost effective system to assess the performance of your network and undertake some operational planning, you probably can't go past the tools you have mentioned in terms of value for money. Depends on your budget and your objectives.
The only disadvantge with EPANET in this instance is that you will not be able to import GIS data directly nor run directly on top of a GIS (unless you get a programmer to help you). But what you can do is make sure that you set up your GIS database to include all fields required in EPANET, then export your GIS data in text format, manipulate the text a little (ie using code or a spreadsheet even) and you can then import that modified text file as a *.inp file to EPANET. (If you decide to go this way I can give you a few tips about asset naming, data types and other issues to be wary of).
If you are handy with fortran you can write some code to automate this process. Of course, this process will mean that you are not modelling directly on top of your GIS, and you might need to update your model based on any modifications to your GIS on a regular basis.
In addition to importing the *.inp file to generate your EPANET network, you can also create a raster image of your GIS (Eg *.bmp or *.wmf) and load it as a background image to your hydraulic model - just to be sure that you've got the pipe network correct. In the raster image you could also include other features such as roads, buildings, property boundaries, reservoirs, PS, zone boundaries etc.
So if what your planning to do is use the GIS, CAD and/or EPANET in conjunction with each other to help you plan and see the impact of operational changes, EPANET can definitely do the trick. You can even load various demand and pipe network scenarios (eg projected demands at 5 year forward intervals and pipe networks which include your capital improvements).
EPANET also does water quality modelling and retention time quite well.
Unless you're really trying to model your system in detail, and it is extremely complex, and you want to have fancy presentation capabilities, I think you're heading in the right direction. Just be careful about the validation issue - you need to be confident that your model is approximating the real system performance as closely as possible - which means having some gauging.
If you are really wanting to run a hydraulic model on top of a GIS then you might want to look at alternative modelling software. For the size of your system it would seem an expensive exercise as licences can cost. But if you're not willing to spend the money there are ways of getting your GIS data into an EPANET model (as outlined above).
KRSServices also makes an invaluable point - if you cannot spare the resources to properly maintain a system you might want to think twice about how much effort you put into developing these tools in the first place.
Hope this helps.
Regards
Sharyn
