You're already ahead of the game by having three sub-woofers. Now to really take advantage, you must optimize the three and minimize the room's negative impact on what you hear. The science involved is very well understood, and quite well documented.
Three subs is a fantastic tool,...you've got to use it. It's actually reasonably easy. Here
is a good, well explained paper, that's not very long, and not too difficult to grasp. Toole really explains the manner in which you can implement selective mode cancellation, by proper placement of subs to really dampen the modal effects at the LP. This is very effective and not too difficult.
Dennis gave great advice above in pursuing the midpoints of three walls, and the two in the front wall (@ 1/4 to 1/3 widths), and the third somewhere along a wall behind you,... wherever measures/works best is perhaps where I'd start. The axis in which you spread the sources is the direction of modal control you're addressing. For example, spreading the two along the front, from the 1/4 points, to the 1/3 points in from the sidewalls addresses the room's width modes. Subsequently, the rear sub addresses the front to back modes.
It's very easy in real-time to measure modal effects with REW, or OmniMic. One can clearly see the peaks and valleys in response by moving front to back, or side to side,....thus you'd know which axis is most problematic. It's not uncommon to have big wide peaks of 15dB, and nulls of 20-30dB. If you leave all your subs along one wall, you'll need to rely solely on EQ to knock down the peaks to minimize the effects of the nulls. Spreading the subs around oftentimes allows you to use dramatically less EQ, and with proper delay/phase adjustments per sub, you'll retain ideal localized and integrated bass with the mains, in an entirely coherent subwoofer system.
Now for three subs, ideally you'll need individual delay for each sub. There's many products out there, the Behringer DCX2496 is inexpensive, and very popular for duty such as this. Also, you'll have plenty of signal shaping options, both individually, and globally across the three subs too.
If you only have an Audyssey based rig, a good one like a two channel capable (MultEQ XT32), then you'll only have two locations to use. You'd need to co-locate two subs, then place the third one where is does the most good modally. Then allow the PrePro and Audyssey perform the freq and time domain EQ'ing. IMO, this is easy, however dramatically much better results may be had by implementing the DCX2496 and rolling your own.
I know that's a lot of info, however optimizing an HT with a multi-sub approach may be one of the biggest values, producing tangible results in this entire hobby. Also, be mindful that the time domain is where you'll find the gold. A strong measure of bass trapping, without negatively affecting the room's liveliness, or system envelopment, will take a good multi-sub system to the next level.
Addressing the time domain behavior of the LF, damping any ringing will add a level of delineation and detail
to the bass, that no amount of equipment could ever add. Bass traps and lots of them,...if needed they can have a ~6mil plastic (or equivalent) covering to allow them to be reflective above a few hundred hz., yet entirely absorptive from that point and below. Anyway, I only bring up bass trapping because a multi subwoofer system such as yours has the potential to approach sub-woofing nirvana,...just sayin'
Best of luck
Two other links of precise interest;
The Geddes method
for a multi sub approach.
paper is truly superb.
Those are focused on the task at hand, if you're interested further on the time domain, bass trapping,.. etc., PM me.