For basic setups here is a video where David shows some simple examples:

First, there is Proximity. Having the subs in the proximity of the tops allows their outputs to be aligned and summed in phase. This helps with "punch" and "impact". If you stack subs in the center, you get the benefit of more even coverage of bass but you lose the benefit of that bass being phase-coherent with the tops over a relatively large coverage area. In big systems, when the tops are flown well above the subs, this alignment issue is a problem even when the subs are left and right because they will also be misaligned to a certain extent from front-to-back due to the height offset. Generally with these bigger systems there is enough equipment and therefore enough sources that the gaps get filled in well enough for the audience to get acceptable results in most locations.  

In smaller systems it can be worse because the tops aren't able to deliver a sufficient portion of the low frequency energy to make the sides sound full when the subs are stacked at a center location. The center-stacking advantage of having even coverage of the bass, without the lobes associated with a left-right placement, is offset by the potential imbalance in the perceived level of bass. With the subs in the center and the tops left and right, the perception can be too much bass in the middle and too little bass on the sides.

The 'best' solution is to have at least 3 subwoofer sources. One left, one right and one center. Instead of having one power alley, and several valleys, you will create 3 power alleys. One in the center, between the left and right, one between left and center and one between right and center. Having three sources creates a situation where you're almost never in a location where an absolute null can occur. It puts full-range, high-impact sound on the outer edges so the overall balance is better to the left and right. It puts power alleys where there used to be valleys. It also puts a bit of extra whomp right in front of the stage where the fervent fans tend to gather.

Now let's consider the difference between indoor and outdoor. Outdoors you're less likely to have surfaces that reflect sound surrounding your system. There's the ground, but not much else. And if your subs are on the ground, the reflections are not generally destructive. Indoors, the reflections of low frequency sound contribute to the comb-filtering effect. Comb-filtering refers to destructive interference caused by an offset of arrival times. Reflected sound arrives later than direct sound and it, too, can be constructive or destructive. The difference in distance from the source to the listener versus the source to the reflecting surface and back to the listener will determine the frequencies that are reinforced or cancelled. The side walls, the back wall behind the stage or speakers, the far wall and the ceiling will all reflect sound. For the most part, reflected sound indoors will be perceived as an increase in the level of higher frequencies due to their shorter wavelengths. On the other hand, due to the longer wavelengths of low frequencies, the reflected low-frequency sound can create large areas with a perceived absence of bass. Due to these reflection-induced nulls, subwoofer placement indoors is less forgiving than outdoors.

For instance, if a subwoofer is placed with its back to a wall, there will be a reflection as the sound leaves the front of the box, wraps around the box and then reflects off the wall. If the same subwoofer is placed with its side to the wall, there is effectively no difference in distance between the source and the reflection, making for a smoother response in the room and a slightly higher level to boot.

When you have a null, no amount of level at the source can overcome it. The remaining SPL will always be the same. +1-1=0, +10-10=0, +100-100=0  The only solution is to change the relationship of the direct sound to the reflecting surface. In other words, to move the source,- because that's usually easier than moving the walls and ceiling. A hard-reflecting, solid surface is effectively a secondary sound source presenting the listener with an inverted, delayed, attenuated signal. The delay is the path length difference from direct to reflected sound. The attenuation is the energy loss over the additional distance and the inversion is the result of the reflection. This problem is potentially worse for people with OCD because if you put everything at the same distances from walls and each other, all the positives and all the negatives add up the same. If you offset things slightly you can have one side null at different frequencies from the other, with the result being some less complete nulls and a more even average level. 

On carts is generally not as good as off carts - unless you're doing the lifting at the end of the night. On a concrete or dirt floor, off the carts will result in a slight, very slight, difference in SPL and impact. The reason is that if the subs can't move back when the cones move forward, more of the energy is transmitted into the air. You can strap multiple subs together with ratchet straps to link their total mass to improve performance. This helps whether they are on carts or not.

Other than minimizing heavy lifting, there are circumstances where on a cart is better, such as if you want the carts to isolate the subs from the surface they're on, like a resonant floor or stage. A resonant floor or stage can make the subs sound muddy by contributing its own tonality to the overall sound. For instance, putting the subs on carts with the wheels unlocked and rotated front-rear will allow the boxes to absorb the cone inertia without transmitting as much energy into the floor.

Avoid putting subs on any stage that has mics on it unless that stage is poured concrete or packed dirt. The only reason for putting subs on a stage, with no mics, is in a room with a ceiling reflection that's killing your low-end energy. This would tend to be a very low, very rigid ceiling, usually less than 12 feet, such as in a basement. If you elevate subs in a room like that, you will probably get better summed response between the floor and ceiling, where the ears are.

This brings to mind the advantages of cardioid deployments. Outdoors, cardioid can keep bass levels down on stage and even keep LF energy from interfering with other performances or even adjacent neighborhoods. Indoors, cardioid deployments can improve frequency response in the audience area by minimizing wrap-around reflections that would otherwise cause destructive interference as they bounced off the back wall and then all around the venue.

In conclusion, I would recommend adding one subwoofer for center placement to a 2 subs and 2 tops setup to help even out bass response in most circumstances. Pay close attention to reflection surfaces and try to use them to your advantage if possible. And consider that a certain amount of asymmetry can work in your favor.