Well, there are a lot of factors involved, nearly every single factor to be exact. You have limitations that are one of the major factors. The others would be driver excursion, power handling, and surface area. The average gain in dB you will recieve nominally from an added enclosure is around 3-5dB. Why is this? To simplify it as much as I can, the compression has a lot to do with it as well as coupling the subs to the box using compression ratios, though some say ratios do not exist....they do. Simply by matching the drivers ability to reproduce the output using an enclosure, the physical compression ratio between one enclosure and another the same size added is closer to 1:1 than you think. And since compression/expansion of the drivers cone creates pressure, if the ratio was less than 1:1 for box to drivers acoustical compression volume, then output normally increases, but this is limited by physical factors mentioned above, not to mention the ability to create backpressure if the design allows it to happen.
Now, going back to the actual question, why is it so hard....is because of electrical/acoustical relation to sound pressure and power. Due to Ohms law, the factors of resistance, current, and voltage are limited to the circuit they are connected to. And it is all about matching the impedance of the subs to the amplifier that gives you the easiest and cheapest way to increase output. But on average, again due to Ohms Law and factors of power in relation to electrical current, you have to nearly double the amount of power to create that same 3dB increase as adding another enclosure. not to mention if the drivers sensitivity is low, the output will suffer. So, matching sub configuration impedance, finding drivers with high sensitivity, high power ratings, and great amount of linear excursion, you will get good results, but limited still by the power factor.
Take this for example: You have a driver that is rated at 90dB sensitivity. The enclosure will add the efficiency and acoustical means of gaining output to that 90dB. And that is based on the designs ability to couple the impedance, control the driver, and placement, and well many many things, but lets take a single frequency, like a burst for example (trying my hardest to keep it simple...it gets very very complicated normally), and feed it power. Now using Ohms law, You figure for lets say its 4 ohms resistance, and you have measured a voltage of 28V, and lets say that is the most your amp will put out as is. So, taking 28V, and resistance, you can figure for power. P=V^2/R. So, 28*28, then divide by 4. You get 196 watts. And to keep it REALLY simple, I will not show the calculations to figure for dB output, lol. But lets say it was measured at 110dB. Now, the dB increase from 90dB to 110dB is 20dB. That is a great increase. Now, to get 113dB (nominally), you would need around 392Watts of power (twice of the 28V power). But voltage does not increase at the same rate as power. But continuing with this, Lets say you want to try to get to 140dB with this particular design. Now, figuring for how many times it takes using 3dB increases, we can figure for power requirements. So, from 113dB to 140dB is alittle less than 10X the 3dB increases. So, we were at 392 watts, now for another 3dB, you need twice the power (I hope you see already where this is going). So, take 392 watts, double it 9 X. You will get more than 200K Watts! Now, the question is not can you get that much power, but can you utilize it with the drivers power level ratings and excursion limitations( that would be a no by the way)? That is why it is so hard to acomplish because the higher the dB levels increase, the more power it will take in relation to the 2X increase in power. So, to get 160dB from this, you will need way over 25M watts of power. Wow right?
So, what we do in order to balance this, is design around efficiency and coupling to get as much dB out of it at 2.83V as possbile before any more power is added. Then you can figure for everything else until one factor is fully limited.