Well, the VISA network currently runs at about 2000tps. I'd imagine the tps for the whole world would be somewhere around roughly 10x that of VISA. If we were to achieve that all with on-chain scaling we'd have to have +3GB blocks! 

now for those that just want a number.. to the OP's question
population= 7bill
tx per month=42(based on real fiat use stats per month as a bases for comparison)

7bill x 42tx=294000000000tx a month
294000000000tx/28/24/6= 72916666.66666667tx/block(10min) = 121,528tx/s (14,468tx/s based on average bitcoin use stats)

now lets clarify

we are not going to have bitcoin as a one world currency - so no doomsday speaches please. bitcoin being a one world currency is worse than 200 fiats combined.

we are not going to have 7billion permenent users tomorrow - so no doomsday speaches please. bitcoin will gain more users but it wont be 7billion tomorrow. bitcoin can grow naturally at a pace that can cope and it certainly wont be 7billion in the next couple decades.

please chillout and refrain from doomsday speaches of short term exaggerations

---

now that is said.
real world stats show the average fiat citizen uses fiat 42 times a month.
at the moment bitcoin users only use bitcoin 5 times a month. which while looking at the amount of merchants accepting bitcoin (300k) is small and manageable.

anyone shouting that we need to lock peoples funds into LN hubs requiring hub managers to sign off on transactions out of fear of 7billion people piling into bitcoin.. seriously needs to wake up.
we dont need to turn bitcoin into bank2.0 via LN.. we do however need to realise that now that core have deemed 4mb 'safe' we should get core to sort out dynamic blocks to allow for real growth of the expected 7tx/s to 28tx/s which will still be under cores recent backtrack(4mb good)
in the form of 4mb base
or
in the form of 2mb base 4mb weight. (14tx/s) thus they also get their segwit feature they love so much.

as for that 28tx/s this would allow
the current 2mill userbase to do 20tx a month instead of 5tx
or
the current 2mill userbase to grow to 8mill userbase, still doing 5tx a month.

then in a few years(before 2020) look at the 'state of the network' to use consensus to grow the block buffer sizes again to allow more potential growth.
and slowly and naturally grow at a pace that can cope.


oh and if you are going to shout doomsdays of datacentres and excess costs.. please refrain from spewing out those delusions.
bitcoin can happy run on a raspberry pi with a data store the size of a fingernail.
technology of normal cheap desktop computers is superier to that so not an issue.
even telecommunication companies 5 year plan has shown to not be an issue for rational onchain growth.

---

now then.. there are 4032 bitcoin blocks a month.
average tx=400byte
294000000000tx = 117600000000000bytes a month
117600000000000byte/4032 blocks= 29,166,666,667bytes = 29,166mb blocks = 29.166gb blocks(3.5gb blocks based on average bitcoin use stats)

most will think 29gb(3.5gb) blocks is crazy. but be realistic. no one in reality will see 7billion people jump into bitcoin tomorrow.
but here is the thing...

only 20 years ago
if you told people that were on dialup when a song(3mb) took 10 minutes to download, that in a game running in HD while livestreaming it to twitch would require 1megabyte a second.. they would shout dont bother making the game online. it would fail.

if you told people that had a max hard drive of 4gb that a game would require 80gb, they would shout dont bother making the game. it will fail.

yep thats right 20 years ago 4gb hard drives were the max available at consumer level.. thats currently 1000x less than todays consumer level
yep thats right 20 years ago 56k internet was the max available at consumer level.. thats 100x less than todays consumer level

so imagine it like this.. 29gb(3.5gb) in 20 years will have the same 'feel' as what we envision as 29mb(3.5mb) today. we will think its childsplay and not an issue.

but as i said even in 20 years we wont have 7billion bitcoiners. and those that are using bitcoin wont be using it 42x a month like they do with fiat.
so that 29gb(3.5gb) block is just a doomsday number of worsecase scenario but has no place in reality and should not be part of any debate today or anytime soon.

No. That's just your uneducated opinion. High birth rate in certain areas does not need to stop, that's just the natural cycle. Demographic transition has 5 steps, i.e. most countries are already in the 4th/5th step while the relatively undeveloped ones are still in the earlier stages. This is what it is in a nutshell:

---

Where there currently is a high birth rate, there usually is a high death rate that follows. Eventually these balance themselves out, not that this matters for Bitcoin.

I'd like to see the source for this number as I have not seen it somewhere else. I'll assume that your calculation serves the purpose of an average number of TPS needed.

Indeed. This thread, and those alike are actually very often just pointless at this time.

The question I would have is if Bitcoin could make those changes to keep up with the demand if it was ever to get that large?  That is a lot of transactions and I want to see it get that high, but how can the system increase the number of transactions it processes?  does this have to do with the mining?

Most of the people on planet earth are uneducated and unbanked, so these numbers are highly inflated. Even if someone else created this for

them, they would not do this task at the same time. This would happen over many years, like banks have been doing it with bank accounts

and credit card applications. In 2012, one in three workers worldwide – some 1.1 billion people – were unemployed. 

VISA is not enough, in order to cater for the necessity of a globally connected economy, we would probably be looking at trillions (yes with a T) of tarnsactions per second. This is only and only possible throught a secondary layer solution like the lightning network. We will Never get anywhere by scaling onchain, no matter how much we increase the blocksize, it will be ridiculously inneficient in trying to fill in the amount of transactions that a global mainstream situation would require, this is why we need segwit anyway, because lihgnint network works better with segwit, so by blocking segwit you block bitcoin's future.