I’ve not written this guide as an all encompassing “turbo size guide”, more to give you a basic understanding of how to read compressor maps as it’s something I’m asked occasionally to spec. This is quite long winded, but I've tried to cover several permutations so you can see how changing one thing affects another.
One of the most useful things to do when trying to choose a turbo is to look at a compressor map. These at first seem like a bit of voodoo but in reality, they’re very straightforward to read once you’ve got the numbers to plug into them.
Here’s a standard compressor map, as supplied by Garrett, for a GT2860RS turbo. I’m using this turbo throughout this guide as it’s a reasonably common turbo for these engines as it’s matched quite well to the engine size and power requirements of most people. Let’s continue the guide to show why it’s so well matched

Terms used throughout this guide
Compressor – Sometimes referred to as the “cold" side of the turbo, this sucks in the intake air and compresses it, on through the intake pipework, to the throttle body.
Turbine - This is the "hot" side of the turbo. Hot exhaust gas passes through it, still expanding after the combustion phase of the engine. This expansion spins a turbine wheel that drives the compressor wheel via a solid shaft through the turbo core. The turbine and compressor wheels are physically connected via this shaft, and the speed of one wheel correlates directly to the speed of the other.
Absolute Pressure - This is air pressure referenced from a pure vacuum. All calculations regarding turbo compressors refer to absolute pressure. Note: 1 atmosphere = 14.7 psia (Absolute pressure in pounds per square inch) = 0 psig (gage pressure in pounds per square inch). Generally, a boost gauge reads in psig, referenced to local atmospheric pressure, so 14psi is actually 28.7psia.
Surge - This is smallest amount of airflow a compressor can supply at a given pressure ratio. Any pressure above this at this airflow, the compressor will "gulp" air. This is bad for the turbo and for ultimate horsepower. This is generally not a problem nowadays on well matched “off the shelf” turbos.
Here is a compressor map with the surge line highlighted in red:

On the X-axis (horizontal) of the compressor maps, you'll see the mass airflow of the compressor expressed as lbs/min. On the Y-axis is the Pressure Ratio.
Pressure ratio is defined as follows:
(Atmospheric Pressure + Boost Pressure)/Atmospheric Pressure
You may notice a pressure ratio of 1.0 is the exact same as atmospheric pressure (14.7 atmosphere pressure + 0 boost pressure / atmosphere pressure = 1). A pressure ratio of 2.0 is equivalent to 1 atmosphere or 14.7 psig in the inlet manifold. In this guide I am assuming the pressure is ambient at the compressor inlet (we’re not factoring in height above sea level etc). Also note pressure drops through the intake system (boost pipes, intercooler etc) will cause power loss compared to the data the mass flowrate of the turbo would suggest from our calculations.
The shaped “rings” on the compressor map are called efficiency islands. These are the areas where the compressor has roughly the same efficiency compressing the air. An inefficient compressor generates heat, so the higher the efficiency the better, as the hotter the air the less power we get.
Next, we need to know the mass airflow of the engine. The important things to understand that play a factor in how much mass airflow an engine consumes are:
• Engine Displacement
• Volumetric Efficiency (how well the engine “breathes”)
• Pressure at the inlet valves (boost)
• Engine RPM
If we alter any of these values (more displacement, different cams, more boost, more revs, ported head, etc etc) you can make the engine consume more air and therefore make more power.
Here’s a link to a calculator tool I’ve created to allow you to change the above settings and view the results. It's an excel file so download this and play with the settings.
https://docs.google.com/open?id=0ByX...jVkY3lVd0JNN0U
It does simplify things since it doesn't vary volumetric efficiency by RPM and so forth, but it is a pretty close guide. I’ve used a VE of 90% in my calculations, as that’s close to what a modern 4 valve per cylinder engine gets in high revs.
For my examples as said, I’ll be working from a GT2860RS compressor map, running on a 1598cc saxo VTS engine, with a VE of 90%, and with a maximum RPM of 7500. For the first data collection point, let’s see what results we get at what might be wastegate pressure - 6 psi (pressure ratio of 1.41).
I evaluate compressor maps by looking at the airflow at 2000 RPM. I find that on the X-axis and then draw a straight line from that point at a PR of 1 to the airflow at 3000 RPM at your desired PR (1.6 in this case). This gives you an idea of how the turbo will look when spooling up, and tells us if it's at risk of surge. From here, the line should stay at that PR (presuming a constant boost level) all the way to the airflow at redline ~18.8 lb/min in the example.

The data points used are 5lb/min at 2000rpm, 7.55 @ 3000rpm, then 18.88 @ 7500rpm.
As you can see, surge is not a consideration at all, and at this pressure, the turbo sits completely inside it’s map. BUT – there’s a lot of headroom here for more power – we have a whole area above and to the right of the graph to work in, so we’re under-utilising the turbo right now.
Let's spice things up, and see what happens when we up the boost to 18 psi(PR of 2.24):

This is MUCH better - you might have heard people tell you a GT2860RS "doesn't like more than 18psi" - here's a great demonstration why. The data points are 8lb/min @ 2000rpm, 12lb/min @ 3000rpm and 30lb/min at 7500rpm. We stay completely inside the graph (although we're visually close to the surge line, in reality deficiencies in the intake system should keep us away from surge), and we're operating across those efficiency islands much more now. At just under maximum RPM, we're perfectly in the turbo's maximum efficiency range of 76%.
Pushing for more boost, and at 20psi:

We're now working the other end of the compressor map - this time outside of the efficiency island. We'll be generating a lot of heat now too, something that will affect performance more than these graphs account for.
So - moving up to a GT3071R turbo. This is a whole different animal. On these engines, capable and proven to make big power, but at the cost of very poor spool. Here's why:
at 6psi:

Well that sucks. We never spool the thing up! Operating way under efficiency here, we're barely using the turbo at all.
Let's give it some boost and see what happens - 18psi:

Ok - so, now we're on the surge side of the map. Bad. We can't spool this turbo fast enough to get it working where we want it until high revs. It takes about 3800rpm (15lb/min) to get the thing operating back inside it's efficiency island, and we don't get peak efficiency until maximum RPM (7500). So what we have (and seat-of-the-pants tells us this) is an extremely laggy turbo, that makes peak power in a very narrow power-band - right at the top of the rev-range.
Ok - I've done some graphs for you - if you want to experiment more, check the calculator tool out, then check out some compressor maps and plot your lines and see what you get. A great place to start for compressor maps is the Garrett website : http://www.turbobygarrett.com/turbob.../turbochargers
I hope that helps you guys getting into boost to chose what you want and to help keep it realistic. Using the guide above, you can see why a GT25 turbo is perfect for 200hp at 6000rpm, and why a GT35 wont be - even though on paper, it can make 600hp
Additional notes/further reading:
If you want to get more advanced, and include the pressure drop across your pipework/intercooler in your calculations, the Garrett website has a PDF outlining how to do this (I've blatantly stolen this from them):
PR = (boost + intercooler drop + atmosphere pressure)/(atmosphere pressure - air filter drop)
Example:
Intake manifold pressure (boost) = 12 psi
Pressure drop over intercooler = 2 psi
Pressure drop through air filter = 0.5 psi
Atmosphere pressure = 14.7 psi at sea level
PR = (12+2+14.7)/(14.7-0.5) = 2.02 in that example.
One of the most useful things to do when trying to choose a turbo is to look at a compressor map. These at first seem like a bit of voodoo but in reality, they’re very straightforward to read once you’ve got the numbers to plug into them.
Here’s a standard compressor map, as supplied by Garrett, for a GT2860RS turbo. I’m using this turbo throughout this guide as it’s a reasonably common turbo for these engines as it’s matched quite well to the engine size and power requirements of most people. Let’s continue the guide to show why it’s so well matched

Terms used throughout this guide
Compressor – Sometimes referred to as the “cold" side of the turbo, this sucks in the intake air and compresses it, on through the intake pipework, to the throttle body.
Turbine - This is the "hot" side of the turbo. Hot exhaust gas passes through it, still expanding after the combustion phase of the engine. This expansion spins a turbine wheel that drives the compressor wheel via a solid shaft through the turbo core. The turbine and compressor wheels are physically connected via this shaft, and the speed of one wheel correlates directly to the speed of the other.
Absolute Pressure - This is air pressure referenced from a pure vacuum. All calculations regarding turbo compressors refer to absolute pressure. Note: 1 atmosphere = 14.7 psia (Absolute pressure in pounds per square inch) = 0 psig (gage pressure in pounds per square inch). Generally, a boost gauge reads in psig, referenced to local atmospheric pressure, so 14psi is actually 28.7psia.
Surge - This is smallest amount of airflow a compressor can supply at a given pressure ratio. Any pressure above this at this airflow, the compressor will "gulp" air. This is bad for the turbo and for ultimate horsepower. This is generally not a problem nowadays on well matched “off the shelf” turbos.
Here is a compressor map with the surge line highlighted in red:

On the X-axis (horizontal) of the compressor maps, you'll see the mass airflow of the compressor expressed as lbs/min. On the Y-axis is the Pressure Ratio.
Pressure ratio is defined as follows:
(Atmospheric Pressure + Boost Pressure)/Atmospheric Pressure
You may notice a pressure ratio of 1.0 is the exact same as atmospheric pressure (14.7 atmosphere pressure + 0 boost pressure / atmosphere pressure = 1). A pressure ratio of 2.0 is equivalent to 1 atmosphere or 14.7 psig in the inlet manifold. In this guide I am assuming the pressure is ambient at the compressor inlet (we’re not factoring in height above sea level etc). Also note pressure drops through the intake system (boost pipes, intercooler etc) will cause power loss compared to the data the mass flowrate of the turbo would suggest from our calculations.
The shaped “rings” on the compressor map are called efficiency islands. These are the areas where the compressor has roughly the same efficiency compressing the air. An inefficient compressor generates heat, so the higher the efficiency the better, as the hotter the air the less power we get.
Next, we need to know the mass airflow of the engine. The important things to understand that play a factor in how much mass airflow an engine consumes are:
• Engine Displacement
• Volumetric Efficiency (how well the engine “breathes”)
• Pressure at the inlet valves (boost)
• Engine RPM
If we alter any of these values (more displacement, different cams, more boost, more revs, ported head, etc etc) you can make the engine consume more air and therefore make more power.
Here’s a link to a calculator tool I’ve created to allow you to change the above settings and view the results. It's an excel file so download this and play with the settings.
https://docs.google.com/open?id=0ByX...jVkY3lVd0JNN0U
It does simplify things since it doesn't vary volumetric efficiency by RPM and so forth, but it is a pretty close guide. I’ve used a VE of 90% in my calculations, as that’s close to what a modern 4 valve per cylinder engine gets in high revs.
For my examples as said, I’ll be working from a GT2860RS compressor map, running on a 1598cc saxo VTS engine, with a VE of 90%, and with a maximum RPM of 7500. For the first data collection point, let’s see what results we get at what might be wastegate pressure - 6 psi (pressure ratio of 1.41).
I evaluate compressor maps by looking at the airflow at 2000 RPM. I find that on the X-axis and then draw a straight line from that point at a PR of 1 to the airflow at 3000 RPM at your desired PR (1.6 in this case). This gives you an idea of how the turbo will look when spooling up, and tells us if it's at risk of surge. From here, the line should stay at that PR (presuming a constant boost level) all the way to the airflow at redline ~18.8 lb/min in the example.

The data points used are 5lb/min at 2000rpm, 7.55 @ 3000rpm, then 18.88 @ 7500rpm.
As you can see, surge is not a consideration at all, and at this pressure, the turbo sits completely inside it’s map. BUT – there’s a lot of headroom here for more power – we have a whole area above and to the right of the graph to work in, so we’re under-utilising the turbo right now.
Let's spice things up, and see what happens when we up the boost to 18 psi(PR of 2.24):

This is MUCH better - you might have heard people tell you a GT2860RS "doesn't like more than 18psi" - here's a great demonstration why. The data points are 8lb/min @ 2000rpm, 12lb/min @ 3000rpm and 30lb/min at 7500rpm. We stay completely inside the graph (although we're visually close to the surge line, in reality deficiencies in the intake system should keep us away from surge), and we're operating across those efficiency islands much more now. At just under maximum RPM, we're perfectly in the turbo's maximum efficiency range of 76%.
Pushing for more boost, and at 20psi:

We're now working the other end of the compressor map - this time outside of the efficiency island. We'll be generating a lot of heat now too, something that will affect performance more than these graphs account for.
So - moving up to a GT3071R turbo. This is a whole different animal. On these engines, capable and proven to make big power, but at the cost of very poor spool. Here's why:
at 6psi:

Well that sucks. We never spool the thing up! Operating way under efficiency here, we're barely using the turbo at all.
Let's give it some boost and see what happens - 18psi:

Ok - so, now we're on the surge side of the map. Bad. We can't spool this turbo fast enough to get it working where we want it until high revs. It takes about 3800rpm (15lb/min) to get the thing operating back inside it's efficiency island, and we don't get peak efficiency until maximum RPM (7500). So what we have (and seat-of-the-pants tells us this) is an extremely laggy turbo, that makes peak power in a very narrow power-band - right at the top of the rev-range.
Ok - I've done some graphs for you - if you want to experiment more, check the calculator tool out, then check out some compressor maps and plot your lines and see what you get. A great place to start for compressor maps is the Garrett website : http://www.turbobygarrett.com/turbob.../turbochargers
I hope that helps you guys getting into boost to chose what you want and to help keep it realistic. Using the guide above, you can see why a GT25 turbo is perfect for 200hp at 6000rpm, and why a GT35 wont be - even though on paper, it can make 600hp
Additional notes/further reading:
If you want to get more advanced, and include the pressure drop across your pipework/intercooler in your calculations, the Garrett website has a PDF outlining how to do this (I've blatantly stolen this from them):
PR = (boost + intercooler drop + atmosphere pressure)/(atmosphere pressure - air filter drop)
Example:
Intake manifold pressure (boost) = 12 psi
Pressure drop over intercooler = 2 psi
Pressure drop through air filter = 0.5 psi
Atmosphere pressure = 14.7 psi at sea level
PR = (12+2+14.7)/(14.7-0.5) = 2.02 in that example.