Elbows - you are quite right that this issue is subject to imprecision in the way we talk about it, and you are right to demand clarity. I think it is best understood with this graph:
All the conventional oil we've discovered is in green. Since it has declined at 5% per annum since it peaked in 1968, it is reasonable to project it forward. When you do that you get 1.9 trillion barrels. Since it is unreasonable to exclude the possibility of future conventional discovery you need to add something. I've added 300 billion barrels. The biggest undeveloped conventional play left is Iraq, which the USGS estimates Iraq is about 78. So 2.2 trillion assumes we can discover another 5 Iraqs of conventional oil (which I think is highly unlikely, but that keeps the analysis conservative). I've actually included a little bit of unconventional - that last little bump on the downslope is mostly deepwater offshore, now declining.
There is an infinite number of future production trajectories, but they are all constrained by the fact that you can't produce more than you have discovered. So the area under the red curve can't be more than 2.2 trillion barrels. The one I've shows assumes a normal distribution centred on 2009. This is a Hubbert curve. You can delay peak by pulling production forward, but that steepens decline. Assuming 1.65% growth, the latest you can maintain production growth is 2030, at which point production falls instantly to zero. This is physically impossible (reservoirs do not exhibit piston like displacement).
Oil demand has grown in recent history at 2% per annum. That's a doubling time of (70/2) 35 years. It is determined by population growth and per-capita energy consumption. BP and IEA differ, but 1.65% annual growth rate still doesn't get non-OECD to even half our standard of living in 30 years, so is conservative.
The gap between demand and the Hubbert conventional curve is the oil that must be discovered and mobilised (the dashed area). The actual production curve you will see going forward is a blend of the Hubbert curve accelerated by whatever they can achieve with incremental capital investment (thus steepening decline in a decade and worsening the economic impact of energy withdrawal) and non-conventionals (arctic, ultradeepwater, shale, biofuels, tar sand).
The slightly tricky conceptual jump is to understand that the Hubbert curve projects historical production. The historical production encodes the technology and commercial conditions from which it arose. So a Hubbert projection is a forecast of what you get assuming technology and economic conditions remain the same. The difference between the Hubbert curve and the demand curve is the amount by which technology and economic conditions have to improve
relative to historical conditions. It is basically a measure of optimism.
It is, by definition, unconventional (our green graph accounts for all historical and reasonably foreseeable conventional conventional discovery). Since you know it takes 10 years between discovery and production, you can work out the non-conventional discovery rate you need to deliver this production and sustain gorwth. I've assumed 5 hypothetical 75 billion discoveries every five years, starting 5 years ago and shown the first 5 cycles. Mobilising 25 Iraqs buys you 14 years. They reckon at a push they may have mobilised most of the only Iraq we have in 14 years.
We would have to recreate the discovery rate that was achieved in the late 1940's and sustain it indefinitely. That is net, of course - since unconventional oil has a fraction of conventional oil EROEI, gross volumes are substantially higher (I've not bothered to estimate since it is unrealistic). It is, of course, rate - they added billions of barrels of tar sand last year to world reserves (actually discovered in the 1920's but declared now to maintain the statistical illusion of discovery growth) but they need to establish a flow from it, which is impossible at these volumes.
The USGS, EIA and IEA (although the IEA are now coming clean) are hilarious. They take that gap and try to break it down into the wedges in your graph that correspond to some known or foreseeable physical production mechanism, from conventional right through to liquidising peasant's lunches to make aviation gasoline. They always end up with a category called some variation of "unconventional, unidentified". It is the euphemism in technical circles for "doesn't exist".
For extra points, you can convert that gap into kwH, and from there into equivalent solar, wind, biomass, etc capacity and work out the annual capacity delivery schedule you'd need to maintain to sustain energy growth. And you can multiply that by £/kwH, tonnes steel/kwH, land/kwH etc. to work out the capital investment and primary resource requirement (to a first approximation - that would neglect the additional capacity you would have to build to compensate for the hydrocarbon diversion to the manufacturing process, but we've discussed that). It's a lot.
Not accounted for here is domestic consumption growth in the major suppliers. For example, Saudi Arabian population growth is 1.5% per annum and energy demand growth is 5% per annum. They use 20% of their own production for domestic energy, including massive desalination. 40% of their water comes from aquifers which have dropped 150m and are believed to be near exhaustion, doubling domestic desalination oil consumption. They currently export 11 million barrels a day. At current export capacity (15 mmbd), domestic consumption growth will eliminate all Saudi export by 2030. They are building 16 nuclear plants within conventional missile attack range of Israel and Iran to safeguard their water supply, indirect evidence of the lack of confidence they have in their (state secret) conventional oil reserves.
