The following data from Laherrere brings up some interesting issues to contemplate with regards to depletion modeling.
Oil companies discovered oil in Cook Inlet (Swanson River) in 1957 and on the North Slope (Prudhoe Bay) in 1968. I found a forecast for North Slope of 22.3 billion barrels and a total extraction of 1.06 billion for Cook Inlet.
For North Slope I used the discovery date and the forecast as a stimulus to the oil shock model, and added the Cook Inlet model separately. For North Slope, I used values for fallow, construction, maturity, and extraction of 0.15 and for Cook Inlet, I used values of 0.2. The shock model production curve looked like:
I found it interesting that a strong discovery stimulus with typical rates adequately describes the curve:
In spite of the conflict between the stochastic premise of the model and the determinism implicit in a single field, the shape largely matches -- except for one significant area. Production only commenced on Prudhoe in 1977, as soon as workers completed the Alaska Pipeline. So we see a sudden surge in production in the actual curve around 1977 which does not show up in the shock model. Since companies worked on construction of the rigs and pipeline in parallel, something has to give. I suppose that extracted oil prior to the completion of the pipeline might have got wasted or stored in reservoirs. Naturally this does not show up in production numbers but it has to pop out somewhere. Otherwise, one must suppress extraction until the pipeline opened up, which would have produced a large shock right around 1977 -- something entirely doable within the context of the oil shock model.
So if I leave the extraction rate at some small number like 0.01 until 1976 and then jump up to 0.15 in 1977, the model looks like this:
In general, a lot of this detail gets washed out as we take larger sets of reservoirs with varying discovery dates, yet the single set provides us with mucho insight -- without invalidating the fundamental premise of the shock model.