How would a select an ARIMA model based on the ACF and PACF? I am trying to estimate a model based on the ACF and PACF. I understand the ACF applies to MA and PACF applies to AR. Do I just count the significant lines and order it that way? I have used a Box Cox transformation and differenced the data. 
I am attempting to forecast monthly oil production in the US. Data chart is posted below.
https://www.eia.gov/dnav/pet/hist/LeafHandler.ashx?n=PET&s=MCRFPUS1&f=M

 A: Sometimes too much of a good thing is not so good !.   Your 1201 monthly values starting at at 1920/1 is such an example. The historical plot is hysterical  suggesting that one might start with the last 109 values starting at 2011/1  ... A recent consistent set of values .
Time series can have both auto-projective and/or deterministic structure. If one limits the solution space to purely arima structure ignoring level shifts and time trend changes and changes in model error variance and possible seasonal pulses and assumes no anomalies ..one can come up VERY SHORT of identifying a possibly useful model. .
There is no need/justification to seasonally difference this data set ... there is just 1 month of the year (Feb) that appears to be consistently exceptional (DOWN).
The 109 most recent values are a cacaphony suggesting the need for data mining yielding an error process that is free of structure.
Here is the Actual/Fit and Forecast for the most recent 109 values  . The model form contains NO ARIMA STRUCTURE  and here 
To summarize there is a seasonal dummy at February of each year AND an identified error variance enlargement  at 2014/7  from http://docplayer.net/12080848-Outliers-level-shifts-and-variance-changes-in-time-series.html  and time trends and level shift indicators. 
The Actual and Cleansed graph is illuminating  . The Forecast graph is here 
The residual ACF is here 
I used AUTOBOX ( a piece of software that I have helped  develop) to automatically dissect your data to signal and noise . The free time series modelling tools fitting a pre-set of arma models and using a simplistic AIC/BIC criteria will sometimes work on data that is uncomplicated ... your selected series , as most are , is complicated.
