zdas04
Mechanical
- Jun 25, 2002
- 10,274
I did a project a few years ago for my former employeer where I was trying to predict reservoir pressure for a gas well from a set of hourly pressure build-up data. I found that for a specific gas reservoir if I had 48 data points then I could use the MathCad GENFIT routine to fit the constants into the equation:
P(bar) = A*ln(hours+B)+C
The "A" "B" and "C" values were different for each well and I was able to "prove" to an acceptable level of confidence that this particular reservoir would have average reservoir pressure at the wellbore within 5 years (call it 40,000 hours) and that the data from the first 48 hours of a buildup would define the equation that allowed prediction of the current reservoir pressure.
Now my former employer is a customer and wants me to write a program for their production techs to use to get the constants for each well (this is a pressure-vessel over-pressure protection exercise not a reserves study). They don't want me to use ODBC calls to Mathcad so I'm in a position of writing a VBA version of GENFIT. Does anyone have any good links to the theoretical basis for GENFIT? I've done some searches and haven't yet been able to find a curve fitting theorm that uses partial derivatives like GENFIT does.
David
P(bar) = A*ln(hours+B)+C
The "A" "B" and "C" values were different for each well and I was able to "prove" to an acceptable level of confidence that this particular reservoir would have average reservoir pressure at the wellbore within 5 years (call it 40,000 hours) and that the data from the first 48 hours of a buildup would define the equation that allowed prediction of the current reservoir pressure.
Now my former employer is a customer and wants me to write a program for their production techs to use to get the constants for each well (this is a pressure-vessel over-pressure protection exercise not a reserves study). They don't want me to use ODBC calls to Mathcad so I'm in a position of writing a VBA version of GENFIT. Does anyone have any good links to the theoretical basis for GENFIT? I've done some searches and haven't yet been able to find a curve fitting theorm that uses partial derivatives like GENFIT does.
David
