GaS_GeoT: A computer program for an effective use of newly improved gas geothermometers in predicting reliable geothermal reservoir temperatures

Geothermal Energy

Science – Society – Technology

Table 6 New improved gas geothermometers proposed in this research work

Neural network model	New gas geothermometer nomenclature (GasG_i)	Input variables					Number of inputs neurons	Number of hidden neurons	Number of outputs neurons	ANN architecture
Neural network model	New gas geothermometer nomenclature (GasG_i)	1	2	3	4	5	Number of inputs neurons	Number of hidden neurons	Number of outputs neurons	ANN architecture
ANN-7	GasG₁	ln(H₂S/CO₂) 46.32%	ln(CH₄/CO₂) 24.84%	ln(H₂/CO₂) 28.84%			3	18	1	[3–18-1]
ANN-9	GasG₂	ln(CO₂/CH₄) 42%	ln(H₂S/CH₄) 31.02%	ln(H₂/CH₄) 26.97%			3	15	1	[3–15-1]
ANN-20	GasG₃	ln(H₂S/CO₂) 39.12%	ln(CH₄/CO₂) 37.68%	ln(H₂/CO₂) 23.2%			3	18	1	[3–18-1]
ANN-22	GasG₄	ln(CO₂/CH₄) 33.05%	ln(H₂S/CH₄) 34.51%	ln(H₂/CH₄) 32.44%			3	34	1	[3–34-1]
ANN-23	GasG₅	ln(CO₂/H₂) 24.6%	ln(H₂S/H₂) 50.2%	ln(CH₄/H₂) 25.21%			3	15	1	[3–15-1]
ANN12	GasG₆	ln(CO₂) 26.18%	ln(H₂S) 30.19%	ln(CH₄) 15.37%	ln(H₂) 28.26%		4	13	1	[4–13-1]
ANN-25	GasG₇	ln(CO₂) 27.83%	ln(H₂S) 12.81%	ln(CH₄) 25.87%	ln(H₂) 33.48%		4	10	1	[4–10-1]
ANN-13	GasG₈	ln(H₂S/CO₂) 25.13%	ln(CH₄/CO₂) 5.11%	ln(H₂/CO₂) 10.29%	ln(H₂S) 42.86%	ln(H₂S/H₂) 16.61%	5	9	1	[5-9-1]

According to the Garson's method (Garson 1991), the relative contribution for each input variable was determined in subsequent calculations of the geothermal reservoir temperatures. This value appears in percentage units below each input variable