REINTERPRETATION OF EXISTING WELLBORE LOG DATA USING NEURAL-BASED PATTERN RECOGNITION PROCESSES

A SIGNIFICANT PORTION OF KNOWN GAS RESERVES IS CONTAINED WITHIN HETEROGENEOUS RESERVOIRS FOR WHICH WELL LOGGING, USING CURRENT ANALYTICAL TECHNIQUES, OFTEN CAN PROVIDE ONLY QUALITATIVE INFORMATION. THIS FACT IS DUE TO THE INDETERMINATE NATURE OF GEOLOGIC SIGNAL PROCESSING, COMBINED WITH THE INHERENT LIMITATIONS OF USING MECHANISTIC APPROACHES TO ANALYZE THE RELATIONSHIPS OF MULTIPLE SIGNALS IN COMPLEX GEOLOGIC FORMATIONS. HYDROCARBON SIGNATURE LOGS (HSLS) IDENTIFY PRODUCING ZONES WITH A GREATER DEGREE OF ACCURACY THAN CAN BE DERIVED USING CONVENTIONAL WELLBORE ANALYSIS, SUBSTANTIALLY INCREASING THE HYDROCARBON DISCOVERY RATE FOR OBSCURE RESERVOIRS. HSLS ARE CREATED USING A PROPRIETARY PROCESS CALLED PROWLS (PATTERN RECOGNITION FOR WELLBORE LOG SUITES). PROWLS IS BASED ON AN EMERGENT PATTERN RECOGNITION TECHNOLOGY CALLED NEURAL COMPUTING, DEVELOPED PRIMARILY THROUGH DEPARTMENT OF DEFENSE-SPONSORED RESEARCH, TO ADDRESS THE PROBLEMS OF IDENTIFYING MILITARY TARGETS IN DIFFICULT ENVIRONMENTS. THIS TECHNOLOGY HAS BEEN SUCCESSFULLY ADAPTED TO IDENTIFY OIL-PRODUCING ZONES IN A WELL USING A SUITE OF CONVENTIONAL WELLBORE LOGS. THE PURPOSE OF THIS PROJECT IS TO ADAPT PROWLS FOR SUBSEQUENT MAPPING OF DEVELOPMENT WELLS. THE PHASE I RESEARCH IS DEVELOPING THE DATABASE FOR THE EXPERIMENT, PRODUCING HSLS THAT INDICATE GAS-PRODUCING ZONES, AND INCORPORATING MAPPING INTO PROWLS TO DEMONSTRATE INCREASED GAS RECOVERY.