Modeling, analysis, and screening of cyclic pressure pulsing with nitrogen in hydraulically fractured wells


Creative Commons License

Artun E., Khoei A. A., Kose K.

PETROLEUM SCIENCE, cilt.13, sa.3, ss.532-549, 2016 (SCI-Expanded) identifier identifier

  • Yayın Türü: Makale / Tam Makale
  • Cilt numarası: 13 Sayı: 3
  • Basım Tarihi: 2016
  • Doi Numarası: 10.1007/s12182-016-0112-7
  • Dergi Adı: PETROLEUM SCIENCE
  • Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus
  • Sayfa Sayıları: ss.532-549
  • Anahtar Kelimeler: Cyclic pressure pulsing, Nitrogen injection, Hydraulically-fractured wells, Experimental design, Artificial neural networks, NEURO-SIMULATION, GAS, OPTIMIZATION, INJECTION, WATER, FIELD
  • Orta Doğu Teknik Üniversitesi Adresli: Evet

Özet

Cyclic pressure pulsing with nitrogen is studied for hydraulically fractured wells in depleted reservoirs. A compositional simulation model is constructed to represent the hydraulic fractures through local-grid refinement. The process is analyzed from both operational and reservoir/hydraulic-fracture perspectives. Key sensitivity parameters for the operational component are chosen as the injection rate, lengths of injection and soaking periods and the economic rate limit to shut-in the well. For the reservoir/hydraulic fracturing components, reservoir permeability, hydraulic fracture permeability, effective thickness and half-length are used. These parameters are varied at five levels. A full-factorial experimental design is utilized to run 1250 cases. The study shows that within the ranges studied, the gas-injection process is applied successfully for a 20-year project period with net present values based on the incremental recoveries greater than zero. It is observed that the cycle rate limit, injection and soaking periods must be optimized to maximize the efficiency. The simulation results are used to develop a neural network based proxy model that can be used as a screening tool for the process. The proxy model is validated with blind-cases with a correlation coefficient of 0.96.