|Theme||Visible||Selectable||Appearance||Zoom Range (now: 0)|
Booncharoen, Pichita (Chevron Thailand Exploration and Production) | Rinsiri, Thananya (Chevron Thailand Exploration and Production) | Paiboon, Pakawat (Chevron Thailand Exploration and Production) | Karnbanjob, Supaporn (Chevron Thailand Exploration and Production) | Ackagosol, Sonchawan (Chevron Thailand Exploration and Production) | Chaiwan, Prateep (Chevron Thailand Exploration and Production) | Sapsomboon, Ouraiwan (Chevron Thailand Exploration and Production)
Abstract In the past few years, over hundreds of wells were drilled in Gulf of Thailand, had faced with the depletion and lost circulation issues resulted from a lack of pressure data. A prior research of reservoir depletion pressure (Fangming, 2009) in oil field, China was obtained from multivariate statistic and regression by using density and neutron porosity log curves in logging-while-drilling data. However, the relative errors are 7.5% from the actual formation pressure. Thus, there are several latent variables in the model like drilling parameters (Rehm, 1971) which part of formation pressure. From 2018 initiative model in Satun-Funan, the classification model was obtained by using mud gas, porosity, water saturation, net sand thickness, net-hydrocarbon-pore thickness and neutron-density separation. However, the limitation is drilling parameters could not account by classifier, and accurate only original pressure category. So, this study has expanded scope to include other reservoir properties and drilling parameters then applied with machine learning on offset well dataset by using three regressors such quantile, ridge and XGBoost regressors. The pore pressure estimation model aims to improve efficiency for making decision in execution phase, increasing confidence in perforation strategy. The model parameters, pay thickness, porosity, water saturation, original pressure from local pressure profile and total gas show are accounted into this model. As of regressor assumption, some facts are conducted to logarithm and perform 2nd polynomial feature for model flexibility. There are three steps for building model such as data manipulation, analysis and deployment. Two purposes of pressure prediction impact algorithm selection, for operational phase, quantile regressor is implemented to provide conservative prediction while Ridge or XGBoost regressors are alternatives for perforation strategy, provide mid case result of pressure prediction. Overall model performance was measured using root mean square error (RMSE) on train & test dataset which show approximately 1.2 and 1.5 ppg range of accuracy respectively from total 12 drilling projects in Pattani basin. Overall model fitting is within reasonable range of generalization capacity to apply with unknown data point (test set). The future model will continue to improve accuracy and manage imbalanced dataset between original pressure and depleted sands.
Abstract Lost circulation is the most common drilling issue for infill drilling projects in Satun-Funan Fields, South Pattani Basin, Gulf of Thailand (GOT). The depleted sand is possible to be a root cause in many wells based on observation from resistivity time-lapse separation in depleted sands or shale nearby. Therefore, the objective of this study is to estimate fracture pressure related to the depleted sand and design an appropriate Equivalent Circulating Density (ECD) threshold for each well to avoid or minimize lost circulation and well control complication during drilling a new well. This study model is using Eaton (1969) equation. There are 3 input parameters which are Poisson's Ratio and pre-drilled estimated depletion pressure and depth. With limitations of no actual fracturing data and limited sonic log, the maximum ECD while lost circulation reading from Pressure While Drilling (PWD) tool and formation pressure test data were used to back-calculate for Poisson's Ratio and identified a relationship with depth. From the total of 68 wells in the Satun and Funan areas, the interpreted Poisson's Ratio ranges from 0.36 to 0.44 and its linear trend is apparently increasing with depth. To minimize the variation of back calculated Poisson's Ratio the local data become an important key for model validation and maintain the similarity of subsurface factors. This interpreted Poisson's ratio trend will be used to calculate for fracture pressure by incorporating with estimated depletion pressure and depth that expect to encounter in each planned well. The lowest fracture pressure in a planned well is used to prepare pre-drilled ECD management plan and a real-time well monitoring plan. Additionally, the model can be adjusted during the operational phase based on the new drilled well result. This alternative model was applied in 4 trial drilling projects in 2019 and fully implement in 6 drilling projects in 2020. The lost circulation can be prevented with value creation from expected gain reserves section is $57M and cost avoidance from non-productive time due to lost circulation is $3.4M. With an effort, good communication and great collaboration among cross-functional teams, the model success rate increases by 12%. However, there are some unexpected lost events occurred even though the maximum ECD lower than expected fracture pressure. This suspect as a combination of limitations and uncertainties on key input parameters and drilling parameters. In the future, the model is planned to expand to other gas fields in the Pattani Basin which will move to more infill phase and have higher chance of getting lost circulation to maximize benefits as the success case in Satun and Funan fields.
Charusrojthanadech, Nunthawath (Department of Civil Engineering, Faculty of Engineering, King Mongkut'sInstitute of Technology Ladkrabang, Bangkok, Thailand) | Yamamoto, Yoshimichi (Tokai University, Graduate School, Science-and-Engineering Hiratsuka-shi, Kanagawa-ken, Japan) | Kawai, Kyohei (Tokai University, Graduate School, Civil Engineering, Japan)
ABSTRACT This paper summarizes a method for estimating the degree and geographic extent of tsunami disaster damage from Indian Ocean Tsunami by using geo-spatial data (such as satellite remote sensing images, aerial photographs, topographic map, ground photos and field survey results) in two case studies of damage to west coast of southern Thailand. The authors learned the following from these case studies:the extent of flooding can be estimated from discoloration of vegetation; damage to buildings can be estimated by deciphering whether the roof was lost or not; and damage to coastal facilities such as seawall can be also estimated by deciphering from geo-spatial information. Moreover, the authors developed a methodology to estimate the risk of a seawall being washed away by using a laboratory experiment and also developed a methodology to show how tsunami damage can be evaluated. INTRODUCTION Tsunami is that cause massive damage occur every few decades somewhere in the world. On December 26, 2004, the 9.0 magnitude Sumatra Andaman mega thrust earthquake spawned gigantic seismic waves or tsunami in the Indian Ocean which caused large scale coastal flooding in various countries such as Indonesia, Thailand, Sri Langka, India, Bangladesh, Malaysia, etc. In Thailand, the area affected by the Tsunami was located in the west coast of six of its southern province. The tsunami caused the death of people including Phang-nga, Phuket, Krabi, Ranong, Trang and Satun. The tsunami caused people were killed, coast plain flooding, resulting in damage to many buildings, coastal facilities, beaches, inlets, etc. Preventing the occurrence of tsunamis is impossible with current technology level, but it is possible to minimize tsunami damage if detailed estimates of potential damage are made.