Jurnal Teknologi Minyak dan Gas Bumi

Jurnal Teknologi Minyak dan Gas Bumi ISSN 0216-6410

JTMGB

Volume 4 Nomor 1 April 2013

Jurnal Teknologi Minyak dan Gas Bumi adalah majalah ilmiah diterbitkan setiap kwartal yang menyajikan hasil penelitian dan kajian para professional yang tergabung dalam Ikatan Ahli Teknik Perminyakan Indonesia (IATMI) dan sebagai media komunikasi anggota IATMI pada khususnya dan mensosialisasikan dunia industri minyak dan gas bumi kepada masyarakat luas pada umumnya.

KEPUTUSAN KETUA UMUM IATMI PUSAT NO: 003/SK/IATMI/II/2013 Penanggung Jawab

:

Ir. Bambang Ismanto

Pemimpin Redaksi

:

Dr. Ir. Ratnayu Sitaresmi

Redaktur Pelaksana

:

Dr. Ir. Usman, M.Eng.

Peer Review

:

Prof. Dr. Ir. Septoratno Siregar (Enhanced Oil Recovery) Prof. Dr. Ir. Pudjo Sukarno (Integrated Production System) Prof. Dr. Ir. Doddy Abdassah (Teknik Reservoar) Dr. Ir. RS Trijana Kartoatmodjo (Teknik Produksi) Dr. Ir. Arsegianto (Ekonomi dan Regulasi Migas) Dr. Ir. Bambang Widarsono (Penilaian Formasi) Dr. Ir. Sudjati Rachmat (Well Stimulation and Hydraulic Fracturing) Dr. Ir. Sudarmoyo (Penilaian Formasi) Dr. Ir. Aris Buntoro (Teknik Pemboran)

Senior Editor

:

Ir. Andry Halim !

" Ir. Junita Musu, M.Sc. Ir. Boni Swadesi " #$

Sekretaris

:

Ir. Bambang Pudjianto (IATMI)

%& $"

Alief S. Syaifulloh, S.Kom. (Sekretariat IATMI)

Sirkulasi

:

Abdul Manan, A.Md. (Sekretariat IATMI)

Jln. Jendral Gatot Subroto Kav. 32-34 Jakarta 12950 – Indonesia. Tel/Fax: +62-21-5203057 website: http://www.iatmi.or.id email: [email protected] Jurnal Teknologi Minyak dan Gas Bumi (ISSN 0216-6410) diterbitkan oleh Ikatan Ahli Teknik Perminyakan Indonesia, Jakarta didukung oleh Fakultas Teknik Pertambangan dan Perminyakan ITB


JTMGB

Volume 4 Nomor 1 April 2013

DAFTAR ISI

%' ' ' '

/'#2

' 0/ & .............................................................................. 1 - 12 Dita Amanda, Taufan Marhaendrajana !" Fracturing Gas Rate Sudjati Rachmat, Ahmad Hadad ..................................................................................... 13 - 23

# $ % & ! ' &

...................................... 25 - 33 Amega Yasutra, Tutuka Ariadji, Zuher Syihab, Pudjo Sukarno

Bambang Widarsono, Kosasih, Junita Trivianty Musu

.................................................... 35 - 47

*+ , -

!

,-. $ Agam Munawar, Donny Hendromurti, Rina Dewi P ......................................................... 49 - 56

KATA PENGANTAR Para pembaca JTMGB yang budiman, pada kesempatan ini atas nama segenap Pengurus IATMI periode 2012-2014 mengucapkan Selamat Tahun Baru 2013, dengan doa dan harapan semoga di tahun 2013 kita semua diberikan kesehatan, kebahagiaan dan rizki yang barokah. Melalui media ini, dengan senang hati kami bisa kembali menjumpai para pembaca dengan aneka materi bacaan ilmiah yang tersaji dalam JTMGB Edisi April 2013 ini. Pada JTMGB edisi April 2013 ini, kita juga ingin membahas persoalan-persoalan sederhana ' ' $" " ! ' ! "% " ' * " %"$ "%"$ bidang reservoar dan produksi yaitu: Studi Teknik Peningkatan Perolehan Minyak dengan Metode "8 ; "$$""<8 = "# *> !" , 1 > ' secara Kontinyu Terintegrasi Sistem & ! " "$ ="$" '"$" $ > 3*+

, -

!

,- . $ 4 Dalam upaya meningkatkan kualitas majalah JTMGB, kami telah mempersiapkan pengajuan

?CL Q "$"' "%"$= "$8"

!";X[ akan dimulai bulan Maret 2013.

Selamat menikmati bacaan edisi kali ini.

(Bambang Ismanto)


Date of issue: 2013-04-28

The descriptors given are free terms. This abstract sheet may be reproduced without permission or charge. Dita Amanda (Institut Teknologi Bandung) Taufan Marhaendrajana (Institut Teknologi Bandung) Studi Teknik Peningkatan Perolehan Minyak dengan Metode Injeksi CO2 Menggunakan Uji Laboratorium dan Simulasi Reservoar 5%1. April 2013, Vol. 4 No. 1, p 1-12

Sudjati Rachmat (Institut Teknologi Bandung) ! " " Methods for Prediction of Post-Hydraulic Fracturing Gas Rate 5%1. April 2013, Vol. 4 No. 1, p 13 - 23

Penelitian ini bertujuan untuk melakukan studi ' ! " ! ' "8 $ 2 untuk meningkatkan perolehan minyak dengan menggunakan uji laboratorium dan simulasi reservoar, sehingga dapat membantu proses pengambilan keputusan dan * ' "$$"" "8 2. Uji laboratorium dilakukan untuk menentukan tekanan tercampu minimum (TTM), melakukan investigasi perilaku fluida pendesak dan pengaruhnya terhadap faktor perolehan, serta mengamati pengaruh keberadaan & di dalam core. Simulasi dilakukan dalam dua tahap %" ] * " * # ] *= 8"" " ' =!" " ]

* " simulasi reservoar dilakukan untuk memetakan hasil uji laboratorium pada skala lapangan. Hasil yang didapatkan dari penelitian ini adalah pengamatan komprehensif mengenai proses pendesakan minyak !$ 2 (swelling, pengurangan viskositas minyak, TTM, dan perbandingan metode uji laboratorium), perolehan minyak dalam skala laboratorium dan skala lapangan, serta perbandingan perolehan minyak skala lapangan dalam berbagai skenario injeksi. TTM yang didapat dari uji laboratorium dan simulasi fluida reservoar adalah 2225 psi. Faktor perolehan minyak dari uji 6+ bernilai 6% lebih rendah daripada hasil uji & akibat pengaruh heterogenitas pada core& " "%Q 2 cenderung berinteraksi dan mengalirkan & terlebih dahulu untuk kemudian mendapatkan kontak dengan minyak residual dalam pori dan mengalirkan minyak ke titik produksi. Pendesakan tersier di laboratorium "$! " " = _`Qq{|}~ # ] * ""8"' """ * "% = "8 2 yang tergantung ' "" " 2 yang ditambahkan. Sementara fenomena swelling dan peningkatan "" * ' $" " =$ = ' "=!" $ 2 pada hidrokarbon. Pengembangan skenario pada simulasi reservoar "$! " " ' " " = `_{~ # " C "$! " ' !" %"$ lebih tinggi dari 7 + 6 " ' ' "%'"2 dapat diperbaiki dengan skenario simultaneous water alternating gas (SWAG) atau diberikan tambahan propana.

% "$ ' * %>! work looping could be started from planning, design, operation, post job treatment, and then evaluation. This complexity provides several approaches for the performance evaluation. In this research, the " $ " ! '' " the approaches. Some parameters initially selected, they are reservoir thickness, water saturation, reservoir porosity, reservoir permeability, pressure gradient, skin factor, productivity index, critical rate factor, cement bonding quality, gel volume, and proppant >! * "' ! "' parameters. The output parameter is the percentage of incremental gas rate. The parameter database: rock !" Q] ! %Q! ' "' are not included in this research. The initial process is to have all data comparable using nodal analysis. Then * " $ " ! > neural network-back propagation (NNBP), adaptive " % " " % #Q "

fuzzy mathematical method. The following processes are performed for the methods, they are Pattern Recognition, Identifying Performance Drivers, and Performance Prediction. There are at least four conclusions can be deducted: (1) Both NNBP and ANFIS are able to recognize the pattern of hydraulic fracturing performance. Furthermore, by reducing the training data and adding testing data, ANFIS is better than NNBP. (2) The performance drivers from both methods are Proppant Mass, Water Saturation, Skin Factor, and Porosity. It is also supported by linear regression method. (3) The fuzzy mathematical concept moderately exhibits the hydraulic fracturing performance. (4) ANFIS method was capable to model the hydraulic fracturing performance by using only four of the most influential parameters. The final network result from ANFIS was utilized into simple application for gas rate prediction after fracturing job. "$* % >8=' in December 2011. It produced excellent match with the actual data.

Kata kunci: #2-6+-&-.#-1

Key words!% "$' " Q neural network, backpropagation, fuzzy mathematical, adaptive neuro fuzzy inference system.

Amega Yasutra (Institut Teknologi Bandung) Tutuka Ariadji (Institut Teknologi Bandung) Zuher Syihab (Institut Teknologi Bandung) Pudjo Sukarno (Institut Teknologi Bandung) Metoda Optimasi secara Kontinyu Terintegrasi Sistem Subsurface-Surface untuk Pengembangan Lapangan Migas 5%1. April 2013, Vol. 4 No. 1, p 26 - 34

L="$ "CLC# !CLC# ?" * "% CLC# Sumatera Coal Bed Methane Reservoirs 5%1. April 2013, Vol. 4 No. 1, p 35 - 47

Metoda dan strategi perencanaan pengembangan lapangan migas mengalami perkembangan yang sangat pesat akhir-akhir ini. Pada awalnya, perencanaan pengembangan lapangan hanya didasarkan pada sisi teknik reservoar saja. Perencanaan pengembangan yang baik membutuhkan masukan dari berbagai sektor dari industri migas, sehingga integrasi aspek-aspek geologiQ $ Q * " ' permukaan mulai terbentuk. Namun, integrasi masih dilakukan secara sekuensial, belum secara utuh di antara aspek-aspek yang terkait tersebut. Fakta memperlihatkan bahwa studi lapangan terintegrasi saat ini umumnya hanya dilakukan sebagai bagian dari proses sekuensial, belum terintegrasi secara kontinyu. Tujuan dari penelitian ini adalah mengembangkan metoda optimasi sehingga dapat diterapkan pada pengembangan lapangan migas melalui pemodelan yang terpadu. Metoda ini memungkinkan variabel yang terkait dengan semua aspek sistem produksi lapangan dapat terwakili yang belum pernah dilakukan pada penelitian sebelumnya. Pemodelan dalam penelitian ini akan melibatkan semua parameter yang terkait dengan reservoar dan sistem sumur produksi, sehingga model tersebut diharapkan dapat merepresentasikan sedekat mungkin keadaan nyata di lapangan. Dengan menggunakan model tersebut, parameter-parameter optimasi yang paling berpengaruh

'

" "$" ! %"$

" digunakan sebagai dasar pertimbangan untuk memilih metoda optimasi yang sesuai. Kemudian hasil akhir tersebut dapat diaplikasikan untuk keperluan optimasi posisi sumur dan jadwal pembukaan sumur produksi dengan mempertimbangkan faktor teknis. Penelitian akan menghasilkan sebuah metoda optimasi pengembangan lapangan yang lebih baik sehingga ' !" " " ' '

"$" Te l a h d i h a s i l k a n o p t i m a si l a p a n g a n d e nga n menggunakan metoda 1 yang di gabungkan dengan .7 * dan Proxy Function memungkinkan untuk mendapatkan hasil yang lebih cepat menuju kondisi optimum dibanding dengan hanya menggunakan 1 4

= !" L ! !" " Q " " Q* Q" ="{ commonly obtained from log analysis-are direct input "$ " "L * However, recent studies on some coal samples taken =" L' '8 !* !" that the commonly used log analysis equations are simply inapplicable for the field’s coal samples. " " $ *

"$" when compared to laboratory results. After a series of re-evaluations and re-measurements on the laboratory results it was convinced that the problem does not lie with the laboratory results but with these ‘conventional’ equations. Therefore modification ' "" = " ' " between measured data (coal samples taken from two "! " !! only equation for ash contents gives accurate results. The other proximate analysis output data - 4. " " Q* Q" ="{ at considerable odd with their corresponding calculated

"'"! $ " '

! " ! "$

" " the equations have produced analogous but different empirical equations to the original equations. These " "% =%! coals, and these better results underline that future log "% "! !* !

"

Kata kunci: optimasi produksi, - 7 -

Key words LQ $ "% Q ' "% Q

"

Agam Munawar (" " Donny Hendromurti (" " Rina Dewi P. (" " Dewatering Facilities for CBM Resources Appraisal, Lesson Learn from Sanga-Sanga Coalbed Methane Field, East Kalimantan 5%1. April 2013, Vol. 4 No. 1, p 49 - 56 ! '' = !" L! m a n y u n i q u e a c t i v i t i e s w h e n c o m p a re d w i t h "* " " " $ ' " " LQ $ p r o d u c t i o n us u a l l y c o m e s a ft e r a n e x t ende d dewatering process. This dewatering process can take % '"%% QL appraisal project will require surface facilities to manage water and gas production. From October 2010 & = ;X;QL " " ' nine dewatering facilities in the Sanga-Sanga Field.

High uncertainties in the early stages of the exploration '! L* "

some redundancy. This approach was successful in * "$ ! $! ' " * = % L wells and keeping the dewatering process running !% "Q! " !* = " collected, it will be easier to size and standardize equipment properly. In other words, the design '! '!%* ] = % "%! purpose of this paper is to discuss major parameters required to be taken into account in the design of

"$ " !!* operational problems during the initial dewatering period in the Sanga-Sanga Field, East Kalimantan, Indonesia. Key words: dewatering, surface facilities, metering, ' "QL

Studi Teknik Peningkatan Perolehan Minyak dengan Metode Injeksi CO2 Menggunakan Uji Laboratorium dan Simulasi Reservoar EOR Study by CO2 Injection using Laboratory Experiment and Reservoir Simulation Dita Amanda, Taufan Marhaendrajana Institut Teknologi Bandung Jl. Ganesha 10, Bandung 40132, Indonesia Telp.: +62816615621, Email: [email protected] Email: [email protected] Abstrak " " " = 8"" "

' ! " ! ' "8 $ 2 untuk meningkatkan perolehan minyak dengan menggunakan uji laboratorium dan simulasi reservoar, sehingga dapat membantu ' ' "$= " ' " " * ' "$$"" "8 2. Uji laboratorium dilakukan untuk menentukan "" ' " Q " "* $ ' ] ' " "' "$!"% ! ' faktor perolehan, serta mengamati pengaruh keberadaan & di dalam core. Simulasi dilakukan dalam

!'%" ] * " *# ] *= 8"" "' =!" "]

* " *

"" " hasil uji laboratorium pada skala lapangan. Hasil yang didapatkan dari penelitian ini adalah pengamatan komprehensif "$ " ' ' " " "% !$ 2 (swelling, pengurangan viskositas minyak, TTM, dan perbandingan metode uji laboratorium), perolehan minyak dalam skala laboratorium dan skala lapangan, serta perbandingan perolehan minyak skala lapangan dalam berbagai skenario injeksi. TTM yang didapat

8 = " ] * !;;;`' ' !" "% 8 core + bernilai 6% lebih rendah daripada hasil uji & akibat pengaruh heterogenitas pada core. Dalam " "%Q2 cenderung berinteraksi dan mengalirkan brine terlebih dahulu untuk kemudian mendapatkan kontak dengan minyak residual dalam pori dan mengalirkan minyak ke titik produksi. Pendesakan tersier di = "$! " " = _`Qq{|}~# ] * ""8"' """ * "% = "8 2%"$ $""$' "" " 2 yang ditambahkan. Sementara fenomena swelling " ' " "$" "" * ' $" " =$ = ' "=!" $ 2 pada hidrokarbon. Pengembangan skenario pada simulasi reservoar menghasilkan " ' " " = `_{~ # " C "$! " ' !" %"$ = ! "$$ 7 + 6 "'' "%'"2 dapat diperbaiki dengan skenario simultaneous water alternating gas (SWAG) atau diberikan tambahan propana. " 2, 6+, &QQC

Abstract %&/

#2 / #2 / &+ 4,- & 7 & 4 7 & -

#2 & / -

& 6 + / 4 + 86 4, 6& + & + & 4 % ' #2 9+ - -- :- - & 4 + & 7 9&

1

2

JTMGB, Vol. 4 No. 1 April 2013: 1-12

6+: 6 4& 6+ &-;;;<- 6+ =>+ &4 -#2 tend to +& + 4" -& ! +& 4% 7 & ?<4@!AB>##4, + / #2 4+ #2 & C 4 ##41 7 + 6+ 4 & J1 &

/ 4 $+8#2-6+-&-.#-1

I. Pendahuluan Metode peningkatan perolehan minyak (Enhanced Oil Recovery) maupun gas dengan menggunakan injeksi gas telah menjadi salah satu praktek umum di dunia sejak tahun 1970-an. Salah satu gas yang sering digunakan untuk meningkatkan ' !" !$ 2"8 2 merupakan metode peningkatan perolehan minyak kedua yang paling banyak dipergunakan di Amerika Serikat setelah 6 karena efek peningkatan ' !"%"$ $" " "8 C 2 dapat dilakukan pada dua kondisi yakni tercampur dan tidak tercampur. Pada kondisi injeksi di atas tekanan tercampur akan didapat harga perolehan minyak yang ' Q " 8 $ 2 diinjeksikan pada kondisi di bawah tekanan tercampur maka akan menyebabkan & atau keadaan tidak tercampur. Pada kondisi tidak tercampur harga perolehan minyak yang didapat akan lebih rendah dibandingkan kondisi tercampur. Sebelum metode peningkatan perolehan minyak diterapkan maka harus dilakukan kajian yang mendalam untuk mengetahui apakah

"8 2 laik digunakan. Kajian yang harus dilakukan mencakup studi laboratorium untuk menentukan tekanan tercampur minimum dan "$ ! " "2 dan minyak, serta simulasi reservoar untuk memprediksi perolehan minyak yang didapat dari hasil injeksi. Di Indonesia, kebanyakan studi yang dilakukan adalah studi simulasi pada tekanan tidak tercampur sebagai pendesakan sekunder. #

"$ " "8 2 sebagai metode pendesakan tersier dalam keadaan tercampur, studi efek saturasi air yang tinggi di reservoar ! ' ' " 2, serta analisis hasil

laboratorium ke dalam skala lapangan perlu dilakukan untuk mendapatkan gambaran interaksi ="{] ! ' ' " "$" ' !" "% "' ' "8 $ 2 ke depannya. II. Metode Data untuk melakukan analisis didapatkan dengan melakukan uji laboratorium dan simulasi. II.1. Uji Laboratorium Uji yang dilakukan di laboratorium dan metode yang digunakan untuk mendapatkannya adalah sebagai berikut. 1. Data Fisik Fluida Minyak yang dipergunakan pada semua eksperimen adalah minyak mentah permukaan (dead oil) tanpa saturasi air terlarut mula-mula. Data API dan viskositas minyak mentah didapatkan dari data penelitian terdahulu. Sementara data densitas didapatkan dari pengukuran menggunakan dan data komposisi minyak didapatkan dari laboratorium $ # " $ 2 yang digunakan !$ 2 murni (99,9%) = ' ]

#

" " "$" !" "% "$" "8 2 "$$""<8 = "# *

3

(Dita Amanda, Taufan Marhaendrajana)

Tabel 2. Perhitungan TTM menggunakan korelasi

Dari kedua korelasi tersebut didapatkan perkiraan tekanan tercampur minimum sehingga dapat ditentukan titik-titik lainnya untuk mensimulasikan keadaaan tidak tercampur dan tercampur. 3. Penentuan Tekanan Tercampur Minimum (TTM)

Gambar 1. Komposisi minyak uji

2. Karakterisasi Core Percobaan dilakukan dengan menggunakan dua alat bantu yaitu & dan core. Tidak seperti slim tube dimana nilai porositas dan permeabilitas telah tersedia, core memerlukan prosedur khusus untuk menentukan kedua nilai tersebut. Nilai volume pori (PV) core dapat diaproksimasi dengan mencatat posisi awal dan akhir pompa selama menginjeksikan & ke dalam core hingga mencapai tekanan operasi. Volume pori merupakan volume tercatat dikurangi dengan volume pada pipa-pipa evakuasi (dead volume). Porositas dihitung dengan menggunakan nilai yang telah didapatkan sebelumnya. Persamaan porositas yang digunakan adalah sebagai berikut:

I

PV K

(1)

Permeabilitas dihitung dengan menggunakan hukum Darcy. Untuk merencanakan tekanan percobaan yang akan digunakan, terlebih dahulu dilakukan perhitungan dengan menggunakan korelasi dari Yellig Metcalfe dan persamaan PRI (Petroleum Research Institute), kedua korelasi ini diambil karena penitikberatan pada temperatur dan bukan komposisi minyak.

Eksperimen dilakukan dengan menggunakan peralatan & dan 6+ yang dimiliki =

" "%" ITB. Kedua peralatan bekerja pada temperatur reservoar (158o "%"! "$ 2 dipompakan dengan pompa Ruska pada laju injeksi tertentu hingga mencapai tekanan operasi kemudian dikunci. Hasil perolehan ditempatkan pada tabung yang memiliki skala. Percobaan dan pengamatan dilakukan pada tiga titik tekanan yang berbeda, percobaan pertama

"

!

" ] ' " 2 ' "$" ] "! minyak di dalam & dan 6+. Percobaan kedua dilakukan pada daerah sekitar daerah tercampur dan percobaan ketiga dilakukan pada daerah di atas titik tercampur. Pada setiap percobaan dilakukan pengamatan jumlah minyak dan gas yang terproduksi sebagai = ' " " $ 2. Hasil perolehan dari kedua metode kemudian diperbandingkan untuk mendapatkan nilai MMP dan melihat perbedaan hasil antara kedua metode uji. 4. Uji Penyapuan Tersier Eksperimen dilakukan dengan menggunakan 6+. Fluida dan gas injeksi dipompakan menggunakan pompa Ruska pada laju injeksi yang diinginkan dan tekanan tertentu. kemudian dikunci dan dipanaskan pada temperatur uji. Hasil perolehan ditempatkan di dalam tabung uji yang dilengkapi dengan skala. disaturasi oleh & kemudian didesak

4


dengan minyak untuk mensimulasikan kondisi inisial. Simulasi pendesakan sekunder dilakukan dengan menginjeksikan & dan injeksi tersier

" "$" "$ "8 "$ 2.

*%"$

$"" ! seperempat 5- dengan ilustrasi ada 3. Analisis Sensitivitas

II.2. Simulasi Fluida Reservoar # ]

"" " kelakuan fluida yang digunakan pada uji laboratorium. Simulasi yang dilakukan antara lain adalah pemodelan kelakuan fasa, tuning C , uji perubahan viskositas terhadap "" "' 2 , uji perubahan

" ! ' "" "' 2, uji volume minyak sebelum dan setelah injeksi 2, dan pemodelan & menggunakan simulasi reservoar untuk mencocokan hasil uji laboratorium dan hasil simulasi.

Gambar 2. Ilustrasi lapangan pada simulasi reservoar

II.3. Simulasi Reservoar

Tabel 3. Data Input yang digunakan pada Simulasi reservoar

Simulasi reservoar dilakukan untuk memetakan properti uji laboratorium ke dalam skala lapangan. Dengan catatan skenario yang digunakan pada simulasi reservoar tidak sama dengan yang dilakukan pada uji laboratorium. 1. Batasan Simulasi ' "$"' !" " kumulatif produksi minyak adalah 30 tahun "8$ 2 diinjeksikan. 8" !" skenario yang memberikan faktor perolehan terbaik. " "'

""' mempertimbangkan masalah keekonomian. 2. Asumsi Simulasi C 2

="%_}`#

%"$ '""${ = 2 %"$

" " " C 2 yang diproduksi akan diinjeksikan kembali C#Q;XX C 2 yang digunakan berada dalam keadaan murni (100%) *!XQ #C%"$

$"" !; "!"$Q;XX} C

" ! !XQ_'

Skenario 1 1 / C C

" "$" "$ "8 "2 ke dalam sumur injeksi dengan laju injeksi yang berbeda-beda. Batasan tekanan injeksi adalah 2300 psi untuk memastikan pendesakan yang terjadi adalah pendesakan tercampur dan sumur produksi diberi batasan water cut sebesar 0,833. "8 2 dilakukan ketika laju produksi minyak dengan menggunakan +

sudah mengalami penurunan (decline rate) yakni pada tahun 2021. Pada tahun ini, injeksi air

! " " "2 - mulai diinjeksikan. Skenario 2 simultaneous water alternating gas (SWAG). SWAG dilakukan dengan menginjeksikan $ 2 dan air pada saat yang bersamaan sehingga menghasilkan air karbonasi untuk "%' "% C 2 diinjeksikan

#

" " "$" !" "% "$" "8 2 "$$""<8 = "# *

5


' $8%"$= = Q%" `X#&Q XX #& " ;`X #& 8 "8 air merupakan fungsi dari laju injeksi gas dengan ' " =$ = !"$Q 2009): Pengaruh Penambahan Propana Terhadap " "8 2. # " Q' = "= 2 juga dapat dilakukan dengan menambahkan propana pada gas injeksi. Penambahan propana di dalam * " ="%~ "" "$"2 sebesar 99 % mol.

through terjadi maka proses penyapuan dengan metode injeksi gas menjadi tidak efektif lagi. Gas akan keluar menuju tabung produksi tanpa menyapu residual oil di dalam &. Core Flow

4. Analisis Ketidakpastian Analisis ketidakpastian dilakukan untuk memperoleh gambaran dampak r pada perubahan parameter. Perubahan parameter yang dilakukan pada penelitian ini adalah

C=[' & pada tekanan 2000 psi

(2)

' ] Q %" ' "$! "8 2 ' ] = ""(live oil) dan dead oil dengan API berbeda. Untuk mengetahui performansi pendesakan pada live oil digunakan minyak dengan kandungan gas terlarut pada nilai API di atas dan di bawah API sampel minyak lab (39,1). yang digunakan adalah sampel minyak pada lapangan Y (API 37) dan Z (API 49). Perbandingan dengan Dead Oil Berbagai API. Untuk melihat pengaruh pendesakan menggunakan $ 2 pada minyak yang lebih berat (API lebih kecil) tetapi masih berada dalam jangkauan ' "$$""2, yakni di atas 22o API. Minyak yang digunakan adalah dead oil dari lapangan W dan X dengan nilai API masing - masing 29 dan 31.

Gambar 4. Produksi kumulatif & pada berbagai tekanan uji

Hasil uji dengan menggunakan core ditunjukkan pada Gambar 5. ' tampak lebih dini terjadi pada 6+ dibandingkan dengan &. Penentuan TTM

III. Hasil III.1. Pengujian Tekanan Tercampur Minimum Slim Tube '

"8"' Gambar 3, yakni mencapai kumulatif tertinggi sebelum gas mencapai ujung & dan keluar pada tabung produksi. Ketika & '!

C=`' 6+ pada tekanan 2000 psi

6


Nilai tekanan tercampur minimum diperoleh dari perpotongan dua kemiringan pada plot kurva RF terhadap tekanan operasi atau dapat juga diperoleh dari menarik garis 95% RF hingga berpotongan dengan plot kurva. Hasil uji & dan 6+ pada tiga tekanan uji disajikan pada Tabel 4 sementara perpotongan kurva RF terjadi pada 2225 psi seperti yang ditunjukkan pada Gambar 7. C=|' + 6 Tabel 4. Hasil uji & (ST) dan 6+

C=}' + 6 pada berbagai tekanan uji = "' !" + 6

Gambar 7. Penentuan TTM

Uji Penyapuan Tersier Hasil pengukuran permeabilitas dan porositas disajikan pada Tabel 5. Sementara hasil + 6 " 2 6 masingmasing seperti pada Gambar 8 dan 10. Sementara perbandingan produksi kumulatif pada berbagai tekanan uji digambarkan pada Gambar 9 dan 11.

Dari hasil profil saturasi + pada Tabel 6, saturasi minyak tersisa dalam core adalah sekitar 0,2. Nilai Sorw cenderung sama walaupun nilai permeabilitas, porositas, dan tekanan inisial core berbeda, atau dengan kata lain dapat dikatakan bahwa sebelum pendesakan tersier dilakukan keadaaan inisial core adalah identik satu sama lain.

Tabel 5. Hasil pengukuran properti core

C=X 2]

#

" " "$" !" "% "$" "8 2 "$$""<8 = "# *

7


Simulasi Fluida Reservoar Fluida digabungkan dan persamaan keadaan (EoS) dituning dengan menggunakan data API (39,1). Fluida hasil regresi diuji ketercampuran dan menghasilkan nilai TTM yang sama yakni 2225 psi dengan menggunakan persamaan Peng dan Robinson. C=' "%' berbagai tekanan uji

L = "$" ' ' ' + 6 dimana minyak terdesak secara langsung oleh air dan diproduksikan (mencapai titik produksi) pada awal pendesakan air, pada ' " " "$" 2 terlihat & lebih dahulu terproduksi hingga PV injeksi tertentu, kemudian minyak mulai terproduksi hingga gas mencapai titik produksi 9& ' :. Produksi kumulatif pada tekanan 2000 psi jauh lebih besar dibandingkan dengan kedua tekanan operasi lainnya, yakni 8,66% di atas faktor perolehan pada tekanan 1000 psi dan 4,85% di atas faktor perolehan pada tekanan 2000 psi. Tekanan 2000 psi adalah tekanan yang mendekati TTM (2225 psi), terlihat bahwa harga faktor perolehan paling besar di sekitar nilai TTM, kemudian menurun dengan bertambahnya dan berkurangnya tekanan injeksi.

1. Viskositas Minyak yang digunakan untuk eksperimen laboratorium disimulasikan pada tekanan reservoar (158oF) untuk memperlihatkan efek penurunan viskositas. Eksperimen juga disimulasikan pada berbagai tekanan untuk memperlihatkan efek tekanan injeksi pada penurunan viskositas. ' "=!"2= ' "$! $" " terhadap penurunan viskositas hingga sekitar 50% mol. Setelah 50% mol, viskositas minyak cenderung datar bahkan sedikit naik terutama pada tekanan di atas 2000 psi seiring dengan ' "=!" 2. Kenaikan hingga 5%

" * ' "8 `X~2 8

|X~2 diinjeksikan.

= _ "' !"2 6

Gambar 13. Perubahan viskositas minyak

2. Densitas

Gambar 12. 0 hasil 6

& " "% =' "=!"2 berbanding terbalik dengan penurunan viskositas minyak. Densitas minyak cenderung naik '" " ""% $" "Q %" hanya sekitar 0,5 - 3%.

8


Gambar 14. Perubahan densitas minyak

Gambar 16. Perbandingan hasil simulasi & dan laboratorium

3. Swelling Pada Gambar 15 terlihat kenaikan swelling seiring dengan kenaikan tekanan saturasi

" "8 2. + pada keadaan inisial adalah 1. Sebagai akibat dari ' "=!"$ 2, hidrokarbon terekstraksi dan mengembang sehingga volume C menjadi lebih besar dari keadaan inisial dan swelling berharga lebih besar dari 1. Pengembangan volume hidrokarbon juga berdampak pada meningkatnya tekanan saturasi.

Gambar 17. Perbandingan hasil simulasi 6+ dan laboratorium.

Simulasi Reservoar # " C

Gambar 15. Swelling minyak

Gambar 18 merupakan kurva perbandingan kumulatif produksi yang menunjukkan bahwa peningkatan kumulatif produksi berbanding "$" 8 "8 $ 2 hingga laju optimum tertentu. Dalam Gambar 18, keadaan ' 8

' 8[XX# 8

' = " ' Q $ 2 tidak sepenuhnya terlarut di dalam minyak residual. Gas yang tidak terlarut kemudian membuat aliran langsung ke titik produksi yang berdampak & ' terjadi lebih cepat.

4. Pemodelan %& Hasil simulasi dan hasil uji laboratorium menunjuk pada nilai TTM yang sama yakni 2225 psi seperti pada Gambar 16 (95% faktor perolehan). Matching juga dilakukan dengan hasil 6+ pada Gambar 17. Matching menghasilkan nilai yang sesuai dengan sedikit over prediksi pada tekanan 3000 psi yang dimungkinkan karena sedikit perbedaan bentuk core yang digunakan.

C=|' C

#

" " "$" !" "% "$" "8 2 "$$""<8 = "# *

9


Berdasarkan kurva produksi kumulatif pada Gambar 18, laju injeksi gas yang dipilih sebagai & ![XX#&8 " dipilih karena nilainya masih jauh dari asumsi

" 2' ! %" _}`#&Q serta karena laju ini menghasilkan faktor perolehan yang paling tinggi yakni sebesar `_Qqq~ $ 2 pada sumur produksi akan diinjeksikan kembali ke dalam * ! ' ' !"2 dan N2. Skenario & mampu memberikan tambahan produksi kumulatif hingga 224,1 MSTB dibandingkan apabila skenario + 6 diteruskan hingga akhir masa kontrak. Penambahan ' !" = " C

'" hingga 9,5% dari perolehan sekunder dan tambahan 7% dibandingkan dengan 7 + 6.

3. Perbandingan Skenario 1 dan 2 Perbandingan dilakukan antara & ' " C ""$ ' "$' pada skenario 2. Kasus 7 + 6 disertakan sebagai pembanding relatif. Hasil perbandingan digambarkan pada Gambar 20.

C=;X ="

"$"C "#C

2. Skenario 2 SWAG Pada Gambar 19 produksi kumulatif minyak terlihat dipengaruhi oleh laju injeksi air ke dalam sumur. Pada laju injeksi gas 250 #&Q & ' terjadi sekitar tahun ;X[Q% X!" !$ 2 mulai diinjeksikan. Sementara pada laju injeksi gas XX#&Q& ' terjadi pada sekitar !";Xq;;!" ! "8 $ 2.

SWAG menghasilkan faktor perolehan yang = !=

="

"$"C "extended + 6. Kenaikkan produksi kumulatif skenario SWAG bahkan hingga 20% bila dibandingkan dengan + 6 " ~ ! ' C Penambahan air dalam jumlah terbatas yang dilakukan dengan metode SWAG mampu ' = ' ' "%'" 2 sehingga & ' terjadi lebih lambat. 4. Pengaruh Penambahan Propana ' "=!"''"[| terhadap kumulatif produksi ditunjukkan pada Gambar 21. Penambahan propana memperbaiki '' C "$"=!" kumulatif produksi minyak sebesar 3,7% dan tambahan faktor perolehan sebesar 2,14%.

C=}' #C

Penundaan & ' menghasilkan penambahan kumulatif produksi sebesar 7% dan peningkatan faktor perolehan sebesar 4% pada tahun 2050. Untuk skenario SWAG pada reservoar

" Q8 "8 XX#& !8 yang paling optimum untuk mendapatkan recovery maksimum.

C=; ="

"$"'' "$" penambahan propana

10


5. Perbandingan dengan Pendesakan Menggunakan # C= ;; ' !" ' %"$ hampir sama antara sampel Y (API 37) dan & (API 39,1). Penyapuan yang lebih baik terlihat pada sampel Y dibandingkan & dengan penambahan faktor perolehan sebesar 17% terhadap faktor perolehan sekunder. " ' "%'"%"$ = != ! "% disebabkan karena kadar gas terlarut di dalam live oil hasil proses rekombinasi.

C=;[' = =$

Meski tidak sebaik performansi & ,

"8 2 masih memberikan peningkatan ' !" %"$ ' $" " %" 12,34% pada sampel X terhadap faktor perolehan sekunder, dan 6,34% pada sampel W. IV. Kesimpulan

C=;;' live oil

# " "8 2 9& : tidak dapat diterapkan pada sampel live oil Z dengan API 49. Sampel Z memiliki kandungan hidrokarbon ringan yang cukup banyak (light oil), sehingga gas injeksi langsung diproduksikan bersamasama dengan produksi gas reservoar. Terlihat tidak terjadi peningkatan produksi minyak setelah !" "8 $ 2. 6. Perbandingan dengan Dead Oil berbagai API Gambar 23 menunjukkan kurva produksi kumulatif dari beberapa sampel minyak dengan API yang berbeda-beda. Gambar menunjukkan ' " 2 yang menurun dalam menyapu minyak residual dengan semakin banyaknya komponen berat di dalam minyak. Kemampuan 2 untuk menurunkan viskositas dan fenomena swelling pada minyak berat dengan menggunakan skenario cenderung tidak efektif dibandingkan dengan kinerja pada minyak ringan.

Berdasarkan seluruh hasil penelitian yang dilakukan, ditarik beberapa kesimpulan sebagai berikut: <8 & pada TTM (2225 psi) dan temperatur reservoar (158oF) menghasilkan " ' !" "% = }`~ <8 6+ dapat dilakukan untuk mendapatkan nilai TTM dengan perolehan maksimum sebesar 88,4% pada tekanan 2225 psi. & "$" " %"$ ' " Q TTM dapat disimulasikan dengan simulator komposional dan mendapatkan hasil yang serupa dengan uji &. ' " "

labotorium 96: menunjukan variasi peningkatan faktor perolehan sebesar {}~ = "8 2. Berdasarkan hasil simulasi pendesakan tersier di laboratorium, "8 2 laik untuk dilakukan. <8 ' " "

= menghasilkan efisiensi perolehan minyak dari 75,4% hingga 89% OOI dan turun menjadi 57% hingga 61% OOIP pada simulasi reservoar. # " C "$"8 "8 $ = [XX#&' "";[XX'

"$$' sebagai & karena memberikan faktor perolehan minyak yang paling tinggi (57,14% OOIP) dengan waktu & '

#

" " "$" !" "% "$" "8 2 "$$""<8 = "# *

11


yang relatif lambat (12 tahun setelah injeksi). " #C

" =$ alter natif tek n ik u n tu k mem p erba i ki = 2 dan mengeliminasi kerumitan oper as i WAG. SWAG m en g h asil ka n kumulatif produksi minyak yang lebih =

="

"$" "$"C "extended + 6, yakni sebesar 20% dibandingkan + "~ ! 'C' tahun 2050.

Daftar Pustaka L" QQ}};'' " ="& in Enhanced Oil Recovery, Vol. 33, Energy "* "" "$ " L QQ;XX { = 2 Application to Improve Oil Recovery. Kansas. $"QQ "Q#Q;XX| C " " * = %=" The Open Petroleum Engineering Journal, 30-41. '= QLQQQ}|` " 2 & ' " #?" of Petroleum Technology, 25 (5), 665-678. !"$QQ}}q! %`th Ed. Mc Graw Hill. !%QQ "QQ! " "Q Q}}; "$ " Green, D. W., & Willhite, G. P., 1998. Enhanced Oil Recovery. SPE. QQ? " QQ}_q !" & ' "=%="& Q?" of Petroleum Technology, 1427-1438. QQ? " "Q?Q " = $QQ}}`<" $" #$ 2 in Aqufers and Oil Reservoir. " $%"* "" "$ " " QQ}`[="&

"$L !" " 8 & $"L "&

Kulkarni, M. M., 2003. Immiscible and Miscible C { & ' ""

" ! " " & ' " Engineering. C#Q;XX2 Sequestration Potential for EOR in Indonesia, Jakarta. Miscibility Pressure: Slim Tube or Rising Bubble Method? SPE. Marhaendrajana, T., Gunadi, B., & Suarsana, P., 2005. Potensi Peningkatan Perolehan Minyak '"$"? ="$& "$"2 Flooding, Jurnal Teknologi Minyak dan Gas Bumi, 1. "QQ= Q?Q}|[="& Flooding, Journal of Petroleum Technology, 396-400. Mungan, N., (1965). Permeability Reduction Through !"$ "'" # " %Q#Qqq}{q`[ QQ!"$QQ;XX}! & "="& "8 " " Oil Recovery. International Journals of Engineering and Sciences, 9 (10), 66-72. Negahban, S., & J, K. V., 1992. Development and Validation of Equation-of-State Fluid Descriptions 2/ Reservoir-Oil System. SPE Reservoir Engineering, 363-368. Scheuerman, R. F., & Bergersen, B. M., 1990. Injection Water Salinity, Formation Pretreatment, and Well operations Fluid Selection Guidelines, Journal of Petroleum Technology, 836 - 845. Stalkup, 1984. Miscible Displacement: SPE Monograph Vol 8. AIME. Tham, B. K., Raif, M. B., Saaid, I. M., & Abllah, E., 2011. The Effects of kv/kh on Gas Assisted Gravity Drainage Process. International Journal of Engineering & Technology, 11 (3), 153-185. Tjahjono, W., & Mardisewojo, P., 1994. Pengaruh Komposisi Minyak, Temperatur, dan Berat Molekul "" ' " 2 dalam Proses Enhanced Oil Recovery (EOR). Jurnal Teknologi Mineral, I (3), 33-57.

12


for Prediction of Post-Hydraulic Fracturing Gas Rate Keberhasilan Aplikasi Metode Kecerdasan Buatan untuk Memprediksi Produksi Gas Paska Perekahan Hidrolik Sudjati Rachmat (1), Ahmad Hadad (2) (1) Institut Teknologi Bandung Jl. Ganesha 10, Bandung 40132, Indonesia Telp.: +62222504955 (2) " " Q q|th-49th Floor Hl. Jend. Gatot Subroto No. 42 Jakarta 12710, Indonesia

Abstract % "$ ' * %>! ' "$ = '"" "$Q $"Q operation, post job treatment, and then evaluation. This complexity provides several approaches for the ' " * ""! !Q! " $ " ! '' " ! ''! Some parameters initially selected, they are: reservoir thickness, water saturation, reservoir porosity, reservoir permeability, pressure gradient, skin factor, productivity index, critical rate factor, cement bonding %Q$ * Q" '''" >! * "' ! "' parameters. The output parameter is the percentage of incremental gas rate. The parameter database: rock mechanic, ] ! %Q! ' "' " " "! !! " ' !* '= "$" "% ! " * " $ " ! >" " {= ''$ " LQ ' * " % " " % #Q "

% mathematical method. The following processes are performed for the methods, they are: Pattern Recognition, Identifying Performance Drivers, and Performance Prediction. There are at least four conclusions can be deducted: (1) Both NNBP and ANFIS are able to recognize the pattern of hydraulic fracturing performance. Furthermore, by reducing the training data and adding testing data, ANFIS is better than NNBP. (2) The performance drivers from both methods are: Proppant Mass, Water Saturation, Skin Factor, and Porosity. It also supported by linear regression method. (3) The fuzzy mathematical concept moderately exhibits the hydraulic fracturing performance. (4) ANFIS method was capable to ! !% "$' " =% "$"%! "] " ' ! "" # " ' '' "$ '

" "$8= "$* % >8=' "& = ;X' " match with actual data. % !% "$' " Q " " Q=''$ "Q%! Q adaptive neuro fuzzy inference system.

Abstrak ' ' ' ' ''-

'/

& /

-

- - + '/ 4 $' &' && ' ' '' ' / ' '4 * -' & ' &

' '' ' / ' '4 + && + - 8 '& - - - & - ' - ' ' - ' ' - / ' '- ' ' - - - '& '

13

14

JTMGB, Vol. 4 No. 1 April 2013: 13 - 23

4 /'

4 * & 8' ' & - ' - & ' 4 + ' &

' 4 $ && ' & 9: ' - +'!& '

9NN:- QQ 9N,:- QQ 4 &' '' '

- 8 - * 4 ' ' & &- 89D: 'NN N, ' / ' '4 -

& ' / N,&& ' NN49;: ' / '

8 - - '' - 4" /

'

49T:$ ' ' '& ' ' /'' ' / ' '49@: N, '' ' / ' '

' &

4 " 'N, ' ' ' ' ''

' '4" /

' & *&;UDD

' 4 $ ' 8 - +'-& '

-QQ -

QQ 4

I. Introduction There many hydraulic fracturing jobs have been done within the research area. Some results were good, some were poor. Evaluation process is always performed after each job done. There are many factors influence to hydraulic fracturing performance. The factors * ! Q * ] Q mechanic, condition of geological regional, well productivity, well condition and structure, "] Q " '''"! Q "! !Q! " " with feedforward backpropagation algorithm adaptive neuro fuzzy inference system and

%! ! '' to evaluate and predict the performance of hydraulic fracturing jobs. This research was initialized with data preparation, and then followed by the main process. The main process covers these three processes: " $" " " % "$ " & * Q"

"

" Most of the processes is carried out by using the computer software, Matlab. The key parameter in hydraulic fracturing is fracture geometry and conductivity. Theoretically, these methods will determine the success of the

hydraulic fracturing jobs through the dimensionless " * % fD). This parameter is designed carefully before conducting the main " =%] "% "$ the job. ! fD equation is governed by fracture width (w), fracture permeability (kf), fracture half length (Xf), and reservoir permeability (k).

(1)

"! Q ! ] "% is shown as below:

(2)

(3)

(4)

# '' "" " $ " !

" {% "$C

15

(Sudjati Rachmat, Ahmad Hadad)

where: ] "% Vf * Q Vi "8 "* Q Vl ] * Q "] Q " Af Q '' "$ Q " ! "] " by several parameters, such as compressibility, viscosity, porosity, etc. $ }" & % ]

(5)

fD < 1, the effective wellbore radius can be approached by: (8)

(6)

" Q " fD, the effective wellbore radius can be approached by: And the combination of above formulas is shown as below:

(9)

(7)

Then, the dimensionless productivity index can be calculated: (10)

where: "] " " compressibility, ft/min "] " " viscosity, ft/min "] " " compressibility, ft/min ' % * ' = %Q

* ' = %Q' * ] * %Q' ' = %Q

'

"= " ] and reservoir pressure, psi * %Q' % ]Q ! ! " = "" fD in Figure.

Where, this dimensionless productivity index will directly affect to the rate after the job. Sometime, the hydraulic fracturing activity is performed in small scale job which usually "=%' "! 8=! the objective only to remove the near wellbore damage. The equation for this treatment as follow:

(11)

where: " Jo " ' * % " JF ' "' * % " XfD

" " ! "$!

16


The saturation changes in the formation can also cause the damage. The low permeability formation usually behave tolerantly only to the damage due to minimum saturation changes. !% ="'! '' "$ for the decreasing productivity due to saturation changing. ' % ' 8 trapped liquid in the formation. This parameter is formed by the pressure difference between wetting phase and non-wetting phase in the formation. Holditch (1978) did the experiment to four low permeability core samples. He got the relation shown in Figure 2.

Figure 2. Holditch (1978) capillary pressure vs water saturation

The empirical equation, aqueous phase trapping (APT) as diagnose tool to estimate the reservoir sensitivity to the water trapping, as below:

Firstly, it is selected eleven parameters as input parameters. These parameters are part of ! "$! >! reservoir characteristic, the second factor is well productivity, then the last is operational fac! " * thickness (h), water saturation (Sw), reservoir porosity (I), reservoir absolute permeability (k). The selected second factor consists of pressure gradient (ppg), skin factor (s), productivity index Q Q" "="

"$ % L ! " ! ! consists of gel volume (Gel), and proppant mass (Prop). This last factor basically is the designable parameter that can be controlled by user. Meanwhile the output parameter is incremental of the production gas rate. Then, the crucial process in data preparation is calibration, with main objective to have all data comparable. It is done by nodal analysis to put all productivity data in one datum or system. ! ! " * %' production system. The last process is data grouping, there are three groups from selected 20 datasets = {;> $' ; ! !! complete eleven input parameters, the second $' _ ! !! "L Q" the last group is the recent dataset which the job performed in December 2010. It only 1 dataset and will be used only for testing dataset. This data is from the job that performed in December 2010 that showed the comprehensive result in early 2011. III.

(12)

From the calculation APT above, then by referring the Table-1, it is known the reservoir sensitivity to the water trapping. II. Data Preparation ! " $ * ! ' &# " & = " &C' "$

Main Process

III.1. Pattern Recognition ! ' =% ' " $" "Q = " "$ " network (ANN) and fuzzy inference system (FIS). The eleven selected parameters and the output were initially processed in the networks. It is actually a supervised learning algorithm within the network using Matlab. It consists of eleven inputs, one hidden layer with one unit in it, and one output. In this research, the type of neural network

# '' "" " $ " !

" {% "$C

17


used is backpropagation (ANNBP). The type generalizes the training rule of “Windrow" =%"

"% between input and output layer, which called as hidden layer. It uses algorithm of descent gradient for each changing of weighting gradient. The architecture of created ANNBP network is shown in Figure 3.

Figure 4. ANFIS Architecture

Figure 3. ANNBP Architecture

The network has following criteria: %' Backpropagation & ; "' "''

"% % <" "

"% " ' * "" " $ $' " Meanwhile, the other technique which % " " % #>! " " system by fuzzy logic concept. Then from this technique, the computerized network is developed with subtractive clustering for data distribution matter, Sugeno FIS, then optimization process using back propagation and least square algorithms. This computerized network also known as adaptive neuro fuzzy inference system (ANFIS). The architecture of created ANFIS network is shown in Figure 4.

The network above also has following certain criteria: # ! #$ " " " ! #= * "$ ' " ! !%= = " " of backpropagation and least square) = ;"= dataset) "' "'' ' They give the excellent matching with the actual data, shown in Figure 5. Then the testing processes were also introduced in this pattern recognition process to choose the best method between the two methods used.

Figure 5. Pattern recognition (without data test)

18


The pattern recognition with the results of testing session is shown in Figures 6a, 6b, 6c, and Table 3.

Figure 6. Pattern recognition (with data test)

It is important to have the number of training datasets be several times larger than the number parameters being estimated to achieve a good generalization capability. Or in other words the D/P > 1 (where D is number of datasets and P is number of parameters).

III.3. Performance Prediction The last process is “performance '

" " !* = prediction purposes, the formation of the ANFIS network is done more comprehensively. It has the validation session to ensure the trainingprocess does not escape the whole existing pattern " "* ! ! %! "$ Then, the fuzzy logic concept also applied in this prediction purpose. The method called fuzzy mathematical method which already has done by several researchers such as Yang, E and Yin Daiyin and Wu Tingting. In their calculation, they used the input from the expert for contribution factor. Since we have no input ' ! ' Q ! "!

"

" !

" " =%''% "$ the Garson equation to available networks from pattern recognition process.

(13)

III.2. Identifying Performance Drivers The second process is “identifying ' " * ! * = networks from pattern recognition process. The networks were run without the input of evaluated parameter, and then calculate the delta R2 from each run The simple linear regression between each input and output also applied in this process. Then, the average number was calculated from above three methods to represent the average "'' "] " ! ' shown in Figures 7 and 8.

Where: 8 $! "$ !" " "' layer to each unit in hidden layer 8 $! "$ "'% ! unit in hidden layer 8 $! "$ "'% hidden layer 8 $! "$ !" "!

" layer to each unit in output layer The datasets used for this purpose are 9+3+1 datasets with eleven input parameters and 10+9+1 datasets with four input parameters. ! "= " "$ Q " validation dataset, and the third is testing dataset. IV. Result and Analysis IV.1. Pattern Recognition

Figure 7. Four most driver parameters: proppant, water saturation, skin factor, and porosity

! =! ! >L and ANFIS are shown in Figure 5. The R2 for

# '' "" " $ " !

" {% "$C

19


both methods are approached to one. These results tell us that the selected eleven input parameters are able to reform (?) the behavior of hydraulic fracturing performance. They also prove that the two methods (ANNBP and ANFIS) are able to recognize the given pattern. Because both methods give almost similar results of training sessions, so it needs testing session to choose the better method between two methods used in pattern recognition. The testing session is performed in three scenario, they use 1, 2, and 3 testing points taken from the 12 datasets available. The results are shown in Figure 6 and Table 3. From the results, it shows that the ANFIS method is better then ANNBP. The differente specially happens by increasing testing points. It can be seen, by using the ANNBP method, the decreasing of R2 is from 0.99090, 0.98797, and then 0.93478 for 1, 2, and 3 testing points, respectively. Meanwhile, the ANFIS method is from 0.99875, 0.98894, and 0.98963. The decreasing of R2 occurred because of reduced training points and increased number of testing points. Therefore, the better method for pattern recognition is ANFIS. This method will be used

"" ' >! ! '

"'' The Figures are summarized in Table-3. IV.2. Identifying Performance Drivers The results from this process are shown in Figures 7 and 8. It can be seen that the amount of proppant, water saturation, skin factor, and reservoir porosity are the most influential ' Q! !!* ! "] " ' "$ more than 10%. These parameters will be used in prediction. ! ' " * ' proppant mass. It makes sense because this parameter is directly represents the fracture volume and the dimensionless fracture conductivity fD). The increasing of proppant mass is a ] " ! $ * Q &"= Q" = " $ ! effective wellbore radius (rw’) also increases. Using equation-10 by Prats (1961), the increasing of rw’ will cause well productivity after the job increasing.

The second performance driver parameter is water saturation. This parameter is the reservoir characteristic, which related to hydrocarbon * " ] { " ""! %' ' = % "Q! ] {

" "'! " "" "] " " the hydraulic fracturing performance. From Figure 2, at same water saturation, the capillary pressure will be greatly increased with decreasing permeability. This phenomenon will be reduced with increasing water saturation. The third performance driver parameter is skin factor. This parameter is the description of the near wellbore damage. Sometime, it is " " ! "$ " = cle well completion. Rae, 4 (1999) for skin by pass treatment (eq.11), mentioned that the skin factor is proportional with post treatment productivity index. So, in other words, with the increase in skin factor, the incremental gas production rate will increase after the job. The fourth performance driver parameter is reservoir porosity. This parameter is the reservoir characteristic, which related to the hydrocarbon * " "$] "%! equations 5, 6, 7, are shown influence of this ' ] L ! '' " " = " ' % " ] Q ! Q ! " "$' %Q] " Q and will reduce fracture propagation, then, ultimately, will reduce the potential incremental gas rate. ! " * %! "] " factors is done using the existing network of previous process (pattern recognition). The result can be seen in Figure 8.

Figure 8. Sensitivity on four most driver parameters

20


IV.3. Performance Prediction The prediction process is performed by ! Q #" "

fuzzy mathematical method. The ANFIS application for 9+3+1 actual datasets with eleven

"'' !" " $ }! $ = ! = ! " ;> X}}||; validation and 0.99571 for testing datasets. This result also supports the pattern recognition purpose. Because of the higher datasets / parameters (D/P), so the network from second datasets $ X = "$ " "= '

"'' ' sets (Figure 9).

! " !

fuzzy mathematical method. It gives the result in Figures 11 and 12. From both graphs, the plots still scatter but showing the good trend. This scattering result is affected by the absent of optimization session for the network parameters and linear grouping of input data. These drawbacks can be overcome by the ANFIS method.

$

%! ! ; datasets)

Figure 9. Evaluation and prediction ANFIS (9+3+1 datasets)

$ ;

%! ! } datasets)

Since the plots show the good trend, it still can be utilized for the prediction and candidate selection, but in preliminary stage. IV.4. Excel Application-Prediction Tool Figure 10. Evaluation and prediction ANFIS (10+9+1 datasets)

The macro application was made for gas rate prediction tool with additional option for

# '' "" " $ " !

" {% "$C

21


proppant mass sensitivity. It used the latest actual data from the job performed in December 2011 ! ! * "! ! "" created. The output (gas rate after hydraulic fracturing) ! '' " XqXX > " while the actual result from the job was 0.470 Mmcfd, meant that the R2 was about 0.96985.

5.

6.

7.

| Figure 13. Excel application - prediction tool

proppant, water saturation, or the skin factor, the greater also the addition of the gas production rate obtained from hydraulic "$ * "" ! "] " of porosity. ANFIS method is able to provide the results that are much more accurate in predicting performance than the modified fuzzy mathematical method, where the R2 obtained by testing is 0.99571. Because the pattern recognition and prediction results obtained are not accurate and are * Q! "!

% mathematical method should only be used as the initial stage of candidate selection of hydraulic fracturing job. ANFIS method is able to model the hydraulic fracturing performance by using only four ! "] " ' ! "Q R2 obtained on validation and testing: 0.97537 and 0.99992, respectively. # ' '' " "$! "" from ANFIS was tested using very last actual data gave excellent match with actual job result with R2 of 0.96985.

V. Conclusion Nomenclature ! " " =''$ " (NNBP) and adaptive neuro fuzzy inference system (ANFIS) are able to recognize the patterns of hydraulic fracturing performance very well. Both produce a RMSE close to 0 and R2 close to 1. 2. Using fewer training data and more testing data, the ANFIS method is better than NNBP for recognizing the patterns. This is because the ANFIS method is combining the methods % " " % #" neural networks (back-propagation). It can be seen in testing three testing datasets obtained by ANFIS the R2 is 0.98963, while the NNBP is 0.93478. [ ! "] " ' "! hydraulic fracturing performance are Amount of Proppant, Water Saturation, Skin factor, and Porosity with an average degree of "] " ![X~Q;[~Q;[~Q" X~ 4. By using available 20 actual datasets, it can be said that with the greater amount of

* c w ' x ' K V Vi Vl t c v cv

I

'r

Dimensionless fracture conductivity Fracture conductivity, md-ft Fracture width, ft Fracture permeability, md Fracture half length, ft Resevoir permeability, md "% Fracture volume, cuft Injection volume, cuft Fluid loss volume, cuft "] Q " Fracture area, sqft Pumping time, min "] " " * %Q " "] " " ' = %Q " "] " " * %" ' = %Q " Porosity Reservoir permeability, md

22

ct

P r ' 'P%

P rw’ JD re rw s

FfD

%i 'a Swi h Sw

PI Gel +/ +/ +/' +/


Reservoir compressibility, 1/psi * ] * %Q' Filtrate permeability, md

"= " ] and reservoir pressure, psi Filtrate viscosity, cp Efective wellbore radius, ft Dimensionless productivity index Reservoir / drainage radius Wellbore radius Skin factor Dimensionless half length Initial aqueous phase trapping Aqueous permeability Initial water saturation Reservoir thickness, ft Water saturation Gradient pressure, ppg Productivity index Dimensionless critical rate Pumped gel volume, gal Pumped amount of proppant, lb Weighting factor from each unit in input layer to each unit in hidden layer Weighting factor from input layer to each unit in hidden layer Weighting factor from input layer to hidden layer Weighting factor from each unit in hidden layer to each unit output layer

Refferences Beggs, D., 2003 Production Optimization using &"% QC" Publications, Tulsa, Oklahoma. Demuth, H., Beale, M., 2000 Neural Network =< !LQ! MathWork, Inc. Economides, M.J., Martin, T., 2007 Modern Fracturing - Enhancing Natural Gas Production, ET Publishing, Houston, TX. = Q Q ; XX } L 8 ' % $ Menggunakan Matlab, Deli Publishing. " QQ}}` # "< "$ !" = " Q<" * % of New South Wales, Sidney NSW 2052. Mohagegh, S. 4 ., 1999 Performance Drivers in Restimulation of Gas Storage Wells, SPE-57453. Rachmat, S., Kharisma, B., 2010 The Application

of Artificial Intellegence on Fracturing "

# "Qst International " " "& "$ !"$%&" National Workshop on Manpower Development in Petroleum Engineering (NWMDPE). Siang, J.J., 2004 Jaringan Syaraf Tiruan & Pemrogramannya Menggunakan Matlab, Penerbit Andi. ! !Q" '' =% !Q ;XX;%$ =< ! LQ! !Q" "$QQ;XX}# "$ " % for Fracturing with Fuzzy Mathematics Method, # !" " "" " "%#% and Knowledge Discovery, Yin Daiyin, Wu Tingting, 2009 Optimizing Well for Fracturing by Fuzzy Analysis Method of Applying ' Q! " " "" " " Information Science and Engineering.

Ahmad Fauzi Hadad, is a student for doctorate degree in Petroleum Engineering of ITB. Hadad holds ST and master degrees from Petroleum Engineering in year sof 2000 and 2011 from university. He is a member of several professional organizations such as Indonesian Petroleum Engineers Association (IATMI) and Society of Petroleum Engineers (SPE). Hadad currently works for Vico Indonesia as reservoir "$ " "L

* "

Table 1. Reservoir sensitivity to APT

# '' "" " $ " !

" {% "$C (Sudjati Rachmat, Ahmad Hadad)

Table 2. The 20 Datasets

Table 3. Pattern recognition

23

24

JTMGB, Vol. 4 No. 1 April 2013

Metoda Optimasi secara Kontinyu Terintegrasi Sistem Subsurface - Surface untuk Pengembangan Lapangan Migas Continuous and Integrated Optimization Method of Subsurface - Surface System in Oil and Gas Field Development Amega Yasutra, Tutuka Ariadji, Zuher Syihab, Pudjo Sukarno Institut Teknologi Bandung Jl. Ganesha 10, Bandung 40132, Indonesia Telp.: +62222504955

Abstrak Metoda dan strategi perencanaan pengembangan lapangan migas mengalami perkembangan yang sangat pesat akhir-akhir ini. Pada awalnya, perencanaan pengembangan lapangan hanya didasarkan pada sisi teknik reservoar saja. Perencanaan pengembangan yang baik membutuhkan masukan dari berbagai sektor dari industri $ Q ! "$$ " $ ' { ' $ $ Q$ Q * " ' ' " terbentuk. Namun, integrasi pada masih dilakukan secara sekuensial, belum secara utuh di antara aspek-aspek yang terkait tersebut. Fakta memperlihatkan bahwa studi lapangan terintegrasi saat ini umumnya hanya dilakukan sebagai bagian dari proses sekuensial, belum terintegrasi secara kontinyu. Tujuan dari penelitian ini adalah mengembangkan metoda optimasi sehingga dapat diterapkan pada pengembangan lapangan migas melalui pemodelan yang terpadu. Metoda ini memungkinkan variabel yang terkait dengan semua aspek sistem produksi lapangan dapat terwakili yang belum pernah dilakukan pada penelitian sebelumnya. Pemodelan dalam penelitian ini akan melibatkan semua parameter yang terkait dengan reservoar dan sistem sumur produksi, sehingga model tersebut diharapkan dapat merepresentasikan sedekat mungkin keadaan nyata di lapangan. Dengan menggunakan model tersebut, parameter-parameter optimasi yang ' "$= ' "$! '

" "$" !%"$

"

$"" =$ ' ="$" untuk memilih metoda optimasi yang sesuai. Kemudian hasil akhir tersebut dapat diaplikasikan untuk keperluan optimasi posisi sumur dan jadwal pembukaan sumur produksi dengan mempertimbangkan faktor teknis. Penelitian akan menghasilkan sebuah metoda optimasi pengembangan lapangan yang lebih baik ! "$$ ' !" " " ' '

"$" !

! " =! ' '"$" dengan menggunakan metoda 1 yang di gabungkan dengan .7 * dan Proxy Function memungkinkan untuk mendapatkan hasil yang lebih cepat menuju kondisi optimum dibanding dengan hanya menggunakan 1 Kata kunci: Optimasi Produksi, 1 , .7 * , *

Abstract %

& 4 & - & Q 4% ' +

4% - - & 4 "+- + 7C - 4,- C + 4 %&/ & &

Q & 4% + & & & 4% + + + - 7 & 4 !

25

26


- & ++& & Q 4,- &Q+ Q + 4 %+ & Q + 4 & Q 1 & +.7 * 7, Q 1 4 $+8 #Q -1 -.7 * - *

I. Pendahuluan Metoda perencanaan pengembangan yang baik merupakan hal yang sangat diperlukan dalam pengembangan suatu lapangan migas 9# Y 1 , * -#*:. Dari hasil kajian perencanaan pengembangan lapangan suatu lapangan dinyatakan layak atau tidak untuk dikembangkan. Pada awalnya perencanaan pengembangan lapangan dilakukan hanya berdasarkan keteknikan reservoar. P er encanaan p en g em b an g an y an g ba i k membutuhkan masukan dari berbagai sektor dari industri migas, sehingga integrasi aspek ' $ $ Q$ Q * " produksi permukaan mulai terbentuk. Tahap perkembangan penyusunan POD dapat dinyatakan sebagai pentahapan berikut: 1. berdasarkan kajian keteknikan reservoar 2. berdasarkan keteknikan GGR (geologi, $ Q * " ' tetapi masih terintegrasi secara sequensial dan manual. 3. berdasarkan keteknikan GGR dan fasilitas produksi terintegrasi secara paralel dan manual 4. berdasarkan keteknikan GGR dan fasilitas produksi secara kontinyu terintegrasi. Studi lapangan terintegrasi pada saat ini biasanya masih dilakukan sebagai bagian proses yang sekuensial. Sering para ahli keteknikan reservoar hanya memodelkan terbatas sekitar & , a h l i k e t e k n i k a n p r o d u k s i memodelkan system sumur dari dasar sumur hingga kepala sumur tetapi terbatas untuk suatu "' 'Q "

" teknik prosesing memodelkan dari kepala

sumur hingga tangki penyimpanan. Ahli teknik produksi dan ahli teknik prosesing tidak mempertimbangkan perubahan reservoar secara kontinyu karena hanya berdasarkan pada kondisi sesaat. Setiap bagian memodelkan tekanan konstan pada ujung akhir dari model sampai masa akhir periode simulasi. Untuk sistem reservoar menggunakan batasan tekanan dasar sumur, untuk sistem sumur produksi menggunakan batasan tekanan kepala sumur sedangkan untuk fasilitas pengolahan menggunakan batasan tekanan separator. Oleh karena itu hasil studi * yang tidak menggunakan teknik terintegrasi sehingga sering terdapat penyimpangan pada saat penerapannya. Pada saat rencana pengembangan lapangan belum berbasiskan simulasi yang terpadu, masih memisahkan antara simulasi reservoar dan sistem sumur produksi sampai permukaan maka pengaruh kinerja sumur dan fasilitas permukaan belum terakomodasi di dalam simulasi. Sebagai contoh dalam pemodelan fasilitas permukaan belum semua ketidak pastian yang terjadi di reservoar dan sistem sumur diperhitungkan dalam perencanaannya. Perubahan produksi sumur akan mempengaruhi sistem sumur dan fasilitas permukaan demikian pula sebaliknya, ! "$$ 8 8

] ' akan menyebabkan rancangan fasilitas pengolahan tidak bekerja secara efisien. Hal ini dapat me nga ki ba t ka n pot e ns i r e s e r voa r yan g sesungguhnya tidak dapat terproduksikan. Di samping itu, penempatan posisi sumur !"% ' ="$"

' " "% " $ # #= {# " "$ ="$"'"$" $

27

(Amega Yasutra, Tutuka Ariadji, Zuher Syihab, Pudjo Sukarno)

9" & K: terbesar dan belum mempertimbangkan & sistem sumur h i n g g a k e s e p a r a t o r. D e m i k i a n p u l a penjadwalan awal sumur berproduksi belum dilakukan secara optimum dan terstruktur yang dapat mengurangi dampak & ' di permukaan. Akibat dari penempatan dan pembukaan sumur yang kurang mempertimbangkan & sumur, maka banyak lokasi yang memiliki cadangan migas besar tidak terkuras secara optimum sehingga menyebabkan terjadinya penyebaran sisa migas yang tidak merata sehingga berakibat pada penurunan perolehan minyak dan gas bumi. Penyebaran secara tidak merata tersebut dapat dikurangi dengan melakukan strategi POD yang terpadu meliputi komponen reservoar, sumur dan fasilitas permukaan. Selain itu keterpaduan seluruh komponen tersebut dapat juga meningkatkan efisiensi perencanaan fasilitas permukaan. Diagram evaluasi dapat dilihat pada Gambar 1

tujuan untuk mendapatkan gambaran pemecahan masalah dengan waktu komputasi yang singkat. Metoda optimasi dari model prototipe ini akan diimplementasikan untuk model dalam skala lapangan. Metoda optimasi yang sering digunakan dalam optimasi adalah dengan menggunakan 1 91:. Dalam proses evaluasinya, GA membutuhkan hasil yang merupakan keluaran dari simulasi. Bila dibutuhkan banyak hasil keluaran, maka simulasi reservoar yang dilakukan akan semakin banyak dan hal ini akan memerlukan waktu yang tidak sedikit dalam melakukan evaluasi optimasi untuk mendapatkan hasil yang optimum. Pengurangan jumlah keluaran simulasi dalam melakukan evaluasi optimasi akan sangat berdampak dalam kecepatan waktu optimasi. Berikut penjabaran beberapa metoda yang sering digunakan dalam penelitian yang berkaitan dengan keteknikan perminyakan.

II. Metoda

Algoritma Genetik (GA)

Untuk optimasi pengembangan lapangan digunakan metoda optimasi yang banyak dikembangkan dalam ilmu matematika, yaitu dengan memilih variabel terbaik dari seperangkat alternatif variabel yang tersedia sesuai dengan fungsi obyektifnya. Dalam penelitian ini dapat dipilih salah satu dari dua fungsi obyektif yaitu faktor perolehan lapangan. Variabel yang menentukan nilai maksimum dari kedua fungsi =% =Q " " Q "" *Q =" "] *Q

" sumur, dan fasilitas permukaan. Namun banyak dari variabel optimasi tersebut merupakan variabel statis yang tidak berubah selama waktu optimasi

""!* = !

" sumur produksi, dimensi pipa salur permukaan, dan jarak sumur ke fasilitas pengolahan. Pada dasarnya penelitian ini ditujukan untuk melakukan optimasi pengembangan lapangan dalam skala yang sebenarnya. Namun untuk saat ini masih menggunakan model konseptual atau model prototipe yang dikembangkan dari sebagian kecil dari data sesungguhnya yang mewakili sistem reservoar dan sistem sumur produksi di lapangan, dengan

Algoritma genetik (GA) adalah pencarian heuristik yang meniru proses evolusi alam. Heuristik telah secara rutin digunakan untuk menghasilkan solusi yang berguna untuk optimasi dan pencarian masalah. Algoritma genetik merupakan kelas yang lebih besar dari algoritma evolusioner (EA), akan menghasilkan solusi untuk masalah optimasi menggunakan teknik yang terinspirasi oleh evolusi alam, seperti warisan, mutasi, seleksi, dan crossover. Dalam algoritma genetika, sebuah populasi dari string (disebut kromosom atau genotipe dari genom), yang menjandikan solusi calon (disebut individu, makhluk, atau fenotip) untuk sebuah masalah optimasi, berevolusi ke arah solusi yang lebih baik. Secara tradisional, solusi direpresentasikan dalam sistem-biner sebagai string dari 0s dan 1s, namun pengkodean cara lain juga dimungkinkan. Evolusi ini biasanya dimulai dari sebuah populasi individu secara acak dan terjadi dalam generasi. Dalam setiap generasi, kesesuaian setiap individu dalam populasi dievaluasi, beberapa individu yang stochastically dipilih dari populasi saat ini (berdasarkan fitness), dan dimodifikasi

28

' " "% " $ # #= {# " "$ ="$"'"$" $ (Amega Yasutra, Tutuka Ariadji, Zuher Syihab, Pudjo Sukarno)

(digabungkan dan mungkin bermutasi secara acak) untuk membentuk populasi baru. Populasi baru kemudian digunakan dalam iterasi selanjutnya dari algoritma. Umumnya, algoritma berakhir ketika jumlah maksimum generasi telah diproduksi, atau tingkat kebugaran yang memuaskan telah dicapai masyarakat. Jika algoritma telah diakhiri karena jumlah maksimum generasi sudah tercapai, solusi yang memuaskan mungkin tercapai atau mungkin tidak tercapai. Algoritma genetika dapat ditemukan aplikasinya

= "$ = " Q $ " Q komputer, teknik, ekonomi, kimia, manufaktur, Q Q "= "${= "$ ""% Experimental Design Dalam statistika, perancangan percobaan 9* .7 : adalah kajian mengenai penentuan kerangka dasar kegiatan pengumpulan informasi terhadap obyek yang memiliki variasi (stokastik), berdasarkan prinsip-prinsip statistika. ""$"' =" '"8 =" bagi peneliti untuk bergerak dari hipotesis menuju pada eksperimen agar memberikan hasil yang valid secara ilmiah. Dengan demikian, perancangan percobaan dapat dikatakan sebagai salah satu instrumen metode ilmiah. Kajian perancangan percobaan adalah pelaksanaan percobaan (eksperimen) terkendali. Dalam percobaan semacam ini, peneliti memberikan sejumlah tindakan (dapat juga ' = " "$" { =8 "%Q diistilahkan sebagai perlakuan atau treatment) pada sejumlah objek yang memiliki variasi pada derajat tertentu. Beberapa pustaka menggunakan istilah 7 design bagi rancanganrancangan yang dibuat untuk kegiatan pengumpulan informasi tidak terkendali, seperti survei, jajak pendapat 9 :, penelitian pengamatan 9 7 :, d a n C !7 . .7 * dilandasi atas sejumlah prinsip statistika mendasar agar analisis yang diterapkan terhadap hasil pengamatan valid secara ilmiah. Proxy Function 7 Y sering digunakan

karena dapat mempermudah proses yang komplek menjadi sederhana dengan akurasi yang baik. Terutama dalam evaluasi resiko yang melibatkan banyak parameter ketidak pastian. G Zangl 4 4 2006 mencoba menggunakan 7 ini guna melakukan optimasi produksi untuk mengurangi running simulasi yang lama untuk mencapai titik optimumnya. Metoda optimasi yang digunakan tetap menggunakan 1 . III. Hasil 1 merupakan metoda yang populer dalam optimasi, tetapi terdapat beberapa kekurangan yang akan memperlambat proses optimasi jika GA di gabungkan dengan simulasi reservoar. Simulasi terintegrasi akan memerlukan waktu cukup lama untuk setiap operasinya. Simulator yang digunakan dalam

" ! # " #' #<LC Metoda optimasi dengan menggunakan G A pa da a w a l pe nye ba r a n popul a sin y a (Gambar 2.) dilakukan secara acak dan ini berpotensi besar untuk melakukan generasi populasi yang lebih banyak guna mendapatkan hasil yang optimum. Untuk mengurangi kelemahan tersebut maka sewaktu generasi populasi awal GA di bantu dengan menggunakan experimental design sehingga sampling yang dilakukan sudah menampung seluruh populasi yang ada. Untuk mempercepat terjadinya kondisi optimum pada optimasi yang dilakukan maka GA dikontrol dengan menggunakan Proxy model 9 : sehingga hasil populasi berikutnya tidak akan jauh dari area optimum nya. Bagan optimasi dengan GA yang sudah

'

!' C=[ Model Model ini merupakan model sederhana yang di dibuat untuk memudahkan simulasi pada awal menentukan metoda optimasi, bentuk model dapat dilihat pada Gambar 4, dimensi dan keterangan model sebagai berikut:

& " XX?


"8"$ = #' ' '

|}XX_;XX Q;_q#L ;=! ;X! ;XX'

Untuk model jaringan diameter pipa dari kepala-sumur ke manifol adalah 3 inci sedangkan jarak kepala-sumur ke manifol berubah ubah tergantung posisi sumur. Posisi manifol menjadi satu lokasi dengan separator. Penjabaran detail untuk optimasi dengan menggunakan GA alurnya dapat dilihat pada Gambar 5a dan 5b. Pada gambar tersebut diperlihatkan alur detail optimasi yang dilakukan pada studi ini jika menggunakan GA saja dan pada Gambar 6a dan 6b jika menggunakan HGA (ED+GA+Proxy Model). Dari hasil simulasi dengan optimasi GA dan ED+GA+Proxy didapatkan hasil keluaran %"$= = ' $" " "$" metoda optimasi gabungan ED+GA+Proxy

! '1 ' biasanya disebut dengan 1 atau "& 1) lebih cepat mencapai kondisi optimum yang diperlihatkan dari dan plot kumulatif produksi yang tertinggi. Hasil yang cepat menuju ke optimum di pengaruhi oleh faktor tambahan ED dan 7 yang mengontrol GA sehingga faktor acak pada GA dapat dikurangi. Hasil perbandingan dari kedua metoda dapat dilihat pada Gambar 7 dan 8. V. Kesimpulan dan Saran Kesimpulan Dari hasil simulasi terlihat bahwa, dengan adanya .7 * dan Proxy Function/ Model, populasi baru akan lebih ter kontrol 9: sehingga kondisi optimum akan lebih cepat tercapai dengan menggunakan 1 atau "& 1 (ED+GA+Proxy). Pada populasi awal HGA sudah terjadi optimasi, sehingga populasi yang diambil lebih mewakili data yang dievaluasi dan lebih cepat mencapai kondisi optimumnya.

29

Saran Perlu diujicoba lebih lanjut dengan data lapangan dan dibandingkan hasil pemilihan posisi sumur dengan posisi sumur yang sudah ada, sehingga dapat di evaluasi apakah hasil optimasi tersebut sudah optimal. Perlu dilakukan & ' dengan software optimasi komersial yang terdapat di pasar sehingga dapat melihat sejauh mana " ' %"$

" Ucapan Terimakasih Penulis berterima kasih kepada para pihak yang telah membantu guna terlaksananya penelitian ini di antaranya adalah Research OGRINDO atas supportnya dan Schlumberger yang telah memberikan student license atas perangkat-lunak kepada Teknik Perminyakan ITB sehingga penulis dapat melakukan penelitian ini. Daftar Pustaka Abdulkarim, A., 2010. Overview of Saudi Aramco’s Intelligent Field Program, Society Petroleum Engineers-Paperno.129706. Q;XX`" $ "$! Furrial FIeld Asset applying Risk and Uncertainity Analysis for the Decision Making, Society Petroleum Engineers-Paper No. 94093 L$!%QQ;XX" $ " " " $ "$ ' Q "

Technologies to Optimally Manage Giant Fields, Society Petroleum Engineers-Paper No. 128469. L "QQ;XXL " ' " "$ an Integrated Asset Operations Modelling System, Society Petroleum Engineers-Paper No. 127893. ' ""$ ! ' "%' = & $"' " Q <" * % QC " * Q[;Q USA. "" Q;XX_" $ " "" Process Facility Models in a Single Simulation Tool, Society Petroleum Engineers-Paper No. 109260. Feroney S., 4 , 2009. Integrated Field DevelopmentImproved Field Planning and Operation Optimization, Society Petroleum EngineersPaper No. 14010.

30


Gendreau, M., July, 2002. An Introduction to Tabu Search, Département d´informatique et de recherche opérationnelle-Université de Montréal. Griess, 2006. Apllication of Global Optimization Techniques for Model Validation and Prediction Scenarios for a North African Oil Field, Society Petroleum Engineers-Paper No. 100193. Guyaguler, 2006. Integrated Optimization of Field Development, Planning, and Operation, Society Petroleum Engineers-Paper No. 102557. Howell, 2006. From Reservoir Through Process, From Today to Tomorrow-The Integrated As set Model, Society Petroleum Engineers-Paper No. 99469. Issaka, M.B., 2008. Real-Time Integrated Field Management at the Desktop, Society Petroleum Engineers-Paper No. 112071. Khan. K., 2006. Optimized Field Development # $%<" <" " % "¡ field-North of Monages, Venezuela, Paper No. IBP1786. Kumar, AJ., 2009. Implementation of Integrated Network Optimization Model for the Mumbai $! { { ' "Q Society Petroleum Engineers-Paper no. 123799.

&LQ;XXL "! "$ " and Automation Practices in the Upstream Business, Society Petroleum Engineers-Paper No.102701. *QQ;XX; ! L%{ " Optimization System Based on Integrated Reservoir and Facility Simulation, Society Petroleum Engineers-Paper No.77643. Madray, M., 2008. Integrated Field Modelling of the Miskar Field, Society Petroleum EngineersPaper No. 113873. Morales, 2010. Using Genetic Algorithm to Optimize

" "C " " * Q Society Petroleum Engineers-Paper No.130999 QC " ! $! ! #' Q 42003. Promoting Real-time Optimization of Hydrocarbon Producing Systems, Society Petroleum Engineers-Paper No. 83978. Purwar, S., 2010. A Method for Integrating Response Surface into Optimization Models with Real ' " # % "C

"$Q# % Petroleum Engineers-Paper No.129566. Rodriguez, R., 2007. Integration of Subsurface, Surface and Economic under Uncertainity in Orocual Field, Society Petroleum EngineersPaper No.107259. Sagli. JR., 2007 Improved Production and Process Optimization Through People, Technology, and Process, Society Petroleum Engineers-Paper No. 110655. #' Q8"Q;XX|" $ Studies Help Asset Teams Make Optimal Field Development Decisions, Society Petroleum Engineers-Paper No.110250. #

QQ;XX}"# " " 8 " ' " " * {]

"$Q Society Petroleum Engineers-Paper No. 125557. #! " ' Q;XX< " $ Production Modelling, Society Petroleum Engineers-Paper No. 128742. #' Q 4- 2002. A critical Overview of '' " "! " " " ' "Q Society Petroleum Engineers-Paper No.77703. Solis R. et al., 2004. Risk, Uncertainity and Optimization for Offshore Gas Asset Planning

" = Q# % "$ " { Paper No. 90177.


Gambar 1. Diagram evaluasi

Gambar 2. Bagan alir optimasi dengan menggunakan GA

Gambar 3. Bagan alir optimasi dengan menggunakan HGA (ED+GA+Proxy Model)

31

32


Gambar 4. Model reservoar

Gambar 5a. Populasi #1 dengan GA

Gambar 5b. Populasi #2 dengan GA


33

Gambar 6a. Populasi #1 dengan HGA menggunakan ED latin hypercube

Gambar 7. Fitness plot GA vs HGA (ED+GA+proxy model)

Gambar 6b. Populasi #2 dengan HGA (ED+GA+proxy model)

Gambar 8. GA vs HGA (ED+GA+proxy model)

34


Coal Bed Methane Reservoirs

di Sumatera Selatan Bambang Widarsono, Kosasih, Junita Trivianty Musu CLC# ? %*X}Q?Q" " Phone: +62217394422, Fax: +62217246150 (email: [email protected]) Abstract = !" L ! !" " Q " " Q* Q" =" {"%= " $"% {

"' "$ " "L * * Q "

" ' "=" L' '8 !* !"! ! "% $"% " '% "'' = ! ' " tend to give far different magnitudes when compared to laboratory results. After a series of re-evaluations and re-measurements on the laboratory results it was convinced that the problem does not lie with the laboratory results = !! ¡"* " " " !

" ' "" = " ' " = " ' " "! " show that only equation for ash contents gives accurate results. The other proximate analysis output data - i.e. " " Q* Q" ="{ " =

!! '"

"$

"'"! $ " ' ! " ! "$

" "! equations have produced analogous but different empirical equations to the original equations. These equations "% =%! Q" ! = " " ! $"% "! !* !

" % LQ$"% Q' "% Q

"

Abstrak *

& & 91:'

&-'

'& & - - 7 & '

' 1 4' ' - '' & &

1 & 9': /'' & + ! ' ' ' ' 4" !

&' &

&&

' & 4 '' ' ! ' '& - ' & + !&4*

' ''& ! - ' ' '' ' ! Z' [&4! & &

'

' / & -

' ' '

' ' '

' !

& ! & ' '

!

1

/ & & 4 $ ' 81- - 7 -' -

35

36


I. Introduction In the last two decades coal bed methane L! * ' " " " " every part of the world. In coal mining industry the contained methane gas in the coal seams were mostly regarded as hazard to the mining activities themselves. Various methane related accidents have occurred throughout the history of coal mining, often resulted in serious fatalities, especially in deep mining activities. "' " %L! % been long considered as a potential alternative to gas production from conventional gas reservoirs. This is true since coal beds are often encountered in petroleum wells, but too deep to be economically exploited through conventional deep mining operations. However, low gas price in the past " ! * L " " most coal beds compared to conventional gas reservoirs prevented its exploitation. Situation in today’s petroleum industry with awareness that world’s gas reserve has its " {! " '! !L has to be given a more appropriate attention. Based on technology available to petroleum and mining industries technology for drilling " ' "L!* = " * ' The various techniques that have been developed, including ones in formation evaluation, have exhibited their usefulness but improvements are indeed required. Recent execution on the ' '8 "" " "=" – South Sumatera, has shown the need. & "$! ! Q ' * ! * $ * % "$ " $ run, but it was doubted whether the common equations/models used in log interpretation are valid for the required reserve assessment. This was proved based on significant difference between results of proximate analysis - ash contents, moisture, volatile matter, and fixed carbon - from laboratory and the corresponding estimates using conventional equations. As fully acknowledged, error in such data will result in " " L " ! %=* ! = " presented in Widarsono et al (2009a and 2009b).

" !

" was produced and is considered as more reliable " * L * " =" "! '' Q! ! study are again to be presented with some scanning electron microscope images added. This addititional data has proved useful in providing " "

"$ " ! ' * coal seams. II. ‘Standard’ Models In a manner similar to conventional oil and gas reservoirs in their early era of exploitation, * "* L * out much using core samples and testing in the =% " "$ ! " coring and testing of all existing wells (especially L * Q! ! ' "

! "] " =% " %Q thought that means had to be established to utilize the most common and widely used method for reservoir evaluation, the wire-line logs. Mullen (1989) proposed a set of interpretation models for proximate analysis data derived from basin-wide database reported by Fassets and Hinds in their 1971 US Geological Survey Professional Paper 676. The original set of equations presented in Mullen (1988) is (1) (2) (3) (4)

U represent the data with ! " " Q ="Q Q * matter, and bulk density, respectively. The proximate analysis data is in weight fraction and bulk density in gr/cc. Although these equations were derived from a certain database they are subject to modification in order to accommodate ‘local effect’, as has been shown in Mullen (1989). The proximate analysis data estimated

$"% #!# L !" *

37

(Bambang Widarsono, Kosasih, Junita Trivianty Musu)

using Equations (1) through (4) enables the estimation of gas contents. There are some ! * = Q"$ !

Kim method (Kim, 1977) of (5)

(6)

(7)

Other fully empirical methods are average gas contents from Mullen (1989) (8)

(9)

(10)

The equations presented above are at present still being used in the interpretation of $ L ' "" calibration using tested cores are suggested but va r ious pr ac tices h av e sh o wn th at t hi s recommended practice is not always carried out. Instead, direct use without any corrective measure is often practiced. III. Case study: Rambutan Field $ "%Q ! =" ' project is situated in the southern part of the South Sumatera sedimentary basin in Sumatera (Figure 1). The basin, which development ! "] " =%! = "! Indo-Australian Plate underneath the Southeast " "$! {% Tertiary, contains sediments from Tertiary terrestrial to marine clastics with minor limestone

(Suwarna et al, 2003). The coal seams that serve as the focus, from which the methane gas is expected, are contained within the tertiary (Oligocene to Pleistocene) Muara Enim formation (Suwarna et al, 2003). The Muara Enim formation comprises approximately 900 m of paralic sandstones with interbedded coal seams, which typically form between 10% and 20% of total formation thickness. Referring to well log data the coal seams are dispersedly distributed throughout the formation. With regard to age the Muara Enim formation is sub-divided, from the oldest to the youngest, into M1, M2, M3, and M4 Subdivisions (Figure 2). From the four Sub-divisions, ; #={

* " ! Q ;Q LQ " (Figure 3) coal seams is the focus of the pilot project (Suwarna et al, 2003). The A1and A2 seams range between 6 and 15 meters in thickness. From studies from outcrops, the A1 seam tends to thin out in the ! "'! ! ! ; tends to be uniform with its 9-15 meters thickness !$!! ! L '' = the thickest with maximum thickness of 19 meter !"* $ * _ %Q most of the seam is present in a single body except some vertical separation in some parts in the "! " ! '' "% the thinnest seam in the M2 Sub-division with thickness from 7 to 11 meters. The seam tends to reach its maximum thickness in the western '! Maceral analysis previously conducted for the coals (from A seams) show that the maceral group is predominantly vitrinite that varies within 71% - 94%, with exinite and inertinite components of 1-7% and 3-10%, respectively. L " ! * " ] " samples from the A seams and the Enim seam (M4 Sub-division) of 0.50-0.52% and 0.380.47%, respectively, it is interpreted that the L' " = " "= $ " " thermogenic types, dominated by the thermogenic one. If the tested coal samples are taken as the representative for the coals in Rambutan project, ! * " ] " $$ !! *% $" ={= " {L ranks. Figures 4 and 5 present scanning electron

38

$"% #!# L !" * (Bambang Widarsono, Kosasih, Junita Trivianty Musu)

microscope (SEM) photos of samples representing " The coals have fairly developed cleat system with averaged cleat spacing of 27.5 cm and 8.2 cm for A and Enim seams, respectively, and cleat aperture of 0.1-0.2 cm. Despite fairly cleated (i.e. moderate permeability), the coals tend to exhibit very low porosity. The high volatile matter contents (39.3-43.4%) shown by the samples indicate characteristics of very low in situ methane co n ten ts. Ho wev er, o t he r characteristics of dull to dull-banded lithotype, vitrinite dominated maceral composition, low to

* " ] " Q contents, and low ash contents indicate moderate * L" " ! "! IV. Laboratory Works A total of 53 core samples were taken from the pilot project’s wells nos. 3 and 4. The sizes of full diameter cores are 3.5 inch and 2.5 inch " Q ' * % ' were retrieved under hydrostatic pressure from mud column in the wells. The samples were to undergo gas contents, proximate analysis, ultimate analysis, and bulk density measurements. As the core samples were retrieved from the wells, they were stored in canisters and isolated completely. Following the standard procedure, total volumes of released gas within the canisters during transportation (Q2) were measured upon arrival at the laboratory through the use of fast desorption method. Simultaneously, gas composition was measured using gas !$'! ' ! " * from the canister and tested for their bulk density using mercury displacement method. Next step was to put the samples into crusher. As the samples were being pulverized the released gas (Q3) was measured thoroughly. Using the recorded Q2 and Q3, the volumes of released gas during the timemeasured core retrieval (Q1) were estimated using fast desorption method. Figure 6 presents an example of Q1 estimation. Summation of Q1, Q2, and Q3 serves as the estimates of total gas contents (V) in the core samples. ! ' "% crushed (grain size < 150 m) some of the pow-

der was taken for proximate analysis. For proximate "% Q "" %' ' was put into furnace, and following the procedures described in ASTM D 3173 – 00, ASTM D 3174 – 00, and ASTM D 3175 – 01, moisture content (VM), ash content (Vash), and volatile matter (VVM) were measured. Proximate analysis measurements are made in air dry basis (adb), which results are then converted into as received (ar) condition imitating in situ condition in wells. All proximate analysis results are presented

" $! " =" '% a subtraction of the combined three parameters from 100%. Notice the difference shown by Mullen’s Equations (1) through (4), in which =" " * merely a product of the other three proximate analysis parameters. For log analysis purposes, it could arguably be considered irrelevant whether * =" ! ' of the other three parameters, but it is indeed volatile matter that actually ‘matters’ – and is measured in the laboratory – since it determines the tested coal’s rank. Apart from the above tests, ultimate analysis tests were also carried out to determine the coal’s elemental contents and calorie values. The resulting data is treated as supplemental information to this study. Main results of the overall laboratory measurements are presented in Table 1. V. Analyses of Data As in standard activities related to the "L "Q=% tests on samples in this study were performed under the most representative conditions. As ! "

" ! " ! "$ data was valid for comparison to its corresponding log analysis results. Discrepancies between the two data sources were expected but it was hoped that they would still be within an acceptable degree not to lead into a need to modify the existing equations. Figures 7 through 10 exhibit comparisons between observed (measured) and calculated ! " " Q ="Q " " Q and volatile matters data for all samples tested.


The comparisons have shown that only equation for ash contents (Equation 1) seems to work properly by yielding accurate Vash estimates. ' " ! ! ! ! * % $" "

$ " Q! ! data tends to differ much compared to observed data. The use of alternate bulk density from density log did not provide any improvements. ' " = " Q1+Q2+Q3) and calculated gas contents were also carried out. Figures 11 through 13 present ! ' "

Q "Q and Mavor et al equations, respectively. From the three comparisons, it is obvious that there is no acceptable degree of accuracy given by the three equations. At this point it is worth emphasizing that the validity of the observed gas contents is very much determined by the validity of the method for determining Q1, the fast desorption method. All comparisons for proximate analysis and gas contents data have proved that the ‘standard’ equations and methods are not applicable for the Rambutan coals. Modifications, even establishments of new equations, upon the ‘standard’ equations are apparently needed if ! $ "! =" "

" ! " L " in the project. VI. Suggested Models For the proximate analysis tests data, Mullen equations for moisture contents (Equation 2) and fixed carbon (Equation 3) " ! " !"

" For volatile matter, Equation 4 is certainly in no need for modification but the resulting estimates were later to be compared with observed VVM from laboratory. This additional source of comparison was later to prove useful in '*

"$ " "$

" Equations 2 and 3.

" " " ;" [ essentially carried out through a series of trials of alternative pairs of slopes and intercepts. Improvement in agreement between calculated data with observed ones is the prime objective with an acceptable degree of agreement between

39

calculated and observed VVM as the constraint. ! ' !

" are the adjustments of Equations 2 and 3 into

Figures 14 through 17 present the comparisons between observed and calculated data produced through the use of Equations 11 through 14. Obvious improvements are apparent when compared to comparisons presented in Figures 8 and 9. Figures 18 and 19 show similar comparison for the resulting volatile matter data obtained using Equation 4. Significant improvement is particularly shown by the comparison in Figure 19 when compared to the old comparison in Figure 10. This underlines the validity of Equations 12 and 14 for Rambu" The comparison in Figure 18 for A seam is indeed less encouraging for the data points are scattered far off the 45 degree line indicating a complete disagreement between calculated and observed VVM. This could be taken as an indication that validity of Equations 11 and 13 are less convincing than Equations 12 and 14. However, when it is considered that the " " !* $" through calibration using laboratory data, it can arguably be taken that the inconsistency is somewhat caused by non-linearity relationship between the four proximate analysis parameters in A seam coals. Regardless the VVM results, E qua t i ons 11 t hr ough 14 c a n s a f e ly b e considered more valid than Equations 2 and 3 for the Rambutan coals. For gas contents, similar approaches !

" "$

40


models/equations. This is especially true for Mullen and Mavor et al equations. For Mullen average gas contents equation, Equation 8 becomes.

to Equation 5. A more theoretical approach should be adopted even though the semi-empirical nature of the equation may make the effort very challenging. VII. Conclusions

and the Mavor et al equation becomes

With VM data in the ad resulted from the new Equations 13 and 14. Figures 20 and 21 present comparisons between the new calculated and measured (i.e. Q1+Q2+Q3) gas contents. Better agreement has been achieved when the two equations are used. For Modified Kim equation, the

" ' " `!$! _=* % !' " " " order variables and constants. From a quick look it is certainly unclear about how the sensitivity trials and adjustments should start from and are focused on. However, provided that the proximate analysis parameters are accepted while expressions of 0.96h and 0.14(1.8h/100) obviously represent pressure and temperature at depths, respectively, it is variables of ko and no that have to be given attention. Through a series of sensitivity trials the ko and no become

Figure 22 presents a comparison between the new calculated gas contents data and the measured data. The reasonably good agreement !! ' *

%!

equation, but with ko and no represented by Equations 17 and 18, for Rambutan coals, at least for the tested samples. It is well understood that adjustments on the ko and no equations are somewhat subjective and non-unique in nature. Other investigators may have attention and focus given to other aspects within and that are related

" "$' analysis and gas contents equations have been established. Despite their varied level of validity, the equations can nevertheless be regarded as more valid for the Rambutan coals. Any future analysis on well log data in Rambutan should use these equations. The use of ‘standard’ equations ! " " L$"% will yield inaccurate results. Modifications over the ‘standard’ equations have reinforced the conviction that different coals should be treated differently in L $ "% Q * " !$! $' "$ " ! '' " % = 8 (i.e. different equations for different groups of " ! " ! =" $" ={= " {L=%" ! ! !

" * coals belonging to those ranks, it is future and further applications that will prove. References ASTM D 3173-00, 2002. Standard Test Method for "! "% #' "

#" " "QXXL= & * Q " !! "Q}q;|{;}`}{<# ASTM D 3174-00, 2002. Standard Test Method ! "! "% #' " #" " "QXXL= & * Q " !! "Q}q;|{;}`}{<# ASTM D 3175-01, 2002. Standard Test Method for "! "% #' " #" " "QXXL= & * Q " !! "Q}q;|{;}`}{<# Fassett, J.E. & Hinds, J.S., 1971. Geology and Fuel Resources of The Fruitland Formation and Kirtland Shale of The San Juan Basin, New " <#C $ #* % Professional Paper 676, p76, 1971. QCQ}__ "$ !" " " L " = ' "&<# Bureau of Mines, Report of Investigations 8245, 22p.


*Q?Q Q?L" QQ}}X "* "' "=

Methane Wells. SPE Paper #90-101, presented ! " " " !" "$Q$% June 10-13. "Q?Q}|}= !" * " " $ "! ! "#"?"L " # %#' #18946, presented at the SPE Joint Rocky Mountain $ " = % * #%' " ! = "Q& "* { Q!{| "Q?Q}||$* " " & = !" %" " " of Geologists,pp 113- 124. # !Q& QQ}| ! & " "$! C " " Procedures and Results. U.S. Bureau of Mines Report of Investigations 8515.

41

#"Q}! Q;XX[* "L Exploration in South Sumatera, Vol. 1A. (in Bahasa Indonesia), unpublished report, Ministery of Energy and Mineral Resources – The Republic of Indonesia. Widarsono, B., Sartadiredja, K., & Musu, J.M., ;XX} "#!# * ! < ¡"* " "$"% " "$= !" &#' ¤;[}q_Q

"$Q' " ! " C " " ! = "Q?{ Indonesia, 4 - 6 August. Widarsono, B., Sartadiredja, K., & Widjayanto, B.A., 2009b. Establishment of More Reliable Equations "= !" "% &C## " " = " to Petroleum Science & Technology, Vol. 32, No. 2, August, pp: 103 – 113.

42


= =% " " '


Figure 2. General stratigraphy of the area of study [from Suwarna et al, 2003]

43

44


Figure 4. SEM photos of sample S3-09 of seam A showing $" ""%" " $!Q'' within vitrinite. Vitrinite is the most abundant maceral content, presents and makes up the groundmass where minor amount of resinite debris (left, green arrows) are

' &§ * %$" content due to presence of the clay minerals.

Sample Depth

: SP-04 L¤[# : 2958.00 – 2958.30 ft

Figure 3. Stratigraphic column with thicness of the seams based on gamma ray and density logs [from Suwarna et al, 2003]

Sample Depth

: S3-09 L¤[# : 1724.10 – 1724.20 ft

Figure 5. Vitrinite occurs in the form of unstructured vitrinite (collinite) (left, brown arrow) and telinite (structured vitrinite) clearly recognized cell walls of more or less

" '" = &§ ! ! $! (organic matter content) compared to seam A samples.


45

$ |' "= " " moisture content (original equation)

$ "' "C ¨ & "$ *& "$ "$

$ _' "= " " ash content (original equation)

$ }' "= " " =" $ " "

$ X' "= " " calculated volatile matter (original equation)

46


$ ' "= " " gas content (Mod Kim, original equation)

$ q' "= " " ="

"

$ ;' "= " " gas content (Mullen, original equation)

$ `' "= " " ="

"

$ [' "= " " gas content (Mavor et al, original equation)

$ ' "= " " " "

"


47

$ _' "= " " " "

"

$ ;X' "= " " $ " " "Q

"

$ |' "= " " *

"

$ ;' "= " " gas content (Mavor et alQ

"

$ }' "= " " *

"

$ ;;' "= " " $ " " Q

"

48


Dewatering Facilities for CBM Resource Appraisal, Lesson Learned from Sanga-Sanga Coalbed Methane Field, East Kalimantan Fasilitas “Dewatering” untuk Apraisal Gas Metana Batubara, Pelajaran dari Lapangan CBM Sanga-sanga Kalimantan Timur Agam Munawar(1), Donny Hendromurti(1), Rina Dewi P(1), Giovanni Foglio(2), BrianSchupp(3) (1)" " Q q|th - 49th Floors Jl. Jend. Gatot Subroto No. 42, Jakarta 12710, Indonesia PO BOX 2828, JKT 10028 Phone: +62215236686 (2)ENI Atrium Mulia Building Jl. H.R. Rasuna Said Kav. B 10-11, Jakarta 12910, Indonesia (3)BP Perkantoran Hijau Arkadia Tower E, 5th Floor Jl. T.B. Simatupang Kav. 88 Jakarta 12520, Indonesia

Abstract ! '' = !" L! "%" * ! "' !"* " " " $ ' ""LQ$ ' " % " " "$' !

"$' " % '"%% QL'' '8 "$ " $ ' "= ;XX& = ;X;QL installed and operated nine dewatering facilities in the Sanga-Sanga Field. High uncertainties in the early $ ! ' "'! L* "

" "%! ''! "* "$! $!' "* = %L " ' "$! "$' "" "$ !% "Q! " !* = " Q = " "

' " '' %"! Q! $"'! '!%* ] = % "%! '' ! '' to discuss major parameters required to be taken into account in the design of dewatering facilities and to show !* ' "'= "$! " "$' "! #"${#"$ Q Kalimantan, Indonesia. % "$Q Q "$Q' "QL

Abstrak 1 & 91:'& '' '/' & '

' ' '

' 4 1-'

& 9+ :4 ' + '

& ! 4" -' 1&' '

'

'4* & #'&;UDU

*&;UD;-K# ' & 9B:

!

4%

' '

+ '

' K# '& 9Q:

'

9 :4 ' &

' /' 9 &: ' ' &/

4 $ ' & ' ' - ' & ' ''

'

'& 4*

' -

& 6'& / 4 ' ' '' ! 0 '

49

50


/' &

K#

/

!

+

!

-$ %- $ $ 8 - '

- ' -'-1

I. Introduction L 8 " * " ' equally between BP and ENI. It is currently executing an exploration program appraising ! %= !" L play in East Kalimantan, Indonesia. With 1,747 ;# $ Q! L# " %* ' ! "$"* " " "#! "$"#! $ "! # ' " and shared usage of resources, equipment, materials, and facilities. Most of the time, the # ' ! $! "Q Kalimantan is on the equator . Geologically, the # " =% "* " "Q which varies the sedimentary facies. The surface " %! " $ ; "%Q L' " "$ Q! ! consist of two multi-well pilots and two single well pilots. The distance from the most southern well to the most northern well is about 33 km. II. Methods <" "* " " Q L need to dewater the coals to produce gas. Water must be removed from the coal to lower the seam pressure below the critical desorption pressure (Mavor et al., 1990). The time to dewater the coals varies, from days to years. This condition L' ' ! $!" " $" "$ ! !! * = conditions. With time, water production should start, so water handling may be managed more easily in the future. " " ! = " " "$ one of the top gas producers in East Kalimantan since the 1970s. Gas processing and water disposal facilities are well established. This condition ! ' L '' '$"QL " than conventional wells do. During the

"$ $ Q L ' =% " % "%QL ! " % Q which are driven by an electric motor. In fact, ! " ' " $ Q! ! do not require electrical power. L{' % "

] "* " "

' "%""* " "% produced water is re-injected into a disposal * # ! ' ] ! " L{' quality has uncertainties, which mean it may contain something that could harm the treatment ' L "$ L water separately from its existing facilities. Produced Water Handling Two big uncertainties exist related to L{' * " % There are wide variations of water production rates from coals in any basin. As a whole, ratio water and gas about 0.31 barrels of water ' ' QXXX !" Q " Q 1993). A Report from 420 wells in the Warrior Basin had initial water production rates of 17 to 1,175 BWPD (Pashin et al., 1990). This is "$ "

"!L{' ! widely range, even at same field. The water production rate may not always high. Southern San Juan basin wells have little amounts of water production (Kaiser and Swartz, 1989). Shallow coals on the eastern edge of the Powder River basin exhibit minimal water production (Seidle, 1991). Furthermore, none of that study covered any water data from basin in Indonesia. ! " " " % "$ ' Q;; ! * ! ! " ""$ XXqQ}XX$&* et al., 1993) Four techniques are possible to dispose of produced coalbed waters: (1) well injection,

& "$ L '' Q " "#"${#"$= !" Q ""

51

(Agam Munawar, Donny Hendromurti, Rina Dewi P)

(2) discharge into surface streams, (3) land '' "Q " q ' " L!

' "8 " as the way to manage big uncertainties in water volume and quality. After producing for more !"[X% Q! "* " "#! many idle wells and depleted reservoirs, which ! % " = "8 " ' " L %' " Q sands). These solids can plug the meters and { " ! ] " Q "$ =

"$! ! ] "

" * ' ' "$L wells has become a study in choosing and modifying a method to give the best procedure to mitigate this problem (Holditch, 1990). To anticipate those solids and their problems, a pond with a geo membrane was constructed (Figure 4). To manage uncertainty in producing the water volume and to accommodate rainwater, the pond was oversized. For safety and security, a fence also encloses the pond. In brief, the water ' ! L ] ! pond, settles down the solids, and is processed, Q " "8 ' "* " " sand. Two injection facilities (Figure 5) were built in two different areas: ! $" "8 " ! pacity of 16,000 bwpd, which is dedicated for ! L ! $" Northern area ! L "8 " !' % of 7,000 bwpd, which is dedicated for the L "! #! " For contingency, these two injection " " conventional water treatments. By having this synergy and cooperation, both parties have the = " Q ! " * " !* "$ problems, they can always use the other team’s facilities. Gas Handling L "" "* " " $ ] " ! "$ conventional gas flowlines may have three different pressure systems:

%' ' ;X{XX' $ ] "$' ' ' XX{;XX' $ ] "$'

' ' ;XX{[`X' $ ] "$'

L ! = "" * % low-pressure system to maximize gas recovery. However, during the life of the well, conventional wells may be recompleted to new production " Q ! ! " " ] " ' !"$ " ' ! "

"Q L " ] " ! "" to conventional wells with no potential for

' ' " " not frequently use a burn gas to burn pits. For % Q ! ' $" $ emissions and redirect all gas molecules from the reservoir to the pipeline. In addition, as the reservoir pressure declines, the population of wellhead compressors has increased dramatically in the last few years to improve reserve recovery. Gas and Water Measurement Measurement is critical to understand pump performance, feed data for appraisal decisions, and to prevent any commercial issues, ' %$ ' "L wells have single line measurement prior entering the production network. This measurement is located on the well site. Gas measurement uses " ! " = " ]' ! = " " in every well to provide the operator with direct $ " " " &

'% (Figure 6). Electrical Power Supply ! " %! ! Q L " !* ' $ " " ''%' ! system (downhole pump) at the wellsite. It is

'"! " '* a clean, reliable, and stable power supply for smooth operations. A local power generator (genset) was the only option that could satisfy power requirements, cost, and time delivery.

52


! * Q L three different sizes of electric motor: _` qX XX The genset 135 KVA diesel engine (Figure 7) " " !L ' ! motors. To reduce fuel truck trip frequency, all gensets have an external fuel tank, which is enough for 7 days operation. III. Operational Problems " " * " " Frequent Dismantling/Re-installing of the & "$ ! ' " '! Q L wells may need to have frequent well interventions for several reasons: ' "$= " "! '' ' "$ "! '' !'' ! * "]

"$ "$ " "$"" = " ! "$ "! '' Almost all well interventions require a workover rig. Before the rig arrives, some facilities will need to be dismantled for safety and operational reasons. Time for dismantling facilities and for re-installing depends on construction crew availability, equipment availability (boom truck, excavator, crane, etc), and the complexity of the job. To minimize job complexity, and crew and equipment availability, L !"$ ' " % (Figure 8) and adopted a portable design pipe support (Figure 9). This new layout places all facilities outside the radius of any rig activity. ! Q"%] " ' be dismantled prior to the rig workover commencing. Gas Present in Water Line L !* ' ] " with meter spools: one for gas and one for water. When the pump is sumped below the bottom ' "Q! L $"" " " acts like a large vertical separator. By gravity,

gases go up through the casing annulus to the surface, while water falls to bottom and enter ! '' " Q ! " ] !$! = "$ to surface. Theoretically, the downhole pump " " " gas. During the pumped off condition, which is ! " ! "" ] * " ! ''

" Q! ""] = $ %" ! risk of gas entering the pump becomes high.

"%Q ! " ! ] "$ = ! pressure is still high and near initial conditions, the water may contain dissolved gas, which will break out of the solution at low pressure points in pump and / or tubing. The condition is not only bad for the downhole pump but also for water metering as it raises pressure losses in ] " * "$ "! ' will help to reduce this risk, but the cheapest ' " % = "$ ! ] * certain level above the intake pump by using '' ' " ! $! ] * may result in higher annular pressures, which could affect gas production, so this needs to be considered during the well design. A bucket test needs to be performed frequently to obtain ] " " $ in the water line. If gas enters the water line, the line will have to be bled off to reduce the pressure, as the gas will become trapped in high ' ! ] " ! '$'!% ! East Kalimantan is rolling hills, so this is a real possibility. Frequent venting can create periods of pressure cycling, which can also affect pump performance. Electrical Power Reliability Impact " L ! = " "$ " ! L '" " _| Q "= ;XX ' =% an electric motor, which has power from a local genset at 135 KVA. The genset requires regular maintenance every 500 hours, which means ! " = ! " " month. During the shutdown, the fluid level between the tubing and casing can equalize, causing a rapid pressure differential across the elastomer. Usually after shut down, once the " "Q = Q ! ! "" ''! " "! condition can lead to failure sequences, such as

& "$ L '' Q " "#"${#"$= !" Q ""

53

(Agam Munawar, Donny Hendromurti, Rina Dewi P)

stripped rods and chunked elastomers. Two possible solutions have been prepared to overcome this problem. First, provide a synchronize panel, which allows the working genset to be switched with a spare genset without any interruption. # " Q " " %' Q which has no relevance to torque and elastomers. ! " ' "Q' " ' $ [ !

* ' "= "" Back Pressure from Conventional Wells. Having a connection with conventional L Q especially after conventional wells change their production zone. New production zones typically have higher pressure than the previous ones. This condition impacts other wells connected

" ! " L $ " " ! been made for Multi Well Pilot #2 to mitigate this problem. This trunkline is used exclusively =%L ! " ' "! trunkline, it is connected to very a low-pressure header line on a gathering station. Problems Associated with High Rainfall During 2010–2012, the rainy season for East Kalimantan was substantial. High rainfall caused slippery roads, slowing down rig movement, operator mobilization, and sometimes disturbing facility installation. It also increased the risk of landslides. The pond water level also quickly increased from the rainwater, "$ ! $! "8 " L has installed a memory recorder on each well to anticipate any data acquisition problem due = ] ' ! ] Q" " !'" ! pressure data. IV. Results L! %"$ dewater ten wells with good surface facility = %# " ! L '" production in October 2010, about 764,213 barrels of cumulative water have been produced and

reinjected into the disposal reservoir. A new facilities design reduced the facility reinstallation time from 5 days to 1 day. The genset routine and unscheduled maintenance result in the most down time. Two possible solutions were proposed to mitigate the impact from this condition. First, provide a synchronize panel, which allows the working genset to be switched to a spare genset !"% " ' " "%Q! synchronized panel is being installed and tested

"! # " Q " " %' !! " *" " ! " ' "Q ! " Q ! ! = " " " demonstrated its ability to handle problems associated with power interruption. The challenges

" % $"L "! simplify and standardize facilities, to optimize the existing ones for less installation time and cost reduction. The early appraisal stage has " $" L well facilities. V. Conclusions The following are the key points from this paper: L * L'' "* " " "$ !{ ] tests may give unreliable results. Only with " ] Q! ! ' years, can the necessary data be collected to analyze the reservoir’s performance, in ' ' = %! QL facilities need to be designed to promote safe and reliable operations while collecting and measuring accurate data to be used in reservoir performance analysis. ; L % $" "! % $ " appraisal program must deal with many operating parameters uncertainties like water volume, water quality, and produced solids. The facilities need to be designed to be as ] = ' = "$ ! " " [ ! L $" "! "* " "L" ! "$ conventional facility’s extra capacity without disturbing conventional operation. When some

54


parameters are still uncertain, and at worst can harm the conventional process (like produced water treatment), it would be better to handle those uncertainties separately (with new facilities). q "QL " !" conventional wells do. During the dewatering $ QL ' =%" % Q! ! electrical power. A local power generator (genset) must be installed when an electrical power network is unavailable in the well area. ! = %$ " "] " the runlife of the downhole pumping system. Generally, downhole pumping systems are vulnerable to power interruption, which results in unplanned shutdowns. Once power interruption happens, the pumps may be

{'Q! !% The presence of a spare genset with a synchronized panel on location can improve electrical power reliability. Failures associated with a downhole pump would be more costly than having a spare genset with a synchronized panel option. 5. Frequent well intervention that requires dismantling / re-installing must be anticipated when designing the facilities. The major equipment must be placed at the outside range of the workover rig working area, so that equipment does not have to be moved. It is necessary to coordinate and work with wellwork teams as the facility design and layout must accommodate frequent rig activity. A multi-disciplinary team, including individuals from production engineering, wellwork, and oper ations teams m u st b e p rese nt i n preplanning, design, and layout meetings. 6. After production data show a consistent trend, some uncertain parameters (water volume and quality) may be unlocked. The following challenges in well design are how to simplify and standardize facilities and to optimize existing ones to design more efficient installations with inherently safer designs. Acknowledgements

! "LQ" " Q

LQQ" $%Q?'"L Q SKK MIGAS and MIGAS for support and permission to publish this paper. Particular $" " ! "L 8 "$ Q! L Ahmed, Rimbo, Maluddin Silitonga, Rachmat Samsoeri, Sudarwadji, Heri Purnomo, Wiarto, " ! " ! "$ " "$" " " L

"Q"Q Achmad Yamil, Iwan WP, and others). References &* QQ# ' "QQ " QQ Q ?Q" "Q&CQ}}[= !" Produced Water Management Strategies in the Black Warrior Basin of Alabama. Proc., " " "= !" #%' Q Vol. I, Birmingham, Alabama, May, p.317-338.

!Q#Q}}X' " ! " seam Reservoirs. Society of Petroleum Engineers, SPE 20670. Kaiser, W.R. and Swartz, T.E., 1989. Fruitland Formation Hydrology and Producibility of = !" "! #"?"L "Q " Q= !" Symposium, Tuscaloosa, Alabama, April, p. 87. " QQ}}[= !" Water Treatment and Disposal Options. Quarterly * !" # !"$%Q December, Volume 11, No. 2, p. 6-17 . *Q?Q "QLQQ?Q}}X "" * "#' " Isotherm Data. Society of Petroleum Engineers, SPE 20728. ! "Q?Q QQQ " "QLQ!" Q R.V., Bolin, D.E., Hamilton, R.P., and Mink, Q}}XC $ * " " = !" Resources, Part II, Black Warrior basin, Annual report. Geological Survey of Alabama, p. 130. # Q?Q}}"${ C & * = % & = # % Engineers, SPE 21488.

& "$ L '' Q " "#"${#"$= !" Q "" (Agam Munawar, Donny Hendromurti, Rina Dewi P)

$ [" *

$ LQ#"${#"$L# (yellow block)

$ qL' '"

Figure 2. Topography at southern part of Sanga-Sanga L#

$ `L "8 "

55

56


Figure 6. Data recorder on well site

Figure 8. New design implementation - applying some distance to major equipments

$ _$ " " %

Figure 9. New design implementation - portable pipe support

UCAPAN TERIMA KASIH Ucapan terima kasih kepada para Mitra Bestari yang telah mengevaluasi, mereview dan memberikan saran perbaikan tulisan-tulisan yang dimuat di majalah Jurnal Teknologi Minyak dan Gas Bumi (JTMGB) edisi penerbitan Volume 4 Nomor 1, April 2013.

1. Prof. Dr. Ir. Septoratno Siregar 2. Prof. Dr. Ir. Pudjo Sukarno 3. Prof. Dr. Ir. Doddy Abdassah 4. Dr. Ir. RS Trijana Kartoatmodjo 5. Dr. Ir. Bambang Widarsono

INDEKS A

+' 13,14, 16, 19

QQ 13, 14, 16, 19

QQ 14

7 35

49 B & '

13, 14, 16, 19 C 6+ 1, 2, 3, 5, 6, 8, 11 CQ;QqQ}QXQ L[`Q[Q[_Q[|Q[}QqXQqQqqQq}Q`XQ`Q 52, 53, 54, 55 D dewatering 49, 50, 54 design 49, 50, 52, 53, 54, 56, E EOR 1, 2, 3, 11, 12 .7 * 25, 26, 27, 29 electricity 49, 51 F QQ 13, 14, 17, 18, 19, 22 Fasilitas Permukaan 49, 50 G 1 25, 26, 28, 30 GMB 35, 49, 50 H 13, 14, 17, 18, 19

I Injeksi Air 50 K Kelistrikan 50 L log analysis models 35 M 35 model log analisis 35

[` metering 49, 52 O Optimasi Produksi 25, 28 # & 49, 52 P * 25, 26 #Q 26, 30 produksi 49, 50 49, 50, 51, 52, 53, 54 Pengurasan 49, 50 pengukuran 50 Permasalahan operasi 50 pengangkatan bantuan 50 perancangan 50 S & 1, 2, 3, 4, 5, 6, 8, 11 SWAG 1, 2, 4, 5, 9, 10, 11 49 W + / 49, 55

JURNAL TEKNOLOGI MINYAK DAN GAS BUMI PEDOMAN PENULISAN

ISI DAN KRITERIA UMUM !! ! "8"%

= !"'=

?" "$ "% "C L (JTMGB) dapat berupa artikel hasil penelitian atau artikel ulas balik / tinjauan (review) tentang minyak dan gas bumi, baik sains maupun terapan. Naskah belum pernah dipublikasikan atau tidak sedang diajukan pada majalah / jurnal lain. Naskah ditulis dalam bahasa Indonesia atau bahasa Inggris sesuai kaidah masing-masing bahasa yang digunakan. Naskah harus selalu dilengkapi dengan Abstrak dalam Bahasa Indonesia dan Abstract dalam Bahasa Inggris. Naskah yang isi dan formatnya tidak sesuai dengan pedoman penulisan JTMGB akan dikembalikan ke penulis oleh redaksi untuk diperbaiki. FORMAT Umum. Seluruh bagian dari naskah termasuk judul abstrak, judul tabel dan gambar, catatan kaki, dan daftar acuan diketik satu setengah spasi pada ! dan ! dalam kertas HVS ukuran A4. Pengetikan dilakukan dengan menggunakan huruf (font) Times New Roman berukuran 12 point. Setiap halaman diberi nomor secara berurutan termasuk halaman gambar dan tabel. Hasil penelitian atau ulas balik/tinjauan ditulis minimum 5 halaman dan maksimum sebanyak 15 halaman, di luar gambar dan tabel. Selanjutnya susunan naskah dibuat sebagai berikut: Judul. Pada halaman judul tuliskan judul, nama setiap penulis, nama dan alamat institusi masing-masing penulis, dan catatan kaki, yang berisikan terhadap siapa korespondensi harus ditujukan termasuk nomor telepon dan faks serta alamat e-mail jika ada. Abstrak. Abstrak / & ditulis dalam dua bahasa yaitu bahasa Indonesia dan bahasa Inggris. Abstrak berisi ringkasan pokok bahasan lengkap dari keseluruhan naskah tanpa harus memberikan keterangan terlalu terperinci dari setiap bab. Abstrak tulisan bahasa Indonesia paling banyak terdiri dari 250 kata, sedangkan tulisan dengan bahasa Inggris maksimal 200 kata. Kata kunci / '+ ditulis di bawah abstrak / abstract dan terdiri atas tiga hingga lima kata. Pendahuluan. Bab ini harus memberikan latar belakang yang mencukupi sehingga pembaca dapat memahami dan dapat mengevaluasi hasil yang dicapai dari penelitian yang dilaksanakan tanpa harus membaca sendiri publikasi-publikasi sebelumnya, yang berhubungan dengan topik yang bersangkutan. Permasalahan. Bab ini menjelaskan permasalahan yang akan dilakukan penelitian ataupun kajian. Metodologi. Berisi materi yang membahas metodologi yang dipergunakan dalam menyelesaikan permasalahan melalui penelitan atau kajian. Hasil dan Analisis. Hanya berisi hasil-hasil penelitian baik yang disajikan dengan tulisan, tabel, maupun gambar. Hindarkan ' "$$""$ = = !"= '

8 " "$" " "$L ' "$$""Q 8 " yang benar-benar mewakili hasil penemuan. Beri nomor gambar dan tabel secara berurutan. Semua gambar dan tabel yang disajikan harus diacu dalam tulisan. Pembahasan atau Diskusi. Berisi interpretasi dari hasil penelitian yang diperoleh dan pembahasan yang dikaitkan dengan hasil-hasil yang pernah dilaporkan. Kesimpulan dan Saran. Berisi kesimpulan dan saran dari isi yang dikandung dalam tulisan. Kesimpulan atau saran tidak boleh diberi penomoran. Ucapan Terima Kasih. Bila diperlukan dapat digunakan untuk menyebutkan sumber dana penelitian dan untuk memberikan penghargaan kepada beberapa institusi atau orang yang membantu dalam pelaksanaan penelitian dan atau penulisan laporan.

JURNAL TEKNOLOGI MINYAK DAN GAS BUMI PEDOMAN PENULISAN DAFTAR PUSTAKA

Acuan. Acuan ditulis dan disusun menurut abjad. Beberapa contoh penulisan sumber acuan: Jurnal Hurst, W., 1934. Unsteady Flow of Fluids in Oil Reservoirs. Physics (Jan. 1934) 5, 20. Buku Abramowitz, M and Stegun, I.A., 1972. Handbook of Mathematical Functions. Dover Publications, Inc., New York. Bab dalam Buku Q?Q}|q!% $ '!$% = ]& Q? ! Q? (eds), Developments and Applications of Geomorphology, Springer-Verlag, Berlin, h.268-317. Abstrak L= QQL $ $$ QLQL " QQ* " QQ " QQ*QQC " QQC $ QQ Iaccarino, S., Innocenti, F., Marinelli, G., Scotti, A., Slejko, D., Sudradjat, A., dan Villa, A., 1983. Magmatic evolution and structural meaning of the island of Sumbawa, Indonesia-Tambora volcano, island of Sumbawa, Indonesia. Abstract 18th IUGG I, Symposium 01, h.48-49. Peta # " 8"QQ#"QC Q#Q " "QQ}}C $ =="$Q# Pusat Penelitian dan Pengembangan Geologi, Bandung. Prosiding Marhaendrajana, T. and Blasingame, T.A., 1997. Rigorous and Semi-Rigorous Approaches for the Evaluation of Average Reservoir Pressure from Pressure Transient Tests. paper SPE 38725 presented at the SPE "" !" " " " ! = "Q#""" Q`©| Skripsi / Tesis / Disertasi Marhaendrajana, T., 2000. Modeling and Analysis of Flow Behavior in Single and Multiwell Bound ed * !&

"Q <" * %Q $ # "Q§ Informasi dari Internet " QQ;XX# "" " * = " ' "$ * "$' www.boston.com/news/local/articles/2006/01/26/sri_lankans_tsunami_drive_blossoms/[26Jan 2006] Software #XX QC ¨ * !"$ Q== "$ "Q<Q}}_ ! '"$ "

"$' "$"$=' $ "$=' $

"%" =$ $= " $ %"$= "$"$

' " ' ! $ «8'$ "$" " " q " " [XX ' Q &«Q Q «Q $C= "=

"

bagian akhir naskah masing-masing pada halaman terpisah. Gambar dan tabel dari publikasi sebelumnya dapat dicantumkan bila mendapat persetujuan dari penulisnya. PENGIRIMAN "

" "$ " '" ! = ""%

'

&%"$ !

'" "$"'$ &

""' " "" " !" dikembalikan untuk diperbaiki jika persyaratan ini tidak dipenuhi. Naskah agar dikirimkan kepada: Redaksi Jurnal Teknologi Minyak dan Gas Bumi

"$ Jln. Jend. Gatot Subroto Kav. 32-34 Jakarta 12950 – Indonesia Pengiriman naskah harus disertai dengan surat resmi dari penulis penanggung jawab/korespondensi (corresponding author) yang harus berisikan dengan jelas nama penulis korespondensi, alamat lengkap untuk surat-menyurat, nomor telepon dan faks, serta alamat e-mail dan telepon genggam jika memiliki. Penulis korespondensi bertanggung jawab atas isi naskah dan legalitas pengiriman naskah yang bersangkutan. Naskah juga sudah harus diketahui dan disetujui oleh salah satu penulis dan atau seluruh anggota penulis dengan pernyataan secara tertulis.

Jurnal Teknologi Minyak dan Gas Bumi

Recommend Documents