PDF Archive

Easily share your PDF documents with your contacts, on the Web and Social Networks.

Share a file Manage my documents Convert Recover PDF Search Help Contact



OF1 2011 .pdf


Original filename: OF1_2011.pdf
Title: Microsoft Word - OF1_2011.docx
Author: Austin Holland

This PDF 1.4 document has been generated by Microsoft Word / Mac OS X 10.6.8 Quartz PDFContext, and has been sent on pdf-archive.com on 10/03/2014 at 18:44, from IP address 46.64.x.x. The current document download page has been viewed 1134 times.
File size: 3.8 MB (31 pages).
Privacy: public file




Download original PDF file









Document preview


 
 
 
 
 
 
 

Examination  of  Possibly  Induced  Seismicity  from  
Hydraulic  Fracturing  in  the  Eola  Field,  Garvin  
County,  Oklahoma  
 
 
Austin  Holland  
 

 

 
 
 
Oklahoma  Geological  Survey    
Open-­‐File  Report  
OF1-­‐2011  
 

 

 

 

 

 

OKLAHOMA  GEOLOGICAL  SURVEY  
Open-­‐file  Report  Disclaimer  
 
This  Open-­‐file  Report  is  intended  to  make  the  results  of  research  
available  at  the  earliest  possible  date  and  not  intended  to  represent  the  
final  or  formal  publication.    The  report  is  an  unedited  copy  prepared  by  
the  author.  

 

 

 

 

 

 

Examination  of  Possibly  Induced  Seismicity  from  
Hydraulic  Fracturing  in  the  Eola  Field,  Garvin  
County,  Oklahoma  
 
 
Austin  A.  Holland  
Oklahoma  Geological  Survey  
Sarkeys  Energy  Center  
100  East  Boyd  St.,  Rm.  N-­‐131  
Norman,  Oklahoma  73019-­‐0628  
 
 
 
 
 
 
 
 
 
 
 
 
 
 
August  2011  
 
Oklahoma  Geological  Survey    
Open-­‐File  Report  
OF1-­‐2011

 

 

 

 
Summary  
 
On  January  18,  2011,  The  Oklahoma  Geological  Survey  (OGS)  received  a  phone  call  
from  a  resident  living  south  of  Elmore  City,  in  Garvin  County,  Oklahoma,  that  
reported  feeling  several  earthquakes  throughout  the  night.    The  reporting  local  
resident  had  also  offered  that  there  was  an  active  hydraulic  fracturing  project  
occurring  nearby.    Upon  examination  there  were  nearly  50  earthquakes,  which  
occurred  during  that  time.    After  analyzing  the  data  there  were  43  earthquakes  large  
enough  to  be  located,  which  from  the  character  of  the  seismic  recordings  indicate  
that  they  are  both  shallow  and  unique.    The  earthquakes  range  in  magnitude  from  
1.0  to  2.8  Md  and  the  majority  of  earthquakes  occurred  within  about  24  hours  of  the  
first  earthquake.    Careful  attention  and  significant  effort  was  put  into  obtaining  the  
most  accurate  locations  possible  and  gaining  a  reasonable  estimate  in  the  error  in  
locations.    The  nearest  seismic  station  is  35  km  away  from  where  the  earthquakes  
occurred.    Formal  errors  in  location  are  on  the  order  100-­‐500  m  horizontally  and  
about  twice  that  for  depth.    Examination  of  different  velocity  models  would  suggest  
that  the  uncertainties  in  earthquake  locations  should  be  about  twice  the  formal  
uncertainties.  The  majority  of  earthquakes  appear  to  have  occurred  within  about  3.5  
km  of  the  well  located  in  the  Eola  Field  of  southern  Garvin  County.    The  Eola  Field  
has  many  structures,  which  may  provide  conduits  for  fluid  flow  at  depth.    The  well  is  
Picket  Unit  B  well  4-­‐18,  and  about  seven  hours  after  the  first  and  deepest  hydraulic  
fracturing  stage  started  the  earthquakes  began  occurring.    It  was  possible  to  model  
95%  of  the  earthquakes  in  this  sequence  using  a  simple  pore  pressure  diffusion  
model  with  a  permeability  of  about  250  mD  (milliDarcies).    While  this  permeability  
may  be  high  it  is  less  than  those  reported  for  highly  fractured  rock.    The  strong  
correlation  in  time  and  space  as  well  as  a  reasonable  fit  to  a  physical  model  suggest  
that  there  is  a  possibility  these  earthquakes  were  induced  by  hydraulic-­‐fracturing.    
However,  the  uncertainties  in  the  data  make  it  impossible  to  say  with  a  high  degree  
of  certainty  whether  or  not  these  earthquakes  were  triggered  by  natural  means  or  
by  the  nearby  hydraulic-­‐fracturing  operation.  
 
Introduction  
 
On  January  18th,  2011,  a  resident  living  in  south-­‐central  Oklahoma  (Garvin  County),  
living  south  of  Elmore  City  contacted  the  Oklahoma  Geological  Survey  (OGS)  to  
report  feeling  several  earthquakes  throughout  the  night  with  the  first  occurring  at  
approximately  6:10  PM  CST  Jan.  17th  and  another  large  one  at  about  2:50  AM  CST  
Jan.  18th.  Upon  examination  there  were  in  fact  earthquakes  in  the  area.    The  resident  
also  reported  that  there  was  an  active  hydraulic  fracturing  project  being  conducted  
in  a  nearby  well.    Examination  of  the  available  seismic  data,  including  EarthScope  
USArray  stations  in  the  region,  quickly  confirmed  that  earthquakes  were  occurring  
in  the  area.    At  this  point  the  OGS  contacted  the  Regional  Manager  for  the  Oklahoma  
Corporation  Commission,  who  indicated  that  there  was  indeed  fracturing  occurring  
at  the  Picket  Unit  B  Well  4-­‐18.      Our  analysis  showed  that  shortly  after  hydraulic  
fracturing  began  small  earthquakes  started  occurring,  and  more  than  50  were  
identified,  of  which  43  were  large  enough  to  be  located.    Most  of  these  earthquakes  
 

1    

 

 
occurred  within  a  24-­‐hour  period  after  hydraulic  fracturing  operations  had  ceased.    
There  have  been  previous  cases  where  seismologists  have  suggested  a  link  between  
hydraulic  fracturing  and  earthquakes,  but  data  was  limited,  so  drawing  a  definitive  
conclusions  was  not  possible  for  these  cases.    The  first  case  occurred  in  June1978  in  
Carter  and  Love  Counties,  just  south  of  Garvin  County,  with  70  earthquakes  in  6.2  
hours.    The  second  case  occurred  in  Love  County  with  90  earthquakes  following  the  
first  and  second  hydraulic  fracturing  stages  (Nicholson  and  Wesson,  1990).  
 
98°15'0"W

98°0'0"W

97°45'0"W

Canadian

97°30'0"W

97°0'0"W

Okfuskee

FNO

Cleveland

35°15'0"N

Pottawatomie

Caddo

W35A

Grady

35°0'0"N

35°0'0"N

Pontotoc

Garvin

Comanche

Seminole

McClain

34°45'0"N

34°30'0"N

96°45'0"W

Oklahoma

W34A

35°15'0"N

97°15'0"W

34°45'0"N

X34A

Murray

Stephens

34°30'0"N

X35A

Johnston

Cotton
34°15'0"N

34°15'0"N

Carter
Jefferson

34°0'0"N

33°45'0"N

Y34A

Love

Marshall
Y35A

34°0'0"N

Bryan

0 4 8 16 24 32
Kilometers
98°15'0"W

98°0'0"W

97°45'0"W

33°45'0"N
97°30'0"W

97°15'0"W

97°0'0"W

96°45'0"W

 

Figure  1  -­‐  Earthquakes  from  1897-­‐2010  from  the  OGS  catalog  (red  crosses).    Yellow  
triangles  are  seismic  stations  from  the  Earthscope  Transportable  Array;  tan  triangles  
are  OGS  seismic  stations.    Faults  are  shown  by  thin  black  lines,  solid  are  faults  
mapped  from  a  surface  expression,  dotted  lines  indicate  subsurface  faults  (Northcutt  
and  Campbell,  1995).    The  main  movement  on  all  of  these  faults  was  in  the  
Pennsylvanian  (Granath,  1989).    Hachured  region  shows  the  location  of  the  Eola  Oil  
Field  (Boyd,  2002).  

 
South-­‐central  Oklahoma  has  a  significant  amount  of  historical  seismicity,  and  has  
been  one  of  the  most  active  areas  within  Oklahoma  since  1977,  when  an  adequate  
seismic  network  was  established.    The  nearest  stations  to  the  earthquakes  were  
several  Earthscope  Transportable  Array  (TA)  stations.    Without  the  TA  stations  only  
 

2    

 

 
a  few  of  the  earthquakes  could  possibly  have  been  located  and  the  uncertainties  in  
the  hypocentral  locations  would  be  quite  large.  
 
Geologic  Setting  
 
The  Eola  Field  lies  at  the  northern  edge  of  the  Ardmore  Basin  and  the  buried  
northwestern  extent  of  the  Arbuckle  Mountains  and  contains  a  highly  folded  and  
faulted  thrust  system  (Swesnick  and  Green,  1950;  Harlton,  1964,  Granath,  1989).    In  
the  Cambrian  this  area  experienced  significant  rifting  associated  with  the  Southern  
Oklahoma  aulacogen  (Keller  et  al.,  1983).        After  the  initial  rifting  the  area  
experienced  thermal  subsidence  and  sedimentation  (Granath,  1989).    The  area  
continued  to  see  periods  of  subsidence  and  sedimentation  with  a  few  periods  of  
erosion  represented  by  unconformities  (Swesnik  and  Green,  1950).    In  the  mid-­‐
Pennsylvanian  the  area  began  to  experience  significant  transpression  associated  
with  the  Ouachita  Orogen  (Granath,  1989).    Because  of  the  areas  complex  tectonic  
history  it  is  quite  likely  that  the  nature  of  faults  has  changed  through  time  and  that  
normal  faults  associated  with  the  aulacogen  and  later  basin  accommodation  were  
reactivated  in  the  mid  to  late-­‐Pennsylvanian  with  a  new  sense  of  motion.    The  
Washita  Valley  fault  is  the  largest  fault  in  the  area,  with  a  surface  trace  of  
approximately  56  km  (Tanner,  1967).    It  is  a  major  fault  that  is  known  to  extend  
nearly  180  km  from  the  Ouachita  thrust  system  in  the  southeast  to  the  Anadarko  
basin  to  the  northwest  (Tanner,  1967).    Estimates  for  the  amount  of  left-­‐lateral  
strike-­‐slip  accommodated  on  this  fault  vary  dramatically,  but  reasonably  range  from  
65  km  (Tanner,  1967)  to  26  km  (McCaskill,  1998).    The  Roberson  fault,  to  south  of  
the  Washita  Valley  fault,  is  a  thrust  fault  with  an  associated  overturned  syncline  
with  significant  shortening  (Swesnick  and  Green,  1950).    The  Reagan,  Eola  and  Mill  
Creek  faults  as  mapped  by  Harlton  (1964)  all  show  significant  components  of  left  
lateral  strike  slip  (Granath,  1989).      The  Eola  field  contains  several  fault  blocks  in  
between  these  major  faults,  with  all  the  faults  in  the  subsurface  having  near  vertical  
dips  (Harlton,  1964).    To  the  southeast  of  the  Eola  Field  is  the  highly  faulted  West  
Timbered  Hills  of  the  northwestern  Arbuckle  Mountains  (Harlton,  1964).          
 
The  Eola  Field  was  discovered  in  1947  with  a  discovery  well  completed  to  a  total  
depth  of  10,234  feet  (3,119  m)  in  the  basal  Bromide  Sandstone.    The  initial  bottom  
hole  pressure  was  about  3800  PSI  and  by  1950  had  declined  to  2,900  PSI  with  seven  
producing  wells  in  the  field  (Swesnick  and  Green,  1950).  

 

3    

 

 

97°30'0"W

97°20'0"W

34°40'0"N

34°40'0"N

Garvin

Mill Creek Fa
ult

Eola Fa
u
R eaga

lt

n Fault

Washita
Valley F
au
Ro

34°30'0"N

ber

son

Fa

lt

ult

34°30'0"N

Carter
0

2

4

97°30'0"W

8

12

16
Kilometers
97°20'0"W

 

Figure  2  -­‐  Fault  map  for  the  Eola  Field,  Oklahoma.    Thick  green  lines  are  faults  
modified  from  Harlton  (1964).    Faults  shown  as  thin  grey  lines  are  from  Stoeser  et  al.  
(2007).    Eola  field  is  colored  a  salmon  color  (Boyd,  2002).  

 
Hydraulic  Fracturing  Operations  at  Picket  Unit  B  Well  4-­‐18  
 
Hydraulic  fracturing  operations  began  on  Monday  January  17,  2011  at  
approximately  6  AM  (CST),  12:00  UTC.    The  hydraulic  fracturing  of  the  well  
consisted  of  a  four-­‐stage  hydraulic  fracturing  operation  with  frac  intervals  of  9,830’-­‐
10,282’,  8,890’-­‐8326’,  7,662’-­‐8,128’,  and  7,000’-­‐7,562’,    with  the  last  frac  stage  
completed  on  January  23,  2011.    The  well  was  then  flushed  until  February  6,  2011.    
Because  the  earthquakes  began  after  the  first  frac  stage  we  will  primarily  consider  
this  stage.    The  first  frac  stage  had  an  average  rate  of  injection  of  88.5  bpm  and  an  
average  injection  pressure  of  4850  psi.    This  stage  also  included  an  acid  stimulation.    
There  was  a  total  of  2,475,545  gallons  of  frac  fluid  injected  and  575,974  lbs  of  
propent.    The  Picket  Unit  B  well  4-­‐18  is  a  nearly  vertical  well  located  at  34.55272  -­‐
97.44580,  elevation  277.4  m,  with  an  API  number  of  049-­‐24797.    The  first  frac  
occurred  in  the  interval  beteween  9,830’  (2,996.2  m),  and  10,282’  (3,134.0  m)  
 
Earthquake  Data  Analysis  and  Methods  
 
The  Garvin  County,  earthquakes  were  analyzed  using  routine  processing  steps  by  
the  OGS  for  earthquake  monitoring.    The  phase  arrivals  were  picked  using  the  
interactive  picking  capabilities  of  the  seismic  software  package  SEISAN  (Havskov  
and  Ottemoller,  1999).    The  earthquake  hypocenters  and  origin  times  were  
determined  using  the  location  program  HYPOCENTER  (Lienert  et  al.,  1986  and  
Lienert  and  Havskov,  1995).    The  OGS  typical  duration  magnitude  calculations  were  

 

4    

 

 
used  to  determine  the  magnitude  of  the  earthquakes  (Lawson  and  Luza,  2006).    The  
velocity  model  used  is  the  same  model  that  is  currently  being  used  by  the  OGS  for  
most  regions  of  Oklahoma,  the  “Manitou  Model”  shown  in  Table  1.    Using  the  
HYPOELLIPSE  method  hypocenter  locations  are  poorly  resolved  because  the  nearest  
station  is  approximately  35  km  away,  phase  arrivals  were  difficult  to  identify  for  
these  events,  and  the  events  appear  to  have  been  shallow.    The  formal  uncertainties  
are  significant  in  the  range  of  several  kilometers  for  these  earthquakes  and  indicate  
that  locations  for  these  earthquakes  should  be  considered  suspect.    The  formal  error  
estimates  from  the  initial  HYPOCENTER  locations  can  be  seen  in  Table  2.    Aside  
from  the  formal  uncertainties  it  is  very  likely  that  the  regional  velocity  model  used  is  
not  quite  appropriate  for  this  area  of  Oklahoma.    In  fact,  a  single  velocity  model  is  
definitely  not  appropriate  across  the  structurally  complex  region  spanning  the  deep  
(>10km)  Ardmore  basin  to  the  immediately  adjacent  Arbuckle  uplift.  
 
In  an  attempt  to  improve  the  hypocenter  determinations  the  Double-­‐Differencing,  
HypoDD,  technique  of  Waldhauser  and  Ellsworth  (2000)  was  employed  to  relocate  
the  earthquakes.  This  approach  takes  advantage  of  the  fact  that  there  is  very  little  
difference  in  phase  traveltimes  between  earthquakes  that  occur  near  each  other.    It  
can  then  use  all  the  earthquakes  to  find  the  centroid  of  the  earthquakes,  while  more  
easily  identifying  and  excluding  bad  phase  arrival  times  at  stations.    Once  the  
centroid  of  the  earthquakes  is  determined  the  relative  location  of  each  earthquake  
to  the  centroid  can  accurately  be  determined.    HypoDD  provides  very  good  
resolution  of  relative  earthquake  locations,  but  the  absolute  earthquake  location  
error  can  be  larger  than  the  formal  error  estimates.    The  singular  value  
decomposition  (SVD)  method  was  used  in  HypoDD  because  this  is  a  small  dataset,  
and  the  SVD  method  provides  a  better  estimate  of  the  formal  uncertainties  
(Waldhauser,  2001).  
 

HypoDD  also  has  the  capability  to  use  waveform  cross  correlation  between  events.    
Waveform  cross-­‐correlations  can  often  dramatically  improve  earthquake  
hypocentral  location  errors,  by  removing  any  human  error  in  phase  arrival  picks.    
This  method  uses  the  similarities  in  waveforms  between  events  to  more  accurately  
measure  arrival  times,  and  has  been  shown  to  dramatically  improve  locations  and  
their  associated  uncertainties.    Cross  correlation  is  simply  finding  the  part  of  the  
recorded  waveform,  which  is  most  like  the  template  waveform  window.    A  template  
waveform  window  is  an  example  waveform  around  either  a  P  or  S-­‐Wave  arrival  
from  an  event  within  the  earthquake  sequence.    Cross  correlations  for  these  
earthquakes  were  attempted.    In  order  to  attempt  the  cross  correlation  a  few  of  the  
larger  events  were  selected  as  templates  and  windows  around  the  P  and  S  phase  
arrivals  were  selected.    These  windows  where  then  run  against  the  corresponding  
template  event  to  determine  how  well  the  data  could  be  cross-­‐correlated  with  the  
entire  waveform  for  the  respective  earthquake.    In  this  test,  the  S-­‐waves  could  
readily  be  identified  using  cross  correlation,  but  the  P-­‐wave  could  not.    The  P-­‐waves  
were  generally  correlating  better  with  S  or  surface  waves  in  the  coda  than  with  the  
P  phase  arrival,  except  for  two  stations  X34A  and  Y34A.    The  inability  to  cross-­‐    
 

 

 

5    

 

 
Table  1  -­‐  Velocity  models  used  in  this  study;  all  have  a  Vp/Vs  ratio  of  1.73.    Manitou  
and  Chelsea  models  are  derived  from  Mitchell  and  Landisman  (1970).    Tryggvason  
and  Qualls  (1967)  model  was  developed  from  the  same  seismic  refraction  line  as  
Mitchell  and  Landisman  (1970).    The  Central  Oklahoma  model  was  developed  from  
traveltime  inversion  for  earthquakes  in  central  Oklahoma.  

Chelsea  
Thickness  (km)  
0.6  
0.4  
2.1  
10.3  
3.0  
1.5  
8.2  
9.1  
11.1  
half-­‐space  
 
 
 

 
 

 

Vp  (km/s)   Vs  (km/s)  
4.00  
2.31  
6.05  
3.50  
5.50  
3.18  
6.08  
3.51  
6.49  
3.75  
6.20  
3.58  
6.72  
3.88  
7.05  
4.08  
7.36  
4.25  
8.18  
4.73  

 
 
 
Manitou  
Thickness  (km)  
1.0  
9.5  
5.1  
2.5  
8.2  
9.1  
11.1  
half-­‐space  

Central  Oklahoma  

 
 
 

Thickness  (km)  
0.5  
0.5  
2.0  
0.5  
1.5  
3.0  
2.0  
5.0  
11.0  
9.0  
11.0  
half-­‐space  
 

Vp  (km/s)   Vs  (km/s)  
4.46  
2.58  
4.60  
2.66  
4.75  
2.75  
6.13  
3.54  
6.16  
3.56  
6.19  
3.58  
6.19  
3.58  
6.20  
3.58  
6.73  
3.89  
7.10  
4.11  
7.36  
4.25  
8.18  
4.73  

 
 
Tryggvason-­‐Qualls  

Vp  (km/s)   Vs  (km/s)   Thickness  (km)   Vp  (km/s)  
5.50  
3.18  
0.54  
4.00  
6.08  
3.51  
13.16  
5.96  
6.49  
3.75  
15.90  
6.66  
6.20  
3.58  
21.84  
7.20  
6.72  
3.88  
half-­‐space  
8.32  
7.05  
4.08  
 
 
7.36  
4.25  
 
 
8.18  
4.73  
 
 
 

6    

Vs  (km/s)  
2.31  
3.45  
3.85  
4.16  
4.81  
 
 
 

 


Related documents


of1 2011
gbu mountain news liii april 5 2014
jerusalem earthquake ppt
thomas snell curriculum vitae current
2 state eathquake complaint
gbu mountain news lxxii august 31 2014


Related keywords