INTERPRETING DATA FOR PERMANENT DOWNHOLE GAUGE - RolandHorne.pdf

1

1

Primary funding is provided by

The SPE Foundation through member donations

and a contribution from Offshore Europe

The Society is grateful to those companies that allow their professionals to serve as lecturers

Additional support provided by AIME

Society of Petroleum Engineers

Distinguished Lecturer Programwww.spe.org/dl

Listening to the Reservoir –

Interpreting Data from Permanent Downhole Gauges

Roland N. HorneStanford University



2

3

• More than 10,000 installed worldwide.

• Usually installed to monitor downhole equipment.

• Data rarely applied for reservoir analysis.

Permanent Downhole Gauges (PDG)

4

Reservoir Engineering Uses1. Reservoir pressure

2. Flowing bottomhole pressure management

3. Replacement for shut-in tests

4. Skin determination

5. Monitoring interference effects

6. Voidage control

7. Tubing hydraulics matching

8. Inflow performance modeling

9. Monitoring well treatments

10. History matching

Kragas, Turnbull and Francis (2004)

3

5

Replacement of Shut-In Tests

• Northstar, Alaska

6 wells, 2 days duration

10,000 STB/d production would be lost per well

� 120,000 STB acceleration per campaign

� 650,000 STB acceleration over field life

And, $1.6 million avoided wireline costs

Kragas, Turnbull and Francis (2004)

6

PDG-Specific Issues

1. Manipulation of huge volumes of data.

2. Deconvolution to see characteristic behaviors.

3. Identification of break points, to separate transients.

4. Changes (such as permeability and skin) as a function of time.

5. Flow rate information.

6. Temperature measurements.

4

7

1. Manipulation & Processing of Data

• Data at 1 second frequency =

32 million data/year per gauge.

• Physical storage and access are a challenge even for today’s databases.

• Access, retrieval and transfer are a challenge even for today’s bandwidths.

Chorneyko (2006)

8

1. Manipulation & Processing of Data

(Athichanagorn et al., 2002)

5500

5600

5700

5800

5900

6000

Pre

ssure

(psi

a)

0 100 200 300 400 500

Time (hour)

aaaaa

aa

a

a

a

a

a

a

a

aaaaaaa

aaaaa aa a a a

BehavioralBehavioral

aberrationaberration

Outliers

Missing dataMissing data

Noisea a a a a

aaaaaaaaaaaaaaaaaaaaaaaa

aaaa a

aaaaaaaaaaaaaaaaaaaaaaaaa

aa a a a a aaaaaaaaaaaaaaaaaaaaaaaaaaaa a a a a a

aa

aa

a a aaaaaaaaaaaaaaaaaaaaaa

aaaaaa a a a a a a a

aaaaaaa

a

a

a

a

a

a

a

a

aaaaaaaaa

aa

a

aa a

a

a a

a

a

a

a

a

a

a a aaaaaaaaa

a

a

a

a

a

a

aaaaaaaaaaaaa a a a a aaaaaaaaaaaaaa

aaaaaaaaaaaaaaaaaaa

aaa

aaaaaaaaaaaaa a a a a a aaaaaaaaaaa

aaaaaaaaaaaaaaaaa a a a a a a a a a aaaaaaaaaaaaaaaaaaaaaaaaaaaa a a a a a

aaaaaaaaaaaaaaaaaaaaaaaaaaa a a a a a a a a a a a a a a a a a a a a a

a

5500

5600

5700

5800

5900

6000

Pre

ssure

(psi

a)

0 100 200 300 400 500

Time (hour)

aaaaa

aa

a

a

a

a

a

a

a

aaaaaaa

aaaaa aa a a a

5500

5600

5700

5800

5900

6000

Pre

ssure

(psi

a)

0 100 200 300 400 500

Time (hour)

aaaaa

aa

a

a

a

a

a

a

a

aaaaaaa

aaaaa aa a a a

BehavioralBehavioral

aberrationaberration

Outliers

Missing dataMissing data

Noisea a a a a

aaaaaaaaaaaaaaaaaaaaaaaa

aaaa a



aa

aa



aaaaaaa

a

a

a

a

a

a

a

a

aaaaaaaaa

aa

a

aa a

a

a a

a

a

a

a

a

a

a a aaaaaaaaa

a

a

a

a

a

a

aaaaa

a a a a aaaaaaaaaaaaaaaaaaaaaaaaa

aaaa a



aa

aa



aaaaaaa

a

a

a

a

a

a

a

a

aaaaaaaaa

aa

a

aa a

a

a a

a

a

a

a

a

a

a a aaaaaaaaa

a

a

a

a

a

a

aaaaaaaaaaaaa a a a a aaaaaaaaaaaaaa

aaaaaaaaaaaaaaaaaaa

aaa




a

aaaaaaaa a a a a aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa

aaa




a

Missing flow data

Pressure

Rate

5

9

Denoising with Wavelets


Noisy signal

Denoised signal

10

Outlier Filtering with Wavelets


Outlier pointsAcceptable points

6

11

2. Deconvolution

• Theoretician’s playground, since 1949.

• Remained impractical until recently.

• Work of von Schroeter, Hollaender and Gringarten (2004), using nonparametric regression, p and q matching, derivative restrictions and smoothness limit constraints.

∫ −∆=∆t

w dtpqtp0

0 )().(')( τττpressure flow constant pressure

reservoir model

12

2. Deconvolution

2700

2720

2740

2760

2780

2800

2820

0 5 10 15 20 25 30 35 40 45

Time, hrs

Pre

ssure

, psi

0

2000

4000

6000

8000

10000

Flo

w r

ate

, bb

l/d

ay

Nomura (2006)

45 hours of data, no transient longer than 5 hours

events

5 hours

7

13

2. Deconvolution

Nomura (2006)

1

10

100

0.001 0.01 0.1 1 10 100

Time, hrs

Pre

ssure

derivative,

psi

TRUE

case1

case2

case3

45

2700

2720

2740

2760

2780

2800

2820

0 5 10 15 20 25 30 35 40 45

Time, hrs

Pre

ssu

re, psi

0

2000

4000

6000

8000

10000

Flo

w r

ate

, bbl/day

45 hour response, inferred from deconvolution

14

2. Deconvolution – Issues

• Model may change over time.

• Buildups and drawdowns may be different.

– Levitan (2005): shut-ins only

– Olsen & Nordvedt (2006): shut-ins only

• Strong dependence on break points.

8

15

3. Transient Identification – Break Points

Wavelet approach: Athichanagorn et al. (2002)

2720

2730

2740

2750

2760

2770

2780

2790

2800

2810

0 5 10 15 20 25 30 35 40 45

Time, hrs

Pre

ssur

e, p

si

Nomura (2006)

16

3. Break Points and Deconvolution

Nomura (2006)

1

10

100

0.001 0.01 0.1 1 10 100

Time, hrs

Pre

ss

ure

de

riv

ati

ve

Wavelet

Adjusted

TRUE

Inaccurate break points are fatal for deconvolution

9

17

3. Break Points and Rate Normalization

Houzé (2006)

Inaccurate break points are also problematic for rate normalization

18

3. Break Points – Approaches

• Wavelets often used for ‘first round’.

• Khong (2001): statistical method.

• Rai (2005 and 2007): – Savitzky-Golay smoothing filter

– Segmentation method

• Olsen & Nordvedt (2005): pattern matching method.

• Nomura (2006): insertion and deletion.

10

19

3. Break Points

Nomura (2006)

case

1

2

3

initial

18

36

11

insertion

136

167

151

deletion

95

81

101

threshold

0.5

0.01

2

1

10

100

0 .001 0 .01 0 .1 1 10 100

T im e, hrs

Pre

ssure

deriva

tive, p

si

T R U E

case1

case2

case3

20

• Constant property solution

• k(t)=constant , s(t)=constant

(Lee, 2003)

4. Permeability and Skin Change in Time

Real data

Data match, constant k and s

11

21

• Quadratic function for permeability

• k(t)= a(t-tL)2+kL , s(t)=s0 + b t

(Lee, 2003)

4. Permeability and Skin Change in Time

Real dataData match, variable k and s

22

4. k and s Changes – Field Observations

• Khong (2001)

• Athichanagorn, Horne and Kikani (2002)

• Richardson, Roux, Quinn, Harker and Sides (2002)

• Lee (2003)

• Haddad, Proano and Patel (2004)

• Coludrovich, McFadden, Palke, Roberts and Robson (2004)

• Chorneyko (2006)

• Olsen and Nordtvedt (2006)

12

23

4. k and s Changes

• de Oliviera and Kato (2004): “analytical models used traditionally for conventional well test interpretation may be too simple to define the pressure and flow rate transients that occur during the extended duration of a permanent downhole gauge record.”

• Using full-scale numerical models is probably what we need, but not widely applied (yet).

24

5. Downhole Flow Rate Gauges

• Both p and q contain measurement errors.

• Match both simultaneously.

• von Schroeter, Hollaender and Gringarten(2004)

• Nomura (2009)

• Ahn and Horne (2008)

2

1

22

)(2

1qaDbRpdobj

cn

r

r

rrrrr−++−= ∑

=

µλ

13

25


Ahn (2008)

450 500 550 600 6508894

8896

8898

8900

8902

8904

8906

8908

8910

Time (seconds)

Pre

ssure

[psia

]

measured

iteration 1

iteration 2

iteration 3

iteration 4

450 500 550 600 6506800

7000

7200

7400

7600

7800

8000

8200

8400

8600

Time (seconds)

Liq

uid

rate

[bbl/d]

measured

iteration 1

iteration 2

iteration 3

iteration 4

Pre

ssu

re (

psia

)

Ra

te (

ST

B/d

)Time (seconds)Time (seconds)

26


p

q

A B

A – flow event B – noise event

Pressure

Rate

14

27

6. Temperature Data

Temperature respondsto flowratechanges

Temperature respondsto flowratechanges

Duru (2008)

q

T

28

6. Temperature Data

Duru and Horne (2008)

Match temperaturehistory �� porosity

15

29

Concluding Remarks (1)

• Permanent downhole gauges are rich sources of reservoir data.

• Not just more, but better!

• Good progress on:– Noise and outlier removal

– Break point identification

– Deconvolution

– Combining rate data

– Utilizing temperature data

• But, more work to do!

30

Concluding Remarks (2)

• The ultimate goal is to achieve a high degree of automation.

• Nobody wants to look at 100 million data points!

• Eventual inclusion in SmartFieldsprocedures.

16

31

Acknowledgements

• Members of the SUPRI-D research consortium on innovation in reservoir testing.

• SUPRI-D graduates:

– Athichanagorn (1999)

– Khong (2001)

– Lee (2003)

– Rai (2005)

– Nomura (2006)

– Duru (2008)

– Ahn (2008)

32

SPE Distinguished Lecturer Program

The SPE Distinguished Lecturer Program is funded principally through a grant from the SPE Foundation.

The society gratefully acknowledges the companies that support this program by allowing their professionals to participate as lecturers.

Special thanks to the American Institute of Mining, Metallurgical, and Petroleum Engineers (AIME) for its contribution to the program.



17

33


Distinguished Lecturer Programwww.spe.org/dl 33

Your Feedback is Important

Enter your section in the DL Evaluation Contest by

completing the evaluation form for this presentation or go online at:

http://www.spe.org/events/dl/dl_evaluation_contest.php

INTERPRETING DATA FOR PERMANENT DOWNHOLE GAUGE - RolandHorne.pdf

Documents

reservoir interpreting

function of time

reservoir analysis

stb acceleration

downhole equipment

reservoir pressure2

shutin tests4

todays bandwidths