Long-distance acoustic emission monitoring of hydraulically induced subsurface cracks in nigorikawa geothermal field, Japan

0375 6505/85$3.00 + 0.00 Pergamon Press Ltd. ,:: 1985CNR.

Geothermics, Vol. 14, No. 4, pp. 539 551, 1985.

Printed in Great Britain.

LONG-DISTANCE ACOUSTIC EMISSION MONITORING OF HYDRAULICALLY INDUCED SUBSURFACE CRACKS IN NIGORIKAWA GEOTHERMAL FIELD, JAPAN H. NIITSUMA,* K. NAKATSUKA,* H. T A K A H A S H I , * N. C H U B A C H I , * H. YOKOYAMA* and K. S A T O t *Faculty of Engineering, Tohoku University, Sendai 980 Japan and "fJapan Metals and Chemicals Co. Ltd., Takizawa 020-01, Japan (Received February 1984; accepted jor puhlication Decemher 1984)

AbstracI--Acoustic emissions (AE) during a massive hydraulic fracturing experiment in a geothermal production well in Nigorikawa field, Hokkaido, Japan, were measured using an acceleration-sensitive downhole AE sonde installed in a 30 m well. Crack extensions at a depth of 1500 m were detected and characterized by long-distanceAE measurements. The mechanism by which productivitywas increased in the geothermal well through hydraulic fracturing has been represented by a model based on AE data and geological considerations. INTRODUCTION Hydraulic fracturing has been successfully applied as a well stimulation technique for both improvement and recovery of productivity in a natural hydrothermal reservoir, as well as for creating fractures in HDR systems (Katagiri et al., 1980; Nakatsuka et al., 1982; Aamodt, 1977). It is important to monitor fracture extension in the rock mass and to determine the configuration of the resulting reservoir, in order to make fracturing effective and to maintain the created reservoir in a stable manner for a long period of operation. The acoustic emission (AE) technique is used to detect a subsurface crack extension in gas storage reservoirs and mines (Hardy, 1981). Since high frequency components of seismicity are measured by highly sensitive downhole detectors, microearthquakes, which are too small to be detected by surface seismometers, can be detected and located in the AlE measurement. The technique is also used to locate hydraulically induced cracks in HDR development projects (Albright and Pearson, 1982; Leydecker, 1981). Deep wells, drilled adjacent to the fractured zone, were used for the measurements. The advantages of using the AE technique in shallow test boreholes are: (1) shallow test boreholes are cheaper and easier to drill, so that an acoustically suitable point can be selected for the measurement and (2) since temperatures are quite low in these wells, long-term or continuous monitoring of crack extension is technically feasible. The drawbacks are that signal quality would be impaired by transmission losses, background noise and reflected waves from the surface. The decrease of prospective angle would also affect the resolution of AE source location in the long-distance measurement. Acoustic emissions during a massive hydraulic fracturing experiment in a geothermal production well in Nigorikawa field (Katagiri et al., 1980), Hokkaido, Japan were measured using an acceleration-sensitive downhole AE sonde installed in a 30 m well. Crack extensions at a depth of 1500 m were successfully monitored by the long-distance AE measurement technique (Takahashi et al., 1980; Niitsuma et al., 1982). The results of these measurements will be described in this paper and a mechanism for improving productivity by fracturing will be discussed. 539

540

H. Niitsurna et al.

H Y D R A U L I C F R A C T U R I N G AS A WELL S T I M U L A T I O N T E C H N I Q U E IN T H E NIGORIKAWA G E O T H E R M A L FIELD A 50 MWe geothermal electric power plant was installed in the Nigorikawa (Mori) area in Hokkaido, Japan, by Dohnan Geothermal Energy Co. Ltd., a subsidiary of Japan Metals and Chemicals Co. Ltd., and was put into operation in November 1982. The location of the field is indicated in Fig. 1. A total of 17 geothermal wells, ranging from 700 to 2400 m in depth were drilled in this area between 1977 and 1981. In order to stimulate the geothermal wells, massive hydraulic fracturing experiments were performed in 10 wells in two separate projects in 1978 and 1980. Although these fracture tests represent the first attempt at stimulating geothermal wells, the resulting increase in productivity or injectivity encourages similar attempts in other geothermal areas.

+ 42°N-~~

s~

140°E 141°E

Fig. I. Location of Nigorikawa geothermal field. Nigorikawa geothermal field is located in a Krakatoan (crater lake) type caldera forming an approximately circular basin of 3 km diameter. The main rock units consist of the pre-Tertiary Kamiiso Group, Miocene formations and Pleistocene caldera-fill deposits (Ide, 1982). The geothermal fluids are stored in fracture zones along the caldera-margins and also in fractures and faults within the Kamiiso Group which consists of slate, chert, limestone and andesitic tuff; this group does not outcrop on the surface but has been recognized in the geothermal wells. Steam and hot water has been produced, at a flow rate of 400 t/h and 1200 t/h under wellhead pressure of 1.27 MPa respectively, from the faults in the Kamiiso Group. An isopach map and well location in the field are shown in Fig. 2; the dashed lines indicate the directions of the three main faults within the Kamiiso Group. The three s t r i k e - s l i p and transverse faults trending from northeast to southwest have a small steeply dipping reverse component. Seven wells, including four which had already been stimulated once in 1978, were stimulated in the spring of 1980. An inorganic gelling agentwas chosen as fracturing fluid. The retrievable packer was set in the 9 ~ inch (245 ram) production liner, usually 60 m below the 133A inch (340 mm) casing shoe. The wells were propped with 20/40 mesh sand. The experiments were carried out in at least two stages with diverting material used between stages. Productivity or injectivity were considerably enhanced by stimulations, as shown in Table 1. L O N G - D I S T A N C E AE M E A S U R E M E N T OF T H E H Y D R A U L I C A L L Y INDUCED CRACKS The long-distance AE measurement was carried out during hydraulic fracturing of D-I well in May 1980. Figure 3 is a N W - S E cross-section of the Nigorikawa field: with the expected

541

Acoustic Emission Monitoring, Nigorikawa Field N

.oo...

1 "'...

-,oo m,.

"'! ......'.,

.............

:,,o?.~°°°'~

,

A ~

C-2

l

~ B

/ ...... / < ~ ) .

/ _.AEwe,,

...../ <)/<, /. /, o o .- ............. ~oo................ .............

" ...........7"'

\

~

.

~

I

....................900............

~

...... ,ooo .........

- " . . ~ , _ ~ U ~ . ' ~ , o o , , -"

"..

-..

J

I;-9

"',

-

"..

:5 "2

.........

....... . . . . . . . .

\,

"

"--._5,oo°-'~

~

-~,.....

....

....-[~oo~

500 X ~"

k

....' ......... -loooN'"

\ o

Fall 1978 stimulations

s o o"

Spring 1980 stimulations

•

B-1

O D-1

S

B-2

O

F-l°

•

B-2

•

D-2

•

C-1=

•

F-3¢

•

C-1°

O

D-6

O

•

F-6b

•

C-2

D-1 D-6

O

Fig. 2. Isopach and location map of ,,,,ells in the Nigorikawa field ((_): production well, ~ : injection well) (Katagiri a l . , 1980).

et

f r a c t u r e l o c a t i o n m a r k e d with an arrow. C r a c k extension was expected f r o m this l o c a t i o n during h y d r a u l i c fracturing, because the m a x i m u m circulation loss d u r i n g drilling, which indicates the existence o f a large pre-existing crack, was o b s e r v e d at a d e p t h o f a b o u t 1400 m. A n A E test b o r e h o l e o f 120 m m d i a m e t e r was drilled to 30 m d e p t h within the M i o c e n e f o r m a t i o n at the l o c a t i o n shown in Fig. 3. The distance between the expected fracture l o c a t i o n a n d the A E m o n i t o r i n g well was a b o u t 1600 m. Figure 4 shows the d o w n h o l e A E sonde. Piezoelectric a c c e l e r o m e t e r s were used as A E t r a n s d u c e r s for the l o n g - d i s t a n c e m e a s u r e m e n t . The acceleration-sensitive i n s t r u m e n t a t i o n is highly sensitive to the higher frequency c o m p o n e n t o f the A E used to detect a n d locate fractures. L o w e r frequency b a c k g r o u n d noise a n d microseisms f r o m o t h e r d i s t a n t seismic sources can also be s u p p r e s s e d in the i n s t r u m e n t a t i o n . Twelve piezoelectric a c c e l e r o m e t e r s

C/") ~4z.

H. Niitsuma et al. Table I. Well test date before and after the hydraulic fracturing test (Katagiri et ul., 1980) lnjeclivity ~est

\kell No.

Production test

Approximate date* Rate (t/h)

B-I

Before Autumn 1978 After Autumn 1978

B-2

Before After Before After

C-la

Before After Before After

C-2

Sleanl flo~

(MPa)

rate (t/h)

0 48

3.4 3.4

Autumn 1978 Autumn 1978 Spring 1980 Spring 1980

69 100 85 112

0.93 0,59 0.54 0.54

Autumn 1978 Autumn 1978 Spring 1980 Spring 1980

45 13(7) 100 92

0.49 0.69 0.54 0.54

10.0 14.9 17.5

Before Autumn 1978 After Autumn 1978

15 82

1.47 1.27

--

0.62 (/.33 (/.54

D-I

Before After Before After

Autumn 1978 Autumn 1978 Spring 1980 Spring 1980

160 t72 255 --

D-2

Before Autumn 1978 After Autumn 1978

23 48

D-6

Before After Before After

F-la F-3c F-6b

Pressure

Autumn 1978 Autumn 1978 Spring 1980 Spring 1980

Hot wateF |'lo\~

20

18.8 20.0 --

.

.

.

. -

43 147 I 18

0.2/) 0,53 (I.77

59.0 139.0 150.0

0.03 0.91 0.77

-

20.0 46.5 56.0 67,2

(I.72

(MPa)

.

-

O. 86

Pressure

rate (t h )

76 107 180 289 .

.

0.59 0.49 0.77 O.77

. .

-

.

.

.

103 220 220

2.4 1).2 0.2 --

Before Spring 198(I After Spring 1980

49 220

0.67 0.67

No continuous flow Not measured 108

Before Spring 1980 After Spring 1980

-110

(I.48

No continuous flov, No continuous flow

Before Spring 1980 After Spring 1980

0 66

0.54 0.44

No continuous flow Not measured 94

20,0 34,7 43.0 61,0

40.0 50.8 125 168

0.20 0.52 0.77 0,77 0.22

O. 1 I

*Indicates tests before and after hydraulic fracturing.

Well stimulatedwith hydraulic fracturing \

SW

500

AEmonitoringlocation NE

. . . . . . . ,- • ~'x~, ...'. : .; ...... ;" "-~,,, Caldera f i l l Miocene strat; ;~deposit ~

Hornblendeandesite / ~.-:~'°~ ..,,I

(tuff, siltstone

-

.'..\...z~

&, s,andstorle.) ...."

....

"?~?~/"~x

-

~

Mi0~ene "st:rata""" ! - "-,

.~..~--Zz.,

~'/

4/' ~

~

I000

~

e

~

~

,

r

t

re-e

1500

Expected Fracture Zone

2000m aldera woll

0

500

l

i

i

I000 m J

Fig. 3. N W - SE cross-section of the Nigorikawa field.

a

~l~t~

li,~stone

r

y

.Ka~iis° Group

Acoustic Emission Monitoring, Nigorikawa Field

543

Cable Head

t

Electronic Circuit Package

Transducer Housing

1

Fixing Arm

Fig. 4. Downhole AE sonde.

(Node Corp. A-26 type; charge sensitivity: 30 pCsZ/m; resonant frequency: higher than 10 kHz; maximum operating temperature: 150°C) were mounted on the transducer housing. For triaxial measurements (Albright and Pearson, 1982), three groups of transducers were directed at right angles to one other. Four transducers in each group were connected electrically in parallel. Figure 5 shows the transducer housing. Three charge-sensitive preamplifiers were mounted in the electronic package. The gain of the preamplifier can be changed by a c o m m a n d signal from ground surface. Overalt sensitivity and frequency range of the sonde are 270 or 27 Vs2/m and 7 Hz to 7 kHz, respectively. A mechanical fixing arm, similar to that made by Los Alamos National Laboratory (Albright and Pearson, 1982), was used. The arm is driven by a high temperature d.c. motor so that the side of the sonde opposite to the arm is pressed against the borehole wall. The weight of the sonde also increases the fixing pressure. Although this fixer is simple and applicable to boreholes of various sizes, its fixing performance was not sufficient for accurate locations of the AE source (Niitsuma et al., 1982). Since the temperature was not high in the shallow test borehole, a normal 50 m multiconductor cable (cabtyre cable) containing three coaxial cables and 24 lines was used as instrumentation cable. The AE signals were amplified by 1 0 - 50 dB in a ground surface station where data recording, waveform monitoring, audible sound monitoring, AE event-ringdown counting and data printout were carried out. The recorded signals were digitized and processed on a personal computer after measurement. AE monitoring was performed before, during (about 1.5 hours) and after fracturing, for a total of 3.5 hours, in order to check the background noise level. Both wellhead pressure and flow rate were also monitored. RESULTS A N D DISCUSSION Twenty-one AE events were detected during hydraulic fracturing by waveform monitoring. The m a x i m u m value of their amplitude is 0.86 Gal. In order to detect the signals covered by background noise, the tone colour of the signal was monitored by ear by setting the reproducing

544

H. Niitsurna et al.

Fig. 5. Phologtaph o!" the Iransdllccl hotlSillg.

speed of the data recorder four times faster than the recording speed. Twenty-two additional small signals were detected by this method. Since no AE event was observed during the hour before and after fracturing, the observed AE signals were, therefore, generated by the fracturing test. The first part of the AE signals that can be clearly distinguished from noise is shown in Fig. 6. The w a v e f o r m is a damped one with a fundamental frequency of about 40 Hz. Although the P-wave arrival time is not so clear due to the inadequate fixing performance of the sonde, the S-wave arrival time can be determined from the X - Y Lissajou pattern shown in Fig. 7. Z - c o m p o n e n t of signals could not be detected because of measurement problems. Figure 8 shows the AE events and their ringdown counts, together with the p r e s s u r e - and flow r a t e - time curves during the fracturing test. The dots in the figure represent the small AE signals detected in the tone colour monitoring. Normally a hydraulically induced fracture formation is inferred by an abrupt pressure drop, denoted as a breakdown, under a condition of increasing or constant flow rate. In this figure, however, one cannot determine the breakdown point clearly from pressure change only. On the other hand, the AE events are presumably generated potentially by a tension elastic rebound at the fracture boundary and rock bursts near the fractured zone. Therefore, the AE technique is effective for monitoring the abrupt breakdown during hydraulic fracturing. All large AE signals observed are listed in Table 2. The signals can be classified into three groups by their waveform. In type A the shear wave is large and clear and the fundamental frequency of the signal is about 40 Hz. In type B the shear wave is less clear than type A and the frequency is slightly lower ( 3 7 - 4 0 Hz). Signals of type C have the lowest fundamental frequency ( 3 0 - 3 4 Hz). A, B and C in Fig. 6 correspond to types A, B and C respectively.

xt

Type (A)

AA~

AA 'VT,ME

Fig. 6. Typical del"ecled AE signals.

Type (B)

X

Type (C)

.la

"'I

e~

aa

546

tt. ,~'iitsuma et al. Y

~--

P WAVE

"r"

4g

1

X

vT i

!

REFLECTEDWAVE

S WAVE

X

i

I

~

X'q

Fig, 7. Lissajou pattern on X-- Y plane.

The distances between sonde and AE source, estimated by the hodogram method (Albright and Pearson, 1982), are also given in Table 2, where the wave velocities are assumed to be vp = 4330 m / s , v~ = 2500 m / s , according to core sample data. Estimated relative directions in the X - Y (horizontal) plane are also indicated in the Table. In Fig. 9 the angle and distance of the AE sources are plotted in polar coordinates. The arrows indicate the sequence of AE generation. Figure 10 shows the profile through the line A - A ' in Fig. 2. Well D-6 was stimulated before well D-1. Both wells D-1 and D-6 are almost on the same plane. F1, F 2 and F3 represent the three m a j o r faults of the Kamiiso Group. LC indicates loss of circulation of mud water during drilling, which indicates that the pre-existing crack in the geothermal reservoir was adequate. Circulation losses mainly occurred along the faults, in the upper part of the limestone layer and in the fractured chert layer (dotted layer in Fig. 10), in the depth range from 1200 to 1400 m. The Kamiiso G r o u p is presumably subdivided into two formations by the unconformity

Acoustic Emission Monitoring, Nigorikawa Field

547

• : Af events detected by tone c o l o r moni torinq

~0 AE Ringdown counts :~ 4O r

30 ,~ 20 Zgl0

-],.

r

I? 5O 2 ,~0

/

* ~

.... o

t

i

I -.

f

I

[~

. . . . . .

T .........

10;40

, .........

"0:5C'

, .........

p .....

fl:O0

" d2= o

...... t

, ....

, ....

, ....

, ....

f ....

, ....

, ....

, t .

6,2 Z~'"

, . , , , ' 7

1110

;I ?0

11:30

1i:40

11:50

' 2 :,3:9

Time

t-ig. 8. Af£ ringdown counts and pressure

and flow rate

time cur~es during l r a c t u r i n g t e s t .

Table 2. AE signals detected during hydraulic fracturing

Fime I sl Stage 10:41 '12" 10:51 '22" I0: 59' 55" 11:07 '45" 11:10'42" 2nd Stage I 1:36'02" I 1:36' 411" 11:46' 20"

I 1:49'40" I 1:55' 25" 11:56'50" 11:57' 12" 12:01 '10"

No.

P-S arrival time difference (ms)

Distance (m)

Located AE source Direction (deg.)

Type

frequency (He)

Fundf, mental

3 6 10 1I 13

271 273 224 209 186

1603 1613 1325 1238 1100

37.0 34.7 33.0 43.3 21.0

A A B A B

37 40 4(} 4() 37

17 19 24 25 26 29 30 31

216 239 268 225 290 263 288 188

1278 1414 1585 1331 1715 1556 1704 1109

43.5 25.1 15 22 34.5 26.0 15 18 23.1 30.0

B A C C C B ( A

39 40 30 3I 30 40 34 411

between the slate layer and the fractured chert layer. The fractured chert layer includes the naturally induced pre-existing crack and networks of pores caused by weathering. It was expected that a crack initiation would occur by the fracturing at the point indicated by the "LC" just below the casing bottom in well D-I. As shown in the figure, D-1 well intersects fault F2 at this point. A mutual relation (i.e. interference) between D-1 and D-6 wells was found after hydraulic fracturing by tracer testing and an analysis of wellhead pressure change. The tracer testing was made by injecting potassium iodide into well D-I under production conditions and by analysing water samples collected from D-6 well. Arrival time from D-I to D-6 was 27 hours and the peak of tracer contained in hot water appeared after 108 hours of tracer injection. Moreover, when a production test was carried out in F-I well, wellhead pressure in both D-I and D-6 wells built up simultaneously. Therefore, the crack presumably extended along fault F2 and the naturally fractured chert layer (dotted layer in Fig. 10). This mechanism is shown schematically in Fig. 12(a) and (b).

548

ft. Niitsuma et al.

(m)

15001

10001

{

~'i

i

i ,

j---

'°°

510

i.J/ z

zW _J--......

s"

"

..-'\

/

/

~

\

/,

45°

~,~~J-/~'-~ 0

j

-" / /

/

60°

(m)

Fig. 9. D i s t a n c e a n d a z i m u t h o f A t : s o u r c e s .

Although the Z-axis data are required for AE source mapping, horizontal distribution of the AE source may be plotted as shown in Fig. l 1, assuming that the crack extended in the manner described above. F1, F2 and F3 in Fig. 12 indicate the location of the faults near the depth of the expected fractured zone (1300- 1500 m). The crack extension process during hydraulic fracturing in D-1 well may be explained by means of the AE data, as follows, referring to Figs 10 and 11. (1) The shear dominant A-type events, Nos 3 and 6, suggest that a shear crack extended from the point of maximum circulation loss near the casing shoe along the fault plane F2. (2) The crack was kinked in the naturally fractured chert layer mentioned above, and extended through the layer towards fault F3, emitting B-type signals (event No. 10), which could indicate opening mode cracking. (3) Shear mode cracking occurred again (No. 11) when the crack reached fault plane F3, where a hydraulically induced crack from D-6 well already existed. (4) The crack then extended through the layer (Nos 13 -30), nearly parallel to the line A - A ' , which coincides with the maximum tectonic stress direction in she Nigorikawa area (Nakamura, 1977). The hydraulically induced cracked layer thus became a new geothermal reservoir. Figure 12 shows the newly created reservoir schematically. This figure assumes that hydraulic fracturing not only increased the permeability of the naturally cracked layer but also formed a two-dimensional reservoir with many inlets of geothermal fluid along the faults. As a result, productivity in both D-1 and D-6 wells was considerably improved. CONCLUSIONS The hydraulically induced crack propagations in the Nigorikawa geothermal field were detected and characterized by acceleration-sensitive long-distance AE measurement in a 30 m borehole. Because of the low temperature and low drilling costs of shallow test boreholes, longterm or continuous monitoring of subsurface cracks is economically and technically feasible.

549

Acoustic Emission Monitoring, Nigorikawa Field

Although some ambiguity still remains, the hydraulically induced crack formed by fracturing in D-1 well is assumed to have propagated first along the pre-existing cracks including faults and naturally fractured layers and then to have been controlled by the main tectonic stress direction in the field. The mechanism by which hydraulic fracturing increased productivity has been represented schematically by a cracking model. The AE monitoring technique could provide useful information on the formation and maintenance of artificial reservoirs for geothermal development. D6 AE D1

A

A'

coldero fill deposit

500

tone

I000-

-stein

-stein

1500-

-stein

2000-

F1

F2

F3

Fig. 10. Profile~hrough lineA

A' of Fig. 2.

550

H. Niitsuma

etal,

/

~ 0

/

/

1I I

ZOOm

""--A

Fig. 11. Horizontal distribution of AE sources.

D-I well

D-6 well

Faults /' "~

D-I

D-6

Fnults

i

D-I

\,

I

J

/

~ i

D-6

ts of fluids

I

Faults (path of fluids)

(a) before stimulation

Hydraulically developed reservoir

(b) after stlmulatlon

Hydraulically developed reservoir

(c) after stlmulatlon ( top vlew )

Fig. 12. Model of reservoir created by hydraulic fracturing.

REFERENCES A a m o d t , R. L. (1977) Hydraulic fracture experiments in GT-I and GT-2. Los Alamos Scientific Laboratory Report LA-6712-MS, Los Alamos, New Mexico. Albright, J. N. and Pearson, C. F. (1982) Acoustic emissions as a tool for hydraulic fracture location: experience at the Fenton Hill hot dry rock site. Sac. Petrol. Engng d. 523 -- 530. Hardy, Jr., H. R. (1981) Application of acoustic emission techniques to rock and rock structures: a state-of-the-art review. Acoustic Emissions in Geotechnical Engineering Practice, STP 750, American Society for Testing and Materials, pp. 4 - 92. lde, T. (1982) Geology in the Nigorikawa geothermal field, Morimachi, Hokkaido, Japan. Geother. Resour. Council Trans. 6, 31 - 3 3 . Katagiri, K., Ott, W. K. and Nutley, B. G. (1980) Hydraulic fracturing aids geothermal field development. ~l"orld Oil 191, 7 5 - 8 8 .

Acoustic Emission Monitoring, Nigorikawa Field

551

I cydeckcr, G. (1981) Seismische Ortung Hydraulisch erzeugter Briiche im Geothermik Frac Projekt |:alkenberg. Bundesanstalt./kir Geowi.ssensch~(/Ten and RohstolJe. Hannover, Archiv-Nr. 86 549. N~tkamttra, K. (1977) Volcanoes as possible indicators of tectonic stress orientation--principle and proposal. J. lo'c. (;eolher. Re.s. 2, I 16. Nakatsuka, K., Takahashi, H+ and Takanohashi, M. (1982) Hydraulic fracturing experiment at Nigorika~,a and l'ractnre mechanics evaluation. Proc. ]~t Japan U.S .Ioint Seminar on H.vdraulic Fracturin,~ and Geothermal t:ner~y, Tokyo, pp. I 17. Niilsuma, H., Nakatsuka, K., Takahashi, H., Chubachi, N., Abe, M., Yokoyama, H. and Sato, R. (1982) In situ AE measurement of hydraulic fracturing at geothermal fields. Proc. Ist Japan US Joint Seminar on 1Yvdraulic t:racturing and Geothermal Energy, Tokyo, pp. 227 241. [+akahashi, H., Niitsuma, H., Tamakawa, K., Abe, H., Sato, R. and Suzuki, M. (1980) Detection of acoustic emission during hydraulic fracturing for geothermal energy extraction. Proc. 5th Int. A E ,~vmp., Tokyo, pp. 443 453.