Digital Rock, Pore Structure and Dynamics:

Physics, Methods and AI

(2019 SPWLA Spring Beijing Workshop)

Date: April 29-30, 2019

Location: China University of Petroleum-Beijing,

18 Fuxue Road, Changping, Beijing, China, 102249

Please note: For information and invitation letter for Chinese visa application,

contact liaoguangzhi@cup.edu.cn

Direction: From Beijing Capital international airport, take taxi to changing district, Beijing

Auspicious Business hotel (北京市昌平区大宅门迎祥商务酒店). Or take subway：first take

airport shuffle to Dongzhimen station, transfer to 13 line to Xierqi station, transfer to Changing

line to Changping Dongguan station, then walk around 20 minutes to hotel.

About CUPB

China University of Petroleum, Beijing, is located in a scenic region of tourist interest in

Changping County, Beijing, about 5 miles to the Ming Tombs, 12 miles to Zhuijiuyu Scenic Spot,

15 miles to the Great Wall, and 28 miles to Yanqing UNESCO Global Geopark. Changping is

dynamic yet laid-back, culturally rich but down to earth. One can easily explore Changping on

foot or by bike with its perfect size. CUPB has a beautiful campus divided into two relatively

independent quarters, which has excellent teaching and learning facilities and well-equipped

residential areas.

Facility and Program

Registration

at Cuigong hotel.

Telephone for CUPB reception: +86-15811205235, +86-18518760349

Amenities:

Breakfast will be provided at hotels

Cuigong hotel cafeteria (12:00 PM-1:30 PM).

Format of talks:

Speakers may load their presentation onto the conference computer before their session or use

their own computer.

07:30 - 08:30

Speaker Institution Title

08:30 - 08:35 Organizer Welcome

08:35 - 09:05 Peixue Jiang Tsinghua University TBA

09:05 - 09:35DongxiaoZhang

Peking University TBA

09:35 - 09:55 Tianzhi Tang CNPC Logging Company TBA

09:55 - 10:15 Moran Wang Tsinghua University TBA

10:15 - 10:45

10:45 - 11:15 Dirk Smit Shell TBA

11:15 - 11:35 Hua WangUniversity ofElectronic Science andTechnology of China

Assessing cement leak paths byanalysis borehole wavefield modes

11:35 - 11:55 Infant RajChina University ofPetroleum

Smart fluids for fracturing

11:55 - 12:15 Jiangfeng LiuChina University ofMining and Technology

Study on Pore Structure andPermeability Prediction of GeomaterialBased on Digital Image and Machinelearning

12:15 - 13:30

13:30 - 13:50 Stefano Aime Harvard UniversityHydraulic fractures in modelheterogeneous solids

13:50 - 14:10 Maojin TanChina University ofGeosciences

Research progress in multi-sourcedata-derived intelligent logsinterpretation method ofunconventional oil and gas reservoir

14:10 - 14:30 Weitao Sun Tsinghua UniversityFluid wetting effect of rough porewall on wave velocity

14:30 - 14:50 Hongjun Xu

PetroChina ResearchInstitute of PetroleumExploration andDevelopment

Investigation on Archie ParametersDependence on Pore Structure by usingNMR Technique

14:50 - 15:10 Guoqiang Wu iRockRock typing and petrophysicalcharacteristics in complex carbonateby using digital rock analysis

Time

Tea Break

Lunch

Session 1

Chair:Lizhi Xiao

Chair:Ying Rao

Session 2

Chai

r:Yu

jun

Feng

Session 3

Digital Rock, Pore Structure and Dynamics: Physics, Methods and AI(2019 SPWLA Spring Beijing Workshop)

China University of Petroleum, 18 Fuxue Road, Changping, Beijing, China 102249

April 29, 2019

Breakfast & Registration

15:10 - 15:30 Junjian LiChina University ofPetroleum

Pore-scale investigation ofmicroscopic remaining oil variation inChina's Continental Reservoir atUltra-High Water Cut

15:30 - 16:00

16:00 - 16:20 Qingyang Lin Imperial collegeiSCAL for Complete RockCharacterization

16:20 - 16:40ThomasCochard

Harvard University Hydraulic fracture dynamics

16:40 - 16:50 Erlong YangNortheast PetroleumUniversity

Study on Microscopic Residual OilDistribution in Fractured LowPermeability Reservoirs (5 mins)

16:50 - 17:10 Linlin WangChina University ofPetroleum

Microstructural insight into thehydromechanical behavior of shales

17:10 - 17:30 Xiran XiangSouthwest PetroleumUniversity

Simulation of Conductivity of NaturalGas Hydrate Reservoir Based onPercolation Network Model

17:30 - 17:50 Jianchao CaiChina University ofGeosciences

TBA

18:00 - 20:30

07:30 - 08:30 Breakfast

Speaker Institute Title

08:30 - 09:00 David Weitz Harvard UnversityPore-scale studies of multiphase flowin porous media with opticallytransparent micromodels

09:00 - 09:20 Yujun Feng Shichuan University

“Water-in-Oil”Associative PolymerEmulsions Used in Multi-pad Frackingfor Tight Oil Reservoir: The FirstExample in China

09:20 - 09:40 Jianmeng SunChina University ofPetroleum

Progress in application of digitalrock for Logging interpretation

09:40 - 09:50 Tao Xu Northeastern UniversityTime-dependent deformation, creep andfracturing of brittle rocks

09:50 - 09:55 Yushi CuiChina University ofPetroleum

Spatially-resolved T2 distributionmapping in heterogeneous rock modelwith phase encode MRI

09:55 - 10:00 Botao LinChina University ofPetroleum

Prediction of Flowback Ratio andProduction in Shale Gas ReservoirsUsing a Neural Network Model (5 mins)

10:00 - 10:30

Session 5

Tea Break

Chair:Yujun Feng

Session 3

Chair:Guangzhi Liao

Chai

r:Yi

qiao

Song

April 30, 2019

Time

Tea Break

Welcome dinner

Session 4

10:30 - 11:00ChristophArns

The University of NewSouth Wales

A fast FFT method for rock-typing ofheterogeneous rock samples viaregional integral geometry

11:00 - 11:20 Gaetan Gerber Harvard UnversityProppants efficiency in fracturedbrittle gels

11:20 - 11:40 Langqiu SunChina University ofPetroleum

Calibration on the pore structure inCT-based rock physical studies

11:40 - 11:50 Can LiangChina University ofPetroleum

Study on Magnetic Resonance RelaxationCharacterizing Rock Wettability

11:50 - 12:00 Yang BaiChina University ofGeosciences

Total organic carbon contentprediction based on committee machinefrom wireline logs (5 mins)

12:00 - 13:30

13:30 - 14:00 Yiqiao SongSchlumberger-DollResearch

Pore connectivity by NMR

14:00 - 14:10 Sheng MaoChina University ofPetroleum

Frequency dependencies of cyclic shearfriction and shear slip of granitefracture

14:10 - 14:30 Kai LiuSouthwest PetroleumUniversity

Amperometric adsorption modelcorrection based on molecularsimulation

14:30 - 14:40 Huajun FanChina University ofPetroleum

Determination of fracture propertiesusing shock-induced Stoneley waves

14:40 - 14:50 Lipeng YanSinopec ResearchInstitute of PetroleumEngineering

Intelligent Processing Technology forDownhole Imaging Data

14:50 - 14:55 Lin HuChina University ofPetroleum

Effect of Lateral Vibration on LWD NMRT2 Distribution

14:55 - 15:00 Lin WangChina University ofPetroleum

NMR Characterization of nano selfassembled micro pore structure andpermeability

15:00 - 15:30

15:30 - 15:50 Yanbin YaoChina University ofGeosciences

Micro-fluid Dynamics in CoalbedMethane Reservoirs

15:50 - 16:10 Jiajia Zhou Beihang University Capillary imbibition in a square tube

16:10 - 16:30 Nan Li CNPC Logging CompanyThe Magnetic Field Strength Effectionon NMR T2 of Rocks

16:30 - 16:50 Xu PangPetrochina HangzhouResearch Institute ofGeology

Application of ECS Logging inLithology Identification of Pre-saltIgneous Rocks in Santos Basin, Brazil

16:50 - 17:10 Yan ZhangChina University ofPetroleum

Localized Laplace NMR for porousmaterials

17:10 - 17:15 Yang XiaChina University ofPetroleum

Dynamics of Multi-Continuum/Discrete-Fracture Interactions of Nonlinear GasTransport in Shale(5 mins)

17:15 - 17:35 Shaowu GaoDassault SystemesShanghai

SIMULIA digital rock in Dassault

17:35 - 17:55 Feng WuSouthwest PetroleumUniversity

Resistivity Simulation and WaterSaturation Calculation of Band ShapedShaly Sandstone

18:00 -

Session 7

The End

Lunch

Session 6

Session 8

Chair:Ruina Xu

Chair:Shouceng Tian

Chai

r:Bi

ng W

ang

Tea Break

Pore connectivity by NMR-detected capillary pressure experiment

Yi-Qiao Song1, Andre Souza1,2 & Muthusamy Vembusubramanian1, & Yiqiao Tang1

1Schlumberger-Doll Research, One Hampshire Street, Cambridge, MA 02139, USA; 2Schlumberger Brazil Technology Integration Center,

Avenida Republica do Chile, 330, 20031-170, Rio de Janeiro, Brazil

Porous media is ubiquitous in nature and porous sedimentary rocks are where energy

(petroleum), water and minerals are extracted in order to support the contemporary

life style of the world. They are complex materials of tortuous pore structures with a

wide range of pore sizes and connectivity owing to geological processes. In particular,

pore connectivity is critical to fluid flow in soil and rocks, however, our

understanding remains simplistic (such as bundle-of-tubes) due to inadequate

measurement techniques. This paper reports a method to decipher how pores are

connected by measuring a pore size-throat correlation map using an NMR-detected

capillary pressure experiment. Our data show that such map can identify a wide

variety of pore connectivity types due to diagenesis processes. This method could be

useful for petroleum engineering, understanding of diagenesis, soils and contaminant

dispersion and the study of other engineering porous materials.

A fast FFT method for rock-typing of heterogeneous rock

samples via regional integral geometry

Han Jiang and Christoph H. Arns

The University of New South Wales, Sydney, Australia

Abstract：The upscaling of pore-scale digital core measurements of heterogeneous

samples to larger scales relevant for the interpretation of well-logging responses

requires the extraction of pore-scale categorical variables and their spatial distribution,

followed by a characterization of the individual categories, typically combined with a

hierarchical tomographic imaging approach.

In earlier work we illustrated the discriminative power of integral geometric measures,

namely the Minkowski functionals, in separating micro-structures originating from

similar generating functions or processes. Furthermore, effective physical properties

for a Boolean process were directly predicted using an effective grain shape.

Combined this suggests that integral geometry may provide an excellent basis to

define different categorical variables on the basis of morphology, and which have well

separated physical properties.

A difficulty in applying this technique is that regional measures should be defined

over significant volumes, such that the Minkowski measures are stable and lead to a

robust classification. This poses a significant computational problem. In this work we

introduce an FFT approach to deriving the local measures and thus making this

pore-scale rock-typing approach practical.

The techniques are demonstrated for a set of Boolean composites and the speed-up

over traditional approaches demonstrated. We furthermore apply the technique to a

thinly laminated sandstone.

Kewords：Integral geometry, micro-CT, rock-typing, regional measures

Research on the pore structure and permeability prediction of

porous media based on digital image processing and machine

learning techniques

Jiang-Feng Liu1*

, Xu-Lou Cao1, Xu Chen, Yang-Guang Wang

1

（1. The State Key Laboratory for GeoMechanics and Deep Underground

Engineering, and School of Mechanics and Civil Engineering, China University of

Mining and Technology, Xuzhou 221116, China）

The pore structure of compacted GMZ bentonite was observed by scanning electron

microscopy (SEM). The porosity was firstly obtained by NMR and MIP, and the

appropriate algorithm (Yen algorithm) was determined by inverse analysis. The

grayscale threshold of the image is determined based on the Yen algorithm, and the

image is further binarized. Further, the pore structure is quantitatively characterized.

The porosity is calculated and its pore size distribution is obtained by discrete and

continuous algorithms, respectively. Based on Hagen-Poiseuille's law and Darcy's law,

a prediction model of permeability is established. Then, the previously extracted pore

parameters are brought into the prediction model to calculate the permeability.

Research indicate that the selected image resolution and image size have a greater

impact on the permeability prediction results. The larger the magnification (the higher

the resolution), the more accurate the prediction of the permeability: 1.5×10-15

m2

(5000×) vs. 1.0×10-15

m2

(gas permeability). Further, based on the FIB/SEM 3D

reconstruction model, and considering the connectivity and tortuosity in the Z

direction, the K-Z and K-T models were combined to calculate the permeability of

GMZ bentonite. The results show that the calculated results are quite different from

those of gas permeability test, but close to those of ethanol test: 0.20×10-18

m2 (K-T

model) vs.0.14×10-17

m2

(K-C model) vs. 4.6×10-17

m2

(ethanol permeability) vs.

1.0×10-15

m2

(gas permeability). The difference between the seepage characteristics of

different fluids stems from the nature of the fluid itself, as well as the interaction

between the fluid and the solid particles. Finally, we develop a digital image feature

recognition and extraction system based on neural network technology, which can

accurately predict the permeability.

Key words: GMZ bentonite; pore structure; permeability; FIB/SEM; K-Z/K-T

model

Fluid wetting effect of rough pore wall on wave velocity of

partially saturated rock

Weitao Sun

( Zhou Pei-Yuan Center for Applied Mathematics, Tsinghua University, Beijing,

100084, China）

The stiffness of the rock skeleton is proportional to the contact strength between the

mineral particles. The contact strength is dependent on the surface energy, which will

be reduced when mineral particles such as quartz absorb fluid molecules. As a

consequence, the skeleton stiffness will decrease when wetting fluid film is formed on

pore wall. This weakening of the rock skeleton stiffness is related to the chemical type

of fluid. The surface energy of oil saturation is much greater than that of water

saturation. Therefore, when the oil saturation is close to 100%, the rock stiffness is

strengthened and the wave velocity rises. The increase of the velocity brought by the

surface energy effect even exceeds the decrease of the velocity caused by the decrease

of water saturation. Therefore, it is often observed in the experiment that there is a

phenomenon of "velocity upwarping" near zero water saturation on Vp-saturation

curve.

When water saturation exceeds a certain threshold, this "velocity ramp up" gradually

disappears, and we call this saturation threshold "maximum relaxation saturation".

This phenomenon of "velocity upwarping" can hardly be explained by conventional

methods, which brings challenges to conventional rock physics models. Therefore, it

is necessary to propose a new velocity prediction technique for partially saturated

fluid based on surface energy effect, which is of great significance for seismic wave

exploration and development of unconventional oil and gas resources.

Kewords: Partially saturated rock, rough pore wall, wetting fluid film, P wave

velocity

Prediction of Flowback Ratio and Production in Shale Gas

Reservoirs Using a Neural Network Model

Botao Lin

(College of Petroleum Engineering, China University of Petroleum, Beijing, 102249,

China）

The flowback management is an important step in fracturing practices of shale gas

reservoirs because a good handle of it not only preserves the conductivity of the flow

paths created by the fracture network but also assists the subsequent production by

minimizing the damage of the fracturing fluid to the fluid-sensitive formation. A

quantitative prediction of its behavior and how it will affect the productivity of a

reservoir, however, are not available yet if accounting for a large group of reservoir

properties and engineering features. This study attempts a mathematical approach to

predict the flowback behavior concerning flow back ratio, and the productivity

represented by first-month production. First, a BP neural network model will be

established to filter out the controlling factors given the values of actual flow back

ratio from the geological properties and engineering characteristics of the sample

wells, which are used to generate a geological index g and an engineering index e.

Secondly, a correlation is generated between the flow back ratio and the gas well as e

by non-linear fitting. In subsequence, the first-month production will be predicted

based on the field recorded flow back ratio and the comprehensive index c, the latter

of which combines the influences of both geological and engineering parameters.

Finally, the stimulated reservoir volumes (SRVs) of several wells will be estimated

using the K-means clustering and Delaunay triangulation to assess their impact on the

flow back ratio and the first-month production data. Meanwhile, there exists an

optimum flow back ratio range for the shale gas region of concern, implying that

drilling paths and fracturing designs can be optimized to achieve that flow back ratio

range so that the productivity of the formation can be maximized. In general, the

proposed approach can be used in shale gas reservoirs to examine the favorable flow

back ratio and guide the engineering designs to enhance production, as long as

adequate reservoir data are accessible.

Microstructural insight into the hydromechanical behavior of shales

Linlin Wang*

1 College of Petroleum Engineering, China University of Petroleum (Beijing), Beijing

102249, China

Hydrocarbon production from unconventional shale-gas reservoirs has increased

dramatically in the last decades; a good understanding of the hydro-mechanical

behavior of the involved rock is of crucial importance. However, characterization of

shales is challenging: these rocks exhibit complex coupled thermo-hydro-chemo

-mechanical behavior and a multi-heterogeneity, the description of which would

strongly benefit from an improved experimental insight on their deformation and

damage mechanisms at micro-scale. We propose here an experimental method for

micro-scale characterization, consisting of in situ tests within the chamber of an

environmental scanning electron microscope (ESEM), and quantification at

micrometric scale of the induced local strains by the combination of high resolution

imaging and digital image correlation techniques (DIC). Moreover, thanks to an

improved surface preparation using ion beam polishing method, the clay particles as

well as secondary inclusions can be observed in comparison with the conventional

mechanical polishing method. On the basis of this method, the hydro-mechanical

behavior of shales can be investigated at their inclusion-matrix- composite

microstructure: this scale is of particular interest because the complex

matrix-inclusion interactions are a key mechanisms governing the deformation and

damage of such rocks. We present some recent observations of the evolution of the

rocks when subjected to hydric and mechanical loads, in particular the key role of the

microstructure on the macroscopic behavior of such rocks.

The Magnetic Field Strength Effection on NMR T2 of Rocks

Li Nan1,2

, Li Xin1,2

, Ge Xinmin2,3

, Cao Xianjun1,2

, Wu Di1,2

, Han Bo1,2

(1 CNPC Logging Company Limited Technical Center, Xi’an 710077, China

2 The Key Laboratory On Logging Of CNPC, Xi’an 710077, China

3 China University Of petroleum, Qingdao 266580, China)

In the field of petroleum exploration, NMR technology is an effective method to

obtain underground fluid information by observing the resonance phenomenon of

hydrogen nuclei under the action of external magnetic field. At present, scholars

mainly focus on NMR acquisition sequences, data processing and methods, but few

scholars pay attention to the influence of external magnetic field strength on NMR

results. In the process of NMR acquisition, the magnetic field strength is proportional

to the NMR frequency. In this paper, the author used the 2MHz, 12MHz and 23MHz

NMR instruments to measure the NMR T2 of the fluid and three types of core samples

which include sandstone, carbonate and volcanic rocks, and used the magnetic

material analyzer to analyze the magnetic substance content of these core samples.

The author finally analyzed the influence of magnetic field strength on core NMR T2.

The experimental results show that: ① the higher NMR frequency, the higher

signal-to-noise ratio acquired; ②as the NMR frequency increases, the NMR porosity

of the same sandstone sample decreases continuously. And the smaller main peak of

the T2 spectrum, the greater influence of frequency; ③the change of NMR frequency

has little effect on the NMR T2 of carbonate rock core with different pore structure；④

the high frequency of NMR has a great influence on the volcanic rocks. The higher

NMR frequency, the smaller NMR porosity of the same volcanic sample, and the

higher magnetic content, the greater effect of frequency changes. The results have

certain guiding significance for core NMR experiment and downhole NMR

acquisition: for carbonate and high porosity sandstone reservoirs, instruments with

higher frequency can be selected to improve the acquisition signal-to-noise ratio and

efficiency; for dense sandstone reservoirs, the low-frequency NMR equipment should

be used if the signal-to-noise ratio meets the requirements; for volcanic reservoirs,

NMR technology is not applicable.

Key word: NMR, Magnetic Field, Frequency, Core, T2

Application of ECS Logging Technology in Lithology

Identification of Pre-salt Igneous Rocks in Santos Basin,

Brazil

Pang Xu1, Wang Hongping

1, Yangliu

1, Wangchaofeng

1, Li Xiaohui

2

(1. Petrochina Hangzhou Research Institute of Geology, Zhejiang Hangzhou,

310023, China

(2. CNPC logging, Xian Shanxi, 710077, China)

Pre-salt igneous rocks are complex and widely developed in Santos Basin, Brazil,

which are the main factors affecting reservoir development and CO2 content.

There are two types of igneous under salt: eruptive rock and intrusive rock. It is

difficult to identify igneous rock lithology with conventional logging data and the

recognition coincidence is low. According core, sidewall core and thin section

identification, the main igneous rock types in the study area are analyzed and

studied. A new lithology indicator curve is constructed and lithology is

determined by using ECS logging and GR Spectrum logging by crossplot and

statistical method. The results of logging identification are extended to

conventional logging, lithology sensitivity curve analysis and neutron-density

crossplot are carried out. The results of this study have been applied in Santos

Basin with good results, providing reliable parameters and strong support for

petroleum exploration and development, reservoir prediction, reservoir physical

properties.

Key words: Santos Basin; pre-salt; igneous rocks; ECS logging; lithology

identification; GR Spectrum logging

Capillary imbibition in a square tube

Tian Yu 1，Jiajia Zhou

1,2，Masao Doi1

(1. Center of Soft Matter Physics and its Applications, Beihang University; 2. School

of Chemistry, Beihang University)

Wetting and evaporation in porous media show similarities to the same process in an

individual channel. Here we study the wetting of a capillary with a square

cross-section. When a square tube is brought in contact with bulk liquid, the liquid

wets the corners of the tube and creates finger-like wetted regions. The wetting of the

liquid then takes place with the growth of two parts, the bulk part where the cross

section is entirely filled with the liquid and the finger part where the cross section of

the tube is partially filled. In previous works, the growth of these two parts has been

discussed separately. Here we conduct the analysis by explicitly accounting for the

coupling of the two parts. We propose coupled equations for liquid imbibition in both

parts and show that the lengths both increase in time following Lucas–Washburn’s law,

but the coefficients are different from those obtained in the previous analysis that

ignored the coupling.

Kewords：capillary imbibition, wetting.

Hydraulic fracture dynamics

T. Cochard1,3, Y. Song2, L. Xiao3, D.A. Weitz1

1 Experimental Soft Condensed Matter Group, Harvard SEAS, Cambridge, MA, USA

2 Schlumberger-Doll Research, Cambridge, MA, USA

3 China University of Petroleum Beijing (CUPB), Beijing, China

Hydraulic fracking of shale reservoirs is a process where a highly pressurized

fluid is injected into the shale matrix, inducing fractures that create a connected path

between the pores filled with hydrocarbons. The dynamics of hydraulic fractures in

high strength materials remain poorly understood, despite the widespread application

of hydraulic fracking. Particularly related to mitigation of seismicity and aquifer

pollution.

Most of the research on hydraulic fracking is focused on the evolution of the

pressure during the fracture propagation. In contrast, in our work, we are developing

an energy balance approach that accounts for both the compressed volume and

pressure of the injected fluid. Importantly, by only considering the pressure evolution

within the system, the total volume of the injected fluid is neglected. Our results show

that the volume of fluid cannot be neglected as the applied pressure upon fracture is at

least 350 times the atmospheric pressure. To accomplish that, we designed a

3D-printed cube using a stereolithography apparatus with an injection system

allowing us to initiate a systematic controlled hydraulic fracture. We use high speed

imaging to follow the fracture propagation and measure the pressure evolution and the

volume of injected fluid being compressed.

As a result, we are able to demonstrate that the desired volume of fractures can

be triggered by the total volume of fluid used during the fracking process. In addition,

we find that by using an energetic approach, we can estimate mechanical parameters

in situ such as the fracture toughness of the material.

iSCAL for Complete Rock Characterization: Using Pore-Scale

Imaging to Determine Relative Permeability, Interfacial

Curvature, Capillary Pressure and Contact Angle

Qingyang Lin, Branko Bijeljic, Martin Blunt

Department of Earth Science and Engineering, Imperial College London, London,

SW7 2AZ, United Kingdom

We propose a new experimental-driven Digital Rock workflow, which we term

iSCAL (Special Core Analysis with Imaging), to combine measurements of

multiphase flow properties with pore-scale imaging. We apply this workflow to a

mixed-wet Bentheimer sandstone. After prolonged contact with crude oil to alter the

surface wettability, a refined oil and formation brine were injected at a fixed total flow

rate at low capillary number but in a sequence of increasing brine fractional flows.

X-ray tomographic images were taken at steady state when the pressure across the

system stabilized. The images were used to compute and obtain porosity including

micro-porosity, saturation, interfacial area, curvature and contact angle. Relative

permeability and capillary pressure were determined as functions of saturation.

We compared the results with a previously published experiment with strongly

water-wet conditions. The oil was connected in layers in the mixed-wet system with

oil relative permeability lower than the water-wet case. The residual oil saturation was

low and was approximately 0.11. The capillary pressure was slightly negative and ten

times lower than the water-wet case, covering a wide range of intermediate saturation.

In particular, we observed that the oil-brine interfaces were not flat, but had two

approximately equal, but opposite, curvatures in orthogonal directions. These

interfaces were approximately minimal surfaces which allow efficient displacement

and imply well-connected phases and better recovery. The average contact angle for

the mixed-wet system was about 8017 from direct measurement from the images.

We also used the principle of energy conservation to derive a

thermodynamically-consistent contact angle, which was approximately 944.

We suggest that the iSCAL workflow can provide a complete rock characterization

and is a compelling complement to traditional petrophysical measurements, which can

lead to a new generation of core analysis. The characterization can further enable the

design of optimal surface recovery.

“Water-in-Oil” Associative Polymer Emulsions Used in

Multi-pad Fracking for Tight Oil Reservoir: The First Example

in China

Yujun Feng 1,2，Yongli Lv

1,3，Sheng Zhang 1

( 1. Sichuan University; 2. Research Institute of Petroleum Engineering, Shengli

Oilfield Co., Sinopec; 3. Shengli Oilfield Shengli Chemicals Co., Ltd.）

Multi-pad hydraulic fracturing is believed a cost-effective procedure to unlock the

tight oil from low-porosity, low-permeability reservoirs. However, the inconvenience

of difficult-dissolving process at surface and crosslinking of the conventional

guar-based fracturing fluid systems cannot satisfy such fracking jobs because of the

massive proppant loading, high flow rate and large volume of the fluids used.

To address these issues, a crosslinking-free and rapid-dissolution fracturing fluid

system based on synthetic hydrophobically associating polymer (HAP) “water-in-oil”

emulsion was developed. The HAPs are derived from classical water-soluble

polymers by incorporating small amount of long hydrophobic side chains onto the

polymer backbone. When above a critical associating concentration, these polymers

can automatically form a three-dimensional transient network by intermolecular

association, reminiscent of cross-linked structures, offering the suspending capacity

for proppants. With inverse emulsion polymerization, the obtained HAP emulsions

can not only get high molecular weight, but also be rapidly dispersed and finally

dissolved within 5 minutes.

It was found concentrated HAP polymer emulsions can be dispersed online with

surface water or even produced fluids to get final designed concentration. Laboratory

rheological study shows that 1% of the as-prepared fracturing fluid can reach more

than 50 mPas at 150 C. Compared with guar-based fluid, the HAP fracturing fluid

can be completely broken, and the viscosity, surface tension, skin damage of the

residual fluid on the permeability are all smaller, while the fluid loss is comparable,

proppant-carrying ability is even better. Most importantly, no further surfactant was

needed to assist the flowback the fluid.

Since September 2013, such associative polymer fracturing fluids were successively

used in 29 wells of 3 well pads, Yan-227, Yan-22 and Bin-37 blocks in Shengli

Oilfield, Sinopec, where the temperature ranges from 110 to 145 C. Totally 60,000

m3 fluids were consumed in these fracking jobs, and 87, 9, and 45 stages were

successively fractured in the horizontal sections, respectively.

Kewords：Hydrofracking; Tight oil; Fracturing fluids; Multi-pad fracking

Dynamics of Multi-Continuum/Discrete-Fracture Interactions of

Nonlinear Gas Transport in Shale

Yang Xia1

1 College of Petroleum Engineering, China University of Petroleum (Beijing), Beijing

102249, China

Advances in multistage fracturing combined with horizontal drilling have made this

technique the driving force for the recent spectacular success in shale gas

development and production. In this paper, a hierarchical approach integrating

discrete fracture networks with multi-continuum concept is proposed to model various

coupling mechanisms of gas nonlinear transport in shale. The hybrid model is

composed of three continuum layers: organic matter, inorganic matter and

micro-fractures in matrix which are treated as a continuum medium, and the discrete

fractures are embedded into the micro-fracture-continuum. The extended finite

element method is employed to decouple the mesh conformity between the mesh of

media and the discrete fractures. The results of long-term well-performance dynamics

are used to show the power and flexibility for simulating the

multi-continuum/discrete-fracture interactions during gas transport in shale. The

effects of fracture geometry and coupling flow mechanisms on gas production are also

analyzed in this paper.

Assessing cement leak paths by analysis borehole wavefield modes

Hua Wang 1，Michael Fehler

2，Aimé Fournier2

( 1.School of Resources and Environment, University of Electronic Science and Technology of

China; 2. Earth Resources Laboratory, Massachusetts Institute of Technology）

Abstract：Evaluation of possible cement leakage pathways is essential for successful

Carbon Capture and Storage (CCS), geothermal engineering, groundwater

development, and oil/gas development. A channel in the borehole cement, which

secures the borehole casing to the formation, may allow fluid to escape. Risk

assessment and remediation decisions about the presence of such channels depend on

channel parameters: radial position from the center of the borehole r; channel

thickness d; azimuthal position of the channel φ; and azimuthal extent of the channel

θ. Conditional cement bond log, which uses only the first arrival at a centralized

borehole receiver, cannot diagnose details of cement leak channels. To accurately

characterize the possible cement leak paths, we use a 3-dimensional finite-difference

method to investigate the use of the abundant data collected by a modernized

monopole sonic tool that contains an array of azimuthally distributed receivers in a

singly-cased borehole with different bonding conditions. We also investigate how to

improve the tool design to acquire even more useful information. We investigate

various receiver geometries, multi-modal analyses of multi-frequency data to discover

the type of logging tool that provides the best information for cement bond evaluation.

Modal dispersion curves and dispersion analysis facilitate the identification of

propagation modes.

We find that the casing modes are strong when interface I (interface between casing

and cement) is partially or fully replaced with fluid. The amplitude dependence on

fluid thickness is small which could lead to ambiguity in interpretation. The casing

modes are different when interface II (interface between cement and formation) is

partially replaced with fluid, because the modes propagate in the mixed material of

steel pipe and cement and the velocities are highly dependent on the cement thickness.

It would highly possibly misjudge cement quality because the amplitudes of these

modes are very small and they propagate with nearly the formation P velocity.

However, it is possible to use the amplitude to estimate the thickness of the cement

sheath because the variation of amplitude with thickness is very clear. We find that an

appropriate choice of wave modes, source frequencies, source polarities, and receiver

locations and offsets provide sensitivity to d, φ, θ. The amplitude of the first arrival

from a monopole source is sensitive to θ. Amplitudes at receivers at different

azimuths are sensitive to φ. While the Stoneley mode (ST1) propagates in the

borehole fluid, a slow Stoneley mode (ST2) appears in the fluid column outside the

casing when cement is partially or fully replaced with fluid. The slow Stoneley mode

(ST2) velocity is sensitive to d, but ST2 is not easy to pick when θ and d are small.

Further improvement is necessary to provide comprehensive information about

possible flow channels in casing cement. Machine learning methods are also

anticipated to be used to pick the specified modes and to calculate dispersion curves

from field data, which will speed up the data processing in field.

Time-dependent deformation, creep and fracturing of brittle

rocks

Tao Xu

(Center for Rock Instability and Seismicity Research, Northeastern University,

Shenyang 110819, China)

Abstract: A numerical meso-scale model for brittle creep considering pore pressure

effect is proposed to describe the time-dependent deformation of heterogeneous brittle

rock under loading conditions at different constant pore pressures. The model

accounts for material heterogeneity and local material degradation properties at a

meso-scale to capture the co-operative interaction between microcracks in the

transition from distributed to localized damage. It describes the spatiotemporal

evolution of acoustic emissions in the rock during the progressive damage process.

The model is validated against experimental data and is used to simulate brittle creep

tests of heterogeneous rock samples under varying constant pore pressures, applied

axial stresses and confining pressures. Our model accurately reproduces the classic

creep behavior observed in laboratory brittle creep experiments. The simulations also

show evidence of a ‘critical level of damage’ before the onset of tertiary creep and

that the initial stages of localization can be seen as early as the start of the secondary

creep phase. Our approach differs from previous macroscopic approaches based on

constitutive laws, and microscopic approaches that focus on fracture propagation. The

model shows that complex macroscopic time-dependent behaviour can be explained

by the small-scale interaction of elements. The fact that the simulations are able to

capture a similar hydro-mechanical time-dependent response of heterogeneous brittle

rocks to that seen in laboratory implies that the model is appropriate to investigate the

non-linear complicated time-dependent behavior of heterogeneous brittle rocks under

coupled hydro-mechanical loading.

Keywords: Brittle creep, time-dependent deformation, fracturing, heterogeneity

Localized Laplace NMR for porous materials

Yan Zhanga, Lizhi Xiao

a,b, Guangzhi Liao

a

a State Key Laboratory of Petroleum Resources and Prospecting,

China University of Petroleum, Beijing, 102249, China b Harvard SEAS-CUPB Joint Laboratory on Petroleum Science, Cambridge, MA

02138, USA

Laplace Nuclear Magnetic Resonance (LNMR) is a powerful method to

characterize porous media [1, 2]. For instance, the D-T2 correlation experiment can be

used to distinguish oil and water in sandstones [3]. Diffusion - diffusion correlation

method was performed to extracted anisotropy information by applying magnetic field

gradients with different directions [4]. However, these multi-dimensional Laplace

methods only obtain integral information about sample, which means the detailed

information in local region will be lost for heterogeneous materials.

Spatially resolved NMR can acquire local details with the help of MRI

experiments. Thus combination of MRI and Laplace NMR is desirable to obtain pore

structure and diffusive fluid information in local part. This is quite suitable for highly

heterogeneous sample such as sedimentary rocks. In our research, back-projection

imaging method was used to acquired local information to implemented T2 and MRI

experiments at the same time, which will save experimental time. To further

accelerate the experiment, compressed sensing was used to subsample the data and

the reconstruction results show good accuracy. The experimental results show that

these methods reveal detailed heterogeneous information for porous media and are

expected to be useful in many areas such as oil industry.

Reference

[1] Y. Zhang, B. Blümich, J. Magn. Reson. 242 (2014) 41-48.

[2] Y. Zhang, B. Blümich, J. Magn. Reson. 252 (2015) 176-186.

[3] M.D. Hürlimann, L. Venkataramanan, C. Flaum, J. Chem. Phys. 117 (2002)

10223–10232.

[4] B. Sun, K.-J. Dunn, A global inversion method for multi-dimensional NMR

logging, J. Magn. Reson. 172 (2005) 152-160.

[5] M. Lustig, D. Donoho, J. M. Pauly, Magn. Reson. Med. 58 (2007) 1182-1195.

[6] S. F. Roohi, D. Zonoobi, A. A. Kassim, J. L. Jaremko, Pattern Recognition 63

(2017) 667-679.

Investigation on Archie Parameters

Dependence on Pore Structure by using NMR Technique

Hongjun Xu1，Falong Hu

1，Chaoliu Li1，Changxi Li, Jun Yu, Chunmei Yang

PetroChina Research Institute of Petroleum Exploration and Development, Beijing

100083

Abstract: A new method based on non-resistivity, NMR, is introduced for the

estimation of Archie equation parameters F and n. Archie equation is a function of

rock pore structure which includes porosity, pore size distribution, and pore throat

size. It has been demonstrated that NMR relaxation measurements provide

information on rock porosity, permeability, pore size distribution, and in some cases,

fluid types. Improvements in NMR have made it possible to be a measure of Archie

equation parameters. In this paper, we have demonstrated that formation factor is

related to the NMR logarithmic mean (T2lm) of rocks at fully brine saturated

condition and the saturation exponent is related to the ratio of T2lm of rocks at

fully-saturated condition to T2lm of rocks at different partially saturations. Based on

the main control factor of rock pore structure, we establish new models to determine

Archie equation parameters from NMR perspective. The new models are verified on a

series of core plugs, and results confirmed that Archie parameters with new models

are accurate and available. To demonstrate the validity and possibilities of new

methods, we also present a field case using the real log data.

Key words：Nuclear magnetic resonance; T2lm; Archie equation; MICP; Pore structure;

Formation factor; Saturation Exponent.

Experimental investigation of proppants efficiency in fractured

brittle gels.

G. Gerbera,b

, E. Julienb, T. Cochard

b,c, W. Steinhardt

b, P. Coussot

a, L. Xiao

c, D. A.

Weitzb

a

Université Paris-Est, Laboratoire Navier (ENPC-IFSTTAR-CNRS), Champs sur

Marne, France b

Experimental Soft Condensed Matter Group, School of Engineering and Applied

Sciences, Harvard University, Cambridge, MA, USA c China University of Petroleum Beijing (CUPB), Beijing, China

During the hydraulic fracturing process, newly created fractures collapse when the

stress accumulated in the surrounding material overcomes the external fluid pressure.

This phenomenon drastically reduces the permeability of the newly fractured matrix

and prevents the recovery of oil and gas trapped in reservoirs. Since their first use in the

80s, proppants (solid particles added to the fracking fluid) have been a key solution to

this problem, by propagating with the fracking fluid in the fractures and limiting their

collapse. Extensive studies have highlighted the importance of the shape and toughness

of proppants to prevent their mechanical failure (crushing) when squeezed by the

fracture. Still, the particle transport within the fracking fluid and their structuration

during collapse remains poorly understood.

Here, we create transparent brittle gels as a model fracking medium and investigate the

dynamics of model proppants (spherical monodisperse beads) as the fracture evolves.

We record the spatial and temporal distribution of the particles by direct imaging, as

well as the evolution of the pressure over time. That allows us to better understand the

fundamentals of proppant displacement during initiation, proppant transport during

propagation, and proppant flow-back and/or accumulation during collapse. We show

that most aspects of these processes are driven by (i) the initial and boundary conditions

of the fracture and (ii) the balance between settling, drag and surface friction forces. In

the end, this study will allow a better identification of the best conditions for optimal

propping efficiency and, in fine, maximum fracture aperture and recovery.

Determination of fracture properties using shock-induced

Stoneley waves

Huajun Fan1,2

, David Smeulders2,3

1, China University of Petroleum-Beijing

2, Delft University of Technology

3, Eindhoven University of Technology

Fractures are of major importance for the productivity of hydrocarbon reservoirs. The

aperture, extension and distribution of fractures are essential for reservoir evaluation

and characterization. Stoneley waves can be used to detect and measure fractures. A

theoretical description for Stoneley wave propagation in the borehole intersected by a

single horizontal fracture provides a quantitative prediction of velocity, attenuation,

reflection and transmission of borehole Stoneley waves. The borehole Stoneley wave

propagation is investigated by means of a so-called shock tube facility. The facility

generates shock-induced Stoneley waves in a broad frequency band under conditioned

circumstances, with excellent repeatability. A water-saturated cylindrical rock sample

which has a centralized borehole is mounted at the test section of the shock tube. A

logging probe is installed inside the borehole thus the pressure profiles in the borehole

can be recorded. The experiments show that recorded Stoneley wave pressure signal

at fixed depth is altered significantly by the change of fracture apertures. The

reflection and transmission of borehole Stoneley waves over the horizontal fractures

are well predicted by theory. This research is directly applicable to fractured reservoir

core samples.

Amperometric adsorption model correction based on molecular

simulation

Kai Liu1

(1.School of Geosciences and Technology, Southwest Petroleum University, Sichuan,

Chengdu 610500)

Isothermal adsorption experiments are currently the most commonly used method in

the study of shale adsorption performance. For shale gas, it mainly exists in the form

of adsorption and free state: free gas is present in nano-scale pores and micro-cracks,

and adsorbed gas is present in organic matter and clay mineral pores. As an important

part of shale gas, adsorbed gas accounts for 20%-85% of the total shale gas. Therefore,

the development of shale gas depends largely on the evaluation of shale gas. The

characteristics of shale reservoirs and fluids under high temperature and high pressure

are special. The reservoirs are mainly nanopores, with various pore sizes and complex

mineral components. The fluids are mainly methane gas and exist in supercritical state.

Some experiments have found that the partial isothermal adsorption curve measured

in the simulated formation environment shows the phenomenon of "inverted

adsorption" which increases first and then decreases with the increase of pressure. The

Langmuir adsorption model cannot directly fit the relevant experimental data.

Based on classical molecular dynamics theory and method, this paper firstly simulates

the adsorption phase density of methane in different pore sizes of organic matter,

calculates the adsorption phase density distribution of methane under different

conditions. And then, based on the single-layer molecular adsorption theory, the

excess adsorption amount is converted into absolute adsorption amount. The method

is used to calculate the actual shale gas supercritical adsorption isotherm. It can be

seen that the difference between the excess adsorption amount and the absolute

adsorption amount under the methane reservoir pressure has more than doubled.

Therefore, the absolute adsorption amount in the actual reserve evaluation can no

longer be replaced by experimental data, which will cause a large error in the

estimation of the reservoir production.

Molecular dynamics simulation calculation is based on the simulation calculation of

Newton's equation of motion, sample extraction of different states in the system, and

then calculate the integral of the configuration system, and calculate the

thermodynamic properties of the simulation system with the system integration result

as the initial state. Other macroscopic properties

About the author: Liu Kai, a master's degree student; research direction: geophysical logging. E-mail:

307800135@qq.com

Rock typing and petrophysical characteristics in complex

carbonate by using digital rock analysis

- An example from Umm Gudair Field, Kuwait

Guoqiang Wu

1,，Sven Roth 1，Nasiru Idowu

1

( 1. iRock Technologies）

Abstract: Umm Gudair field is a giant carbonate reservoir in West Kuwait with over

45 years of production. Even though plenty of data is available, reservoir

characterization of Minagish Oolite formation, the main reservoir of Umm Gudair

field, has been a challenge due to the strong heterogeneity. Based on the CT image of

337 ft fresh whole-core and 66 core plugs, 7 types of rocks are distinguished honoring

geological aspects and petrophysical data. The study also captured megascopic vuggy

pores and potential thin flow barriers, which are difficult to capture through

conventional log and laboratory data. Static properties including porosity and pore

size distribution and dynamic properties including permeability and two-phase fluid

flow data of each type of rocks are simulated based on digital rock analysis. All the

results are delivered with high quality in a relative short time and can used for

geological model and reservoir simulation. This study proves that digital rock analysis

is a quick and effective method in improving reservoir characterization.

Kewords: Rock typing; Digital rock analysis; Complex carbonate reservoir

Research progress in multi-source data-derived intelligent logs interpretation method of

unconventional oil and gas reservoir

Maojin Tan1,Yang Bai

1,Qian Wang

1,Yujiang Shi

2,Gaoren Li

2,Jing Wu

1,Xiuping Wei

3

(1. School of Geophysics and Information Technology of China University of

Geosciences, Beijing; 2. China Petroleum Changqing Oilfield Exploration and

Development Research Institute; 3. Sinopec Petroleum Exploration and

Development Research Institute)

Abstract

Geophysical logging technology can continuous and accurate in-situ

petrophysical parameters such as electrical, acoustic, nuclear, and nuclear magnetic

resonance information for oil or gas reservoir evaluation, which like an "eye"

detecting the formation. How to extract reservoir information from these parameters is

the main task of logs interpretation and formation evaluation. In organic shale and

tight sandstone reservoirs, fluid occurrence forms and mineral composition are both

complex, the previous method are not suitable for unconventional oil and gas

reservoir. Therefore, we must explore new logs interpretation methods.

For organic shale, total organic carbon (TOC) content is important for reservoir

hydrocarbon generation capacity evaluation. We propose a method based on radial

base function (RBF) neural network for calculating total organic carbon content.

Through investigating the sensitivity relationship between logging data and TOC, we

optimized the sensitive logging data and compared the predictions results of different

sensitive inputs. The research results show that the predicted results from the RBF

method is in good agreement with core experiment, and it is better than the ΔlogR

method. Aiming at the mineral content prediction of organic shale problem, we study

the optimized logs interpretation method of organic shale, including pattern search

method, genetic algorithm, and simulated annealing method. The logs interpretation

method based on the RBF interpolation method, namely the two-dimensional RBF

method, was also studied. These optimization method above can predict various

mineral contents such as clay, quartz, calcite, pyrite, and kerogen. The prediction

accuracy of the algorithms is significantly improved.

In tight sandstone reservoirs, the porosity and permeability is low, and the

logging responses is not sensitive to the fluids. It is difficult to identify the fluid types

from well logging response. The committee machine (CM) is a compound intelligent

algorithm developed in recent years. To identify fluid of tight sandstone, we selected

the BP neural network, probabilistic neural network (PNN) and decision tree classifier

to constitute the classification committee machine, and applied the voting method to

make the final fluid identification. To predict parameters of tight sandstone, we select

the BP neural network, the extreme learning machine (ELM) and wavelet neural

network (WNN) to constitute the regression committee machine, and used the

weighted average method to output the final reservoir parameters. Some case studies

show that the committee machine can combine these individual intelligent algorithms

together through a good decision-making mechanism, which is superior to each

individual intelligent algorithm. Moreover, the committee machine can effectively

avoid overfitting and falling into local minimum. Therefore, the committee machine

method is more advanced and the predicted results are more accurate.

The model-based logs interpretation methods are difficult for unconventional oil

and gas reservoir such as organic shale and tight sandstone. Through multi-source

data including logging data, testing results, and petrophysical experimental data, the

machine learning algorithm can effectively realize feature extraction and quantitative

evaluation of reservoirs, which is an inevitable trend of logs interpretation

development.

ACKNOWLEDGEMENTS: This work presented is sponsored by National Natural

Science Foundation of China (41774144, U1403191), National Major Projects

"Development of Major Oil& Gas Fields and Coal Bed Methane(2016ZX05014-001)”

Keywords: Multi-source data; organic shale; tight sandstone; machine learning;

committee machine

Prediction of Flowback Ratio and Production in Shale Gas

Reservoirs Using a Neural Network Model

Botao Lin

1 College of Petroleum Engineering, China University of Petroleum (Beijing), Beijing

102249, China

The flowback management is an important step in fracturing practices of shale gas

reservoirs because a good handle of it not only preserves the conductivity of the flow

paths created by the fracture network but also assists the subsequent production by

minimizing the damage of the fracturing fluid to the fluid-sensitive formation. A

quantitative prediction of its behavior and how it will affect the productivity of a

reservoir, however, are not available yet if accounting for a large group of reservoir

properties and engineering features. This study attempts a mathematical approach to

predict the flowback behavior concerning flow back ratio, and the productivity

represented by first-month production. First, a BP neural network model will be

established to filter out the controlling factors given the values of actual flow back

ratio from the geological properties and engineering characteristics of the sample

wells, which are used to generate a geological index g and an engineering index e.

Secondly, a correlation is generated between the flow back ratio and the gas well as e

by non-linear fitting. In subsequence, the first-month production will be predicted

based on the field recorded flow back ratio and the comprehensive index c, the latter

of which combines the influences of both geological and engineering parameters.

Finally, the stimulated reservoir volumes (SRVs) of several wells will be estimated

using the K-means clustering and Delaunay triangulation to assess their impact on the

flow back ratio and the first-month production data. Meanwhile, there exists an

optimum flow back ratio range for the shale gas region of concern, implying that

drilling paths and fracturing designs can be optimized to achieve that flow back ratio

range so that the productivity of the formation can be maximized. In general, the

proposed approach can be used in shale gas reservoirs to examine the favorable flow

back ratio and guide the engineering designs to enhance production, as long as

adequate reservoir data are accessible.

Resistivity Simulation and Water Saturation Calculation of Band

Shaped Shaly Sandstone

Feng Wu, Cong Yao, Linlin Cong, Yanping Xi

School of Geoscience and Technology, Southwest Petroleum University, Chengdu

610500, China;

There are two types of shaly sandstone reservoirs in W oil field: high resistivity

reservoir and low resistivity reservoir. However, the causation of the low resistivity

reservoir is unclear and water saturation calculation is not accurate enough. At first,

the micro scale to core scale distribution characteristics of shale is studied by using

SEM, cast thin section, CT scan and core photos. Then, based on typical thin section

phostos, the electrical conductivity model of low resistivity shaly sandstone is

constructed, and the effect of shale on resistivity of shaly sandstone is simulated by

finite element method. Finally, based on the simulation results, an improved water

saturation calculation model is proposed. The results show that: (1) The shaly

sandstone contains high content of irregular band shaped shale, and the clay content

of the adjacent sandstone area is very low. (2) At high water saturation, the electrical

conductivity of shaly sandstone is mainly controlled by the pure sandstone area, and

the resistivity of shaly sandstone increases with the increase of clay content. At

medium to low water saturation, the electrical conductivity of shaly sandstone is

mainly is mainly controlled by the shale area, and the resistivity of shaly sandstone

decreases with the increase of clay content. (3) High content and irregular band

shaped shale is one of the main causes of low resistivity reservoir. (4) The improved

water saturation model has higher accuracy and better correlation coefficient between

calculation water saturation and sealed coring water saturation.

Keywords: band shaped shale, low resistivity reservoir, shaly sandstone, numerical

simulation, resistivity

Email: wufengzh@swpu.edu.cn; wfswpu@126.com

Total organic carbon content prediction based on committee machine from

wireline logs

Yang Bai1, Maojin Tan

1,2

(1. School of Geophysics and Information Technology of China University of

Geosciences, Beijing,100083, China;2. Key laboratory of Geo-detection (China

University of Geoscience), Ministry of Education, Beijing,100083, China)

Abstract

Well logging technology is an important mean of shale gas reservoirs evaluation.

It is a very effective method to quantitatively evaluate the total organic carbon (TOC)

content from wireline logs. Compared with conventional gas reservoirs, the

geophysical characteristics such as reservoir space and the petrophysical

characteristics such as porosity of shale reservoirs are different. Some scholars

proposed many TOC content prediction methods based on well logs. These methods

mainly include empirical formula method, ∆logR method and neural network method.

However, shale gas is little sensitive to geophysical logging response, and the linear

methods is not suitable, and the neural network method is unstable. To this end, we

propose a committee machine to predict the TOC content of shale gas reservoirs.

Aiming to improve the accuracy and robustness of the prediction system, the

committee machine (CM) introduces the human committee's decision-making

mechanism (voting method) into machine learning. A combiner is used to evaluate the

performance of all experts and provide weights for each expert's predictions. The

committee machine gets the final predictions through the combiner. According to the

purpose of logging interpretation, the committee machine can be divided into the

classification committee machine and the regression committee machine. The

classification committee machine may be used for fluid identification, and the

regression committee machine may be used to predict reservoir parameters. For the

shale TOC prediction, we use the regression committee machine.

In this work, we first construct the regression committee machine. The Elman

neural network, the Extreme learning machine (ELM) and the Generalized regression

neural network (GRNN) are used as committee machine experts, and the weighted

average method is used as a combiner. Through investigating the sensitivity

relationship between wireline logs and TOC, we optimized the normalized sensitive

logging data (GR, RD, AC, CNL, DEN, PE, U, TH, K, KTH, THU, etc.) and the

corresponding TOC content as input data. Through the training of committee machine,

we obtained the TOC calculation model. Finally,using the model, we compared the

performance of the committee machine and individual experts through a test set. The

results show that the committee machine has a lower relative error and more stable

performance.

Keywords: Organic shale; total organic carbon; machine learning; committee

machine

Acknowledgements: This work presented is sponsored by National Natural Science

Foundation of China (41774144, U1403191), National Major Projects "Development

of Major Oil& Gas Fields and Coal Bed Methane(2016ZX05014-001)”

Pore-scale investigation of microscopic remaining oil variation

in China's Continental Reservoir at Ultra-High Water Cut

Junjian Li1

Baoyang Cheng，Hanqiao Jiang

1 College of Petroleum Engineering, China University of Petroleum (Beijing), Beijing

102249, China

At present, most of China's oil fields have entered the late ultra-high water cut stage,

with an average water content of more than 95%. The reserves and distribution of

remaining oil are complex and difficult to exploit Therefore, it is of great significance

to understand the remaining oil and exploit the remaining oil. Many macroscopic

production rules and seepage characteristics of reservoirs are comprehensive

reflections of fluid migration under the microstructure and pore scale of the reservoir.

The microstructure of the rocks and the properties of the fluids are fundamental, while

the macroscopic features are expression. Ultra-high water cut oilfields should improve

oil recovery by studying not only the production dynamics in the macroscopic level,

but also the dynamic characteristics of multiple fluids in the microscopic level. In this

paper, the topology structure of the micro-residual oil in four different homogeneous

cores under different displacement phases is analyzed by Micro-CT in-situ scanning.

The micro-residual oils are classified into cluster flow, porous flow, column flow,

droplet flow and membrane flow. During the water flooding process, as the cluster

flow is continuously displaced or broken into other four kinds of discontinuous phase

residual oil, only the proportion of the continuous cluster flow is decreasing, while the

proportion of the remaining four micro-residual oils have different degrees increase.

According to the topology structure of the micro-residual oil after water-flooding, the

distribution pattern of the micro-residual oil was established. Different methods are

used to increase the pore-wave coefficient for different residual oil distribution

patterns: For the case where the continuous residual oil content is dominant,

hydrodynamic methods such as increasing injection speed are continued; For the case

where the discontinuous residual oil content is dominant, the method of surfactant

flooding is adopted; for the case with moderate continuous residual oil content,

microsphere flooding or asp flooding is adopted.

Kewords: ultra-high water cut microscopic residual oil distribution pattern

pore-wave coefficient

Frequency dependencies of cyclic shear friction and shear slip of

granite fracture

Mao Shenga,, Pu Li

a, Xiaoying Zhuang

b, Jianbo Wang

a, Shouceng Tian

a

a. State Key Laboratory of Petroleum Resources and Prospecting, (China University of

Petroleum-Beijing), Beijing 102249, P.R. China

b. Institute of Continuum Mechanics, Appelstraße 11, 30167 Hannover

Abstract: Cyclic hydraulic fracturing with cyclic injection is a recently suggested new

concept, which is alternative to the conventional hydraulic fracturing with continuous

injection. Cyclic hydraulic fracturing can help to reduce the induced seismicity and

also improve the shear slipping and permeability. The influence of cyclic force on

shear friction was found to be an essential factor for fracture slip however this upperic

has not been well addressed. This paper carried out an experimental study of cyclic

shear friction on granite fracture. A cyclic normal force in sinusoidal function and a

constant shear velocity were applied on the rock specimen. The required shear force

and displacement were measured real time so that the time-dependent friction can be

obtained successfully. Results indicate that the frictional strength is determined by the

normal force to the cyclic variation. High frequency normal force stimulates the shear

slip with a lower frictional coefficient of granite fracture. However, the amplitude of

normal force displays irregular influence on the frictional coefficient. Through

microstructure analysis, numerous scratches and a powder-lubrication layer were

produced along the sheared asperity between the fracture surfaces. Moreover, the

powder layer becomes finer when the frequency of normal stress is increased, which

induces lower frictional coefficient. The present work offers more insights to the

mechanisms of cyclic hydraulic fracturing enhancing effectiveness of shear

stimulation.

Keywords: Cyclic loading, shear friction, powder lubrication, normal stress

frequency

Study on Microscopic Residual Oil Distribution in Fractured Low

Permeability Reservoirs

Erlong Yang，Yujia Fang

（Key Laboratory of Enhanced Oil Recovery, Ministry of Education, Department of

Petroleum Engineering, Northeast Petroleum University, Daqing China 163318）

Abstract: In order to improve the oil displacement efficiency of fractured low

permeability reservoirs and explore the distribution law of residual oil. In this paper,

photoetched glass and core anatomical models were used to conduct microscopic

displacement experiments with three oil displacement systems: polymer flooding,

polymer/surfactant binary flooding and polymeric surfactant flooding. The residual oil

is divided into 5 types, the oil displacement effects of different systems and the

variation laws of various remaining oil saturation were obtained. The results show that

after water flooding, the residual oil saturation of fractured model is higher than that

of non-fractured model, and the distribution proportion of cluster residual oil is the

largest. The polymeric surfactant flooding has the largest enhanced oil recovery value,

and the displacement effect of all kinds of residual oil is the best, followed by binary

flooding and polymer flooding, in addition, cluster and column residual oil saturation

decreases the most. Compared with other oil displacement systems, polymeric

surfactant has better plugging effect on fractures, and its viscoelasticity and shearing

action can better carry out the residual oil. The above research can provide guiding

significance for further development of low permeability oil layers.

Key words: microscopic residual oil; fracture; low permeability; polymeric surfactant;

residual oil saturation

Sponsors

教育部非常规油气国际合作联合实验室

（Joint Laboratory of Unconventional Oil and Gas for International

Cooperation of the Ministry of Education）