### Create a StudySoup account

#### Be part of our community, it's free to join!

Already have a StudySoup account? Login here

# Art Appreciation ART T113

WVU

GPA 3.88

### View Full Document

## 32

## 0

## Popular in Course

## Popular in Art

This 70 page Class Notes was uploaded by Ms. Samanta Collins on Saturday September 12, 2015. The Class Notes belongs to ART T113 at West Virginia University taught by Staff in Fall. Since its upload, it has received 32 views. For similar materials see /class/202681/art-t113-west-virginia-university in Art at West Virginia University.

## Reviews for Art Appreciation

### What is Karma?

#### Karma is the currency of StudySoup.

#### You can buy or earn more Karma at anytime and redeem it for class notes, study guides, flashcards, and more!

Date Created: 09/12/15

SPE 83446 Identification of Contaminated Data in Hydraulic Fracturing Databases Application to the Codequot Formation in the DJ Basin Andrei Popa Shahab Mohaghegh Razi Gaskari and Sam Ameri WEST VIRGINIA UNIVERSITY OUTLINE III Objective I Introduction I Background II Methodology III Results amp Discussion II Conclusions SPE 83446 Popa Mohaghegh Gaskari and Ameri OBJECTIVE IITo introduce a new methodology for identification of contaminated data in hydraulic fracturing databases I This methodology Integrates clustering techniques neural network modeling and an iterative process to achieve the convergence goal SPE 83446 Popa Mohaghegh Gaskari and Ameri www BACKGROUND El Part of a comprehensive study of best practices identification for restimulation El SPE 77597 Identification of Successful Practices in Hydraulic Fracturing Using Intelligent Data Mining Tools Application to the Codell Formation in the DJ Basin El Incorporates stateoftheart in data mining knowledge discovery and dataknowledge fusion techniques SPE 83446 Popa Mohaghegh Gaskari and Ameri BACKGROUND III Methodology includes five steps 1 2 3 4 5 Data Quality Control Fuzzy Combinatorial Analysis Production Data Analysis Neural Model Building Successful Practices Analysis SPE 83446 Popa Mohaghegh Gaskari and Ameri n INTRODUCTION El Databases are not always accurate and useful Missing data I Incorrect data or contaminated data a lmproperl incomplete data collection I Data entry errors a Lack of proper interpretation In Gauge reading errors I Data collection not done by trained personnel in lnattention n Automatic data collection without verification l quality control 1 Simply random natural chaos SPE 83446 Popa Mohaghegh Gaskari and Ameri n INTRODUCTION El Artificial Neural Networks I Distributive parallel processing system capable of solving nonlinear problems III Fuzzy Clustering I Grouping of similar objects classifying elements of a data set into groups using similarity criterion I Entropy I Measure of the lack of order in a system or the degree of lack of similarity between elements SPE 83446 Popa Mohaghegh Gaskari and Ameri INTRODUCTION IZIGoaI To train a neural network capable of predicting PostRestimulation Peak Production SPE 83446 Popa Mohaghegh Gaskari and Ameri n INTRODUCTION 73 39Biarbonate ppm 7 4 Peak Vlsc 7 Lat Orig mf SandQMibs Long Refrac Date ViscShear 100 30Min TotHandness ppm Calcium ppm 39 i Angate BPM Est Uit GOR quotNo Ni Peifs iiAngsi 71j67 Vi Eh ai 1003561 in 7 17 109cc F39en 39 7 18 7 TDSoIid ppm 19 MMCf SPE 83446 Popa Mohaghegh Gaskari and Ameri 22 23 21 25 rs used in the study 1 pr Frac Type No COPerfs E laide ppm Nl Perfed H Water pHLab quotP re Re rac M39Cf39d Cum MMcf Water Source Total Perfs Sulfate ppm New Perfs Sodium ppm 737 38 4o 41 42 Magnesium pp 7 Vi oS heat quot1 JO 0M i Pre Fraci DP iPost FmdSiiP MIb20 40 Communication 7 METHODOLOGY I First attempt at analysis led to impossible interpretation of data III Data quality control then initiated SPE 83446 Popa Mohaghegh Gaskari and Ameri METHODOLOGY IIldentification of outliers I Using Fuzzy Curves together with Regression plots n12 wells were eliminated as belonging to naturally fractured part of field SPE 83446 Popa Mohaghegh Gaskari and Ameri METHODOLOGY IIldentification of outliers I Using Fuzzy Curves together with Regression plots n12 wells were eliminated as belonging to naturally fractured part of field SPE 83446 Popa Mohaghegh Gaskari and Ameri mm METHODOLOGY Fuzzy Curve Analysis for outlier identification Fuzz C l bi a m i AWYSB Fuzzy Cmrfhiminxia Analysis i i 300 240 250 260 Origl fll Sandilibs quot quotim MIND4 SPE 83446 Popa Mohaghegh Gaskari and Ameri mwm METHODOLOGY EIRegression plots resemble shotgun Regresinn Analysis 3 D D 395 9 3 13 x x e g 5 E E Clrig FIuidMgal Orig FluidMeal SPE 83446 Popa Mohaghegh Gaskari and Ameri mwm METHODOLOGY ySIS EIRegression plots resemble sotgun M Ln a 0 L0 C T u ACT BOEI39MII B 8 5 3500 4000 4500 5000 5500 5000 0500 7000 7500 A39uigPsi SPE 83446 Popa Mohaghegh Gaskari and Ameri mwm METHODOLOGY D F re uenc distribution have no trend Framerquot is quuc my SPE 83446 Popa Mohaghegh Gaskari and Ameri METHODOLOGY IIAgain unable to train a Neural Network Model I Several types of neural networks I Different neural network architectures I Combinations of parameters I Best results Correlation coefficient 017 R2 007 SPE 83446 Popa Mohaghegh Gaskari and Ameri METHODOLOGY IZIWe concluded that Data can not be trained in its present form Ellntelligent system approach was considered for identification and clean up of the erroneous records in the dataset SPE 83446 Popa Mohaghegh Gaskari and Ameri mwm METHODOLOGY II NeuroCluster Data Classification System I Integrates clustering techniques neural networkmodeling and iterative process to achIeve convergence I Separates data into 3 categories nGood data nSlightly contaminated data n Bad data I Uses most influential parameters SPE 83446 Popa Mohaghegh Gaskari and Ameri WWWMETHODOLOGY 1 Cluster data using output of system 2 Train ANN using cluster info membership functions and entropy 3 Iterative process 4 Repeat step 3 for each record in dataset SPE 83446 Popa Mohaghegh Gaskari and Ameri METHODOLOGY 1 Cluster data using output of system III Since the output is used during the cluster analysis the information generated by this analysis carries the signature of the system behavior as a whole inputoutput SPE 83446 Popa Mohaghegh Gaskari and Ameri METHODOLOGY 2 Train ANN using cluster info membership functions and entropy III Results of step 1 is used during the network training This means the system output is contributing to the input albeit indirectly El This constitutes a bootstrapping method where output is indirectly present in the system input SPE 83446 Popa Mohaghegh Gaskari and Ameri WWWMETHODOLOGY 3 Iterative process Select a step for the range of output system Sweep output range while firing neural network For each value of assumed output Calculate cluster number for assumed value Calculate entropy Fire neural network Calculate error NM output Assumed Output Display all 3 curves on single graph Identify position where data error reaching zero and entropy is minimum for constant cluster label representation SPE 83446 Popa Mohaghegh Gaskari and Ameri METHODOLOGY ElThis methodology is based on the following hypothesis I In a well behaved system the output should be able to contribute to its own prediction and identification SPE 83446 Popa Mohaghegh Gaskari and Ameri quot RESULTS amp DISCUSSION El Output range for the output parameter during identification was between 500 2500 scf II Three fuzzy clusters were considered III Entropy was defined as the ratio between the smallest and largest membership function resulted after clustering The range of the entropy was between 0 and 1 El Error curve between 1000 and 1500 scf SPE 83446 Popa Mohaghegh Gaskari and Ameri quot RESULTS amp DISCUSSION IZIGood data Diff Assumed ActPeak Predicted ActPeak vs Assumed ActPeak Curve 1 Cluster Curve 2 Entropy NN prediction Original Output Curve 3 Error Curve Assumed ActuIPeak SPE 83446 Popa Mohaghegh Gaskari and Ameri quot RESULTS amp DISCUSSION Diff Assumed ActPeak Predicted ActPeak vs Assumed ActPeak NN prediction Original Output Curve 2 Entropy 1000 1500 5j 000 3500 Curve 1 Cluster Curve 3 Error Curve Assumed SPE 83446 Popa Mohaghegh Gaskari and Ameri quot RESULTS amp DISCUSSION Diff Assumed ActPeak Predicted ActPeak vs Assumed ActPeak Curve 2 Entropy Curve 1 Cluster NN prediction Original Output Curve 3 Error Curve ActulPeak SPE 83446 Popa Mohaghegh Gaskari and Ameri quot RESULTS amp DISCUSSION Diff Assumed ActPeak Predicted ActPeak vs Assumed ActPeak NN prediction Original Output Curve 2 Entropy Curve 1 Cluster Curve 3 Error Curve Assumed ActulPeak SPE 83446 Popa Mohaghegh Gaskari and Ameri quot RESULTS amp DISCUSSION Diff Assumed ActPeak Predicted ActPeak vs Assumed ActPeak NN prediction Original Output Curve 2 Entropy Curve 1 Cluster 3000 3500 Curve 3 Error Curve Assumed ActulPeak SPE 83446 Popa Mohaghegh Gaskari and Ameri alumquot RESULTS amp DISCUSSION Diff Assumed ActPeak Predicted ActPeak vs Assumed ActPeak NNprediction Ori inalOu ut Curve 1 Cluster g tp Curve 2 Entropy 1500 2000 Curve 3 Error Curve Assumed ActulPeak SPE 83446 Popa Mohaghegh Gaskari and Ameri RESULTS amp DISCUSSION IZIData classified 88 good 26 slightly contaminated 60 bad IZIResults Correlation coefficient 0876 R2 082 SPE 83446 Popa Mohaghegh Gaskari and Ameri quot RESULTS amp DISCUSSION Training Dina Sul hrzuion Dala Scl g 8 390 Actual quot Predicted E Fl a u lt E El 393 E E u 3 x Q E A Prmlicml M Rnsrjmulmian Actual Peak I Q Past SPE 83446 Popa Mohaghegh Gaskari and Ameri CONCLUSIONS III A NeuroCluster Data Classification System was introduced to identify contaminated data III The combination of two intelligent tools neural networks and fuzzy Cmean clustering provides a simple solution to problems like data classification as presented in this paper SPE 83446 Popa Mohaghegh Gaskari and Ameri CONCLUSIONS I The applicability of this methodology was verified using a synthetic dataset developed using a commercial fracture simulator I The application of this methodology can be extended to any type of database for identification of contaminated data SPE 83446 Popa Mohaghegh Gaskari and Ameri SPE 83446 Identification of Contaminated Data in Hydraulic Fracturing Databases Application to the Codequot Formation in the DJ Basin Andrei Popa Shahab Mohaghegh Razi Gaskari and Sam Ameri WEST VIRGINIA UNIVERSITY OUTLINE III Objective I Introduction I Background II Methodology III Results amp Discussion II Conclusions SPE 83446 Popa Mohaghegh Gaskari and Ameri OBJECTIVE IITo introduce a new methodology for identification of contaminated data in hydraulic fracturing databases I This methodology Integrates clustering techniques neural network modeling and an iterative process to achieve the convergence goal SPE 83446 Popa Mohaghegh Gaskari and Ameri www BACKGROUND El Part of a comprehensive study of best practices identification for restimulation El SPE 77597 Identification of Successful Practices in Hydraulic Fracturing Using Intelligent Data Mining Tools Application to the Codell Formation in the DJ Basin El Incorporates stateoftheart in data mining knowledge discovery and dataknowledge fusion techniques SPE 83446 Popa Mohaghegh Gaskari and Ameri BACKGROUND III Methodology includes five steps 1 2 3 4 5 Data Quality Control Fuzzy Combinatorial Analysis Production Data Analysis Neural Model Building Successful Practices Analysis SPE 83446 Popa Mohaghegh Gaskari and Ameri n INTRODUCTION El Databases are not always accurate and useful Missing data I Incorrect data or contaminated data a lmproperl incomplete data collection I Data entry errors a Lack of proper interpretation In Gauge reading errors I Data collection not done by trained personnel in lnattention n Automatic data collection without verification l quality control 1 Simply random natural chaos SPE 83446 Popa Mohaghegh Gaskari and Ameri n INTRODUCTION El Artificial Neural Networks I Distributive parallel processing system capable of solving nonlinear problems III Fuzzy Clustering I Grouping of similar objects classifying elements of a data set into groups using similarity criterion I Entropy I Measure of the lack of order in a system or the degree of lack of similarity between elements SPE 83446 Popa Mohaghegh Gaskari and Ameri INTRODUCTION IZIGoaI To train a neural network capable of predicting PostRestimulation Peak Production SPE 83446 Popa Mohaghegh Gaskari and Ameri n INTRODUCTION 73 39Biarbonate ppm 7 4 Peak Vlsc 7 Lat Orig mf SandQMibs Long Refrac Date ViscShear 100 30Min TotHandness ppm Calcium ppm 39 i Angate BPM Est Uit GOR quotNo Ni Peifs iiAngsi 71j67 Vi Eh ai 1003561 in 7 17 109cc F39en 39 7 18 7 TDSoIid ppm 19 MMCf SPE 83446 Popa Mohaghegh Gaskari and Ameri 22 23 21 25 rs used in the study 1 pr Frac Type No COPerfs E laide ppm Nl Perfed H Water pHLab quotP re Re rac M39Cf39d Cum MMcf Water Source Total Perfs Sulfate ppm New Perfs Sodium ppm 737 38 4o 41 42 Magnesium pp 7 Vi oS heat quot1 JO 0M i Pre Fraci DP iPost FmdSiiP MIb20 40 Communication 7 METHODOLOGY I First attempt at analysis led to impossible interpretation of data III Data quality control then initiated SPE 83446 Popa Mohaghegh Gaskari and Ameri METHODOLOGY IIldentification of outliers I Using Fuzzy Curves together with Regression plots n12 wells were eliminated as belonging to naturally fractured part of field SPE 83446 Popa Mohaghegh Gaskari and Ameri METHODOLOGY IIldentification of outliers I Using Fuzzy Curves together with Regression plots n12 wells were eliminated as belonging to naturally fractured part of field SPE 83446 Popa Mohaghegh Gaskari and Ameri mm METHODOLOGY Fuzzy Curve Analysis for outlier identification Fuzz C l bi a m i AWYSB Fuzzy Cmrfhiminxia Analysis i i 300 240 250 260 Origl fll Sandilibs quot quotim MIND4 SPE 83446 Popa Mohaghegh Gaskari and Ameri mwm METHODOLOGY EIRegression plots resemble shotgun Regresinn Analysis 3 D D 395 9 3 13 x x e g 5 E E Clrig FIuidMgal Orig FluidMeal SPE 83446 Popa Mohaghegh Gaskari and Ameri mwm METHODOLOGY ySIS EIRegression plots resemble sotgun M Ln a 0 L0 C T u ACT BOEI39MII B 8 5 3500 4000 4500 5000 5500 5000 0500 7000 7500 A39uigPsi SPE 83446 Popa Mohaghegh Gaskari and Ameri mwm METHODOLOGY D F re uenc distribution have no trend Framerquot is quuc my SPE 83446 Popa Mohaghegh Gaskari and Ameri METHODOLOGY IIAgain unable to train a Neural Network Model I Several types of neural networks I Different neural network architectures I Combinations of parameters I Best results Correlation coefficient 017 R2 007 SPE 83446 Popa Mohaghegh Gaskari and Ameri METHODOLOGY IZIWe concluded that Data can not be trained in its present form Ellntelligent system approach was considered for identification and clean up of the erroneous records in the dataset SPE 83446 Popa Mohaghegh Gaskari and Ameri mwm METHODOLOGY II NeuroCluster Data Classification System I Integrates clustering techniques neural networkmodeling and iterative process to achIeve convergence I Separates data into 3 categories nGood data nSlightly contaminated data n Bad data I Uses most influential parameters SPE 83446 Popa Mohaghegh Gaskari and Ameri WWWMETHODOLOGY 1 Cluster data using output of system 2 Train ANN using cluster info membership functions and entropy 3 Iterative process 4 Repeat step 3 for each record in dataset SPE 83446 Popa Mohaghegh Gaskari and Ameri METHODOLOGY 1 Cluster data using output of system III Since the output is used during the cluster analysis the information generated by this analysis carries the signature of the system behavior as a whole inputoutput SPE 83446 Popa Mohaghegh Gaskari and Ameri METHODOLOGY 2 Train ANN using cluster info membership functions and entropy III Results of step 1 is used during the network training This means the system output is contributing to the input albeit indirectly El This constitutes a bootstrapping method where output is indirectly present in the system input SPE 83446 Popa Mohaghegh Gaskari and Ameri WWWMETHODOLOGY 3 Iterative process Select a step for the range of output system Sweep output range while firing neural network For each value of assumed output Calculate cluster number for assumed value Calculate entropy Fire neural network Calculate error NM output Assumed Output Display all 3 curves on single graph Identify position where data error reaching zero and entropy is minimum for constant cluster label representation SPE 83446 Popa Mohaghegh Gaskari and Ameri METHODOLOGY ElThis methodology is based on the following hypothesis I In a well behaved system the output should be able to contribute to its own prediction and identification SPE 83446 Popa Mohaghegh Gaskari and Ameri quot RESULTS amp DISCUSSION El Output range for the output parameter during identification was between 500 2500 scf II Three fuzzy clusters were considered III Entropy was defined as the ratio between the smallest and largest membership function resulted after clustering The range of the entropy was between 0 and 1 El Error curve between 1000 and 1500 scf SPE 83446 Popa Mohaghegh Gaskari and Ameri quot RESULTS amp DISCUSSION IZIGood data Diff Assumed ActPeak Predicted ActPeak vs Assumed ActPeak Curve 1 Cluster Curve 2 Entropy NN prediction Original Output Curve 3 Error Curve Assumed ActuIPeak SPE 83446 Popa Mohaghegh Gaskari and Ameri quot RESULTS amp DISCUSSION Diff Assumed ActPeak Predicted ActPeak vs Assumed ActPeak NN prediction Original Output Curve 2 Entropy 1000 1500 5j 000 3500 Curve 1 Cluster Curve 3 Error Curve Assumed SPE 83446 Popa Mohaghegh Gaskari and Ameri quot RESULTS amp DISCUSSION Diff Assumed ActPeak Predicted ActPeak vs Assumed ActPeak Curve 2 Entropy Curve 1 Cluster NN prediction Original Output Curve 3 Error Curve ActulPeak SPE 83446 Popa Mohaghegh Gaskari and Ameri quot RESULTS amp DISCUSSION Diff Assumed ActPeak Predicted ActPeak vs Assumed ActPeak NN prediction Original Output Curve 2 Entropy Curve 1 Cluster Curve 3 Error Curve Assumed ActulPeak SPE 83446 Popa Mohaghegh Gaskari and Ameri quot RESULTS amp DISCUSSION Diff Assumed ActPeak Predicted ActPeak vs Assumed ActPeak NN prediction Original Output Curve 2 Entropy Curve 1 Cluster 3000 3500 Curve 3 Error Curve Assumed ActulPeak SPE 83446 Popa Mohaghegh Gaskari and Ameri alumquot RESULTS amp DISCUSSION Diff Assumed ActPeak Predicted ActPeak vs Assumed ActPeak NNprediction Ori inalOu ut Curve 1 Cluster g tp Curve 2 Entropy 1500 2000 Curve 3 Error Curve Assumed ActulPeak SPE 83446 Popa Mohaghegh Gaskari and Ameri RESULTS amp DISCUSSION IZIData classified 88 good 26 slightly contaminated 60 bad IZIResults Correlation coefficient 0876 R2 082 SPE 83446 Popa Mohaghegh Gaskari and Ameri quot RESULTS amp DISCUSSION Training Dina Sul hrzuion Dala Scl g 8 390 Actual quot Predicted E Fl a u lt E El 393 E E u 3 x Q E A Prmlicml M Rnsrjmulmian Actual Peak I Q Past SPE 83446 Popa Mohaghegh Gaskari and Ameri CONCLUSIONS III A NeuroCluster Data Classification System was introduced to identify contaminated data III The combination of two intelligent tools neural networks and fuzzy Cmean clustering provides a simple solution to problems like data classification as presented in this paper SPE 83446 Popa Mohaghegh Gaskari and Ameri CONCLUSIONS I The applicability of this methodology was verified using a synthetic dataset developed using a commercial fracture simulator I The application of this methodology can be extended to any type of database for identification of contaminated data SPE 83446 Popa Mohaghegh Gaskari and Ameri

### BOOM! Enjoy Your Free Notes!

We've added these Notes to your profile, click here to view them now.

### You're already Subscribed!

Looks like you've already subscribed to StudySoup, you won't need to purchase another subscription to get this material. To access this material simply click 'View Full Document'

## Why people love StudySoup

#### "There's no way I would have passed my Organic Chemistry class this semester without the notes and study guides I got from StudySoup."

#### "I bought an awesome study guide, which helped me get an A in my Math 34B class this quarter!"

#### "There's no way I would have passed my Organic Chemistry class this semester without the notes and study guides I got from StudySoup."

#### "It's a great way for students to improve their educational experience and it seemed like a product that everybody wants, so all the people participating are winning."

### Refund Policy

#### STUDYSOUP CANCELLATION POLICY

All subscriptions to StudySoup are paid in full at the time of subscribing. To change your credit card information or to cancel your subscription, go to "Edit Settings". All credit card information will be available there. If you should decide to cancel your subscription, it will continue to be valid until the next payment period, as all payments for the current period were made in advance. For special circumstances, please email support@studysoup.com

#### STUDYSOUP REFUND POLICY

StudySoup has more than 1 million course-specific study resources to help students study smarter. If you’re having trouble finding what you’re looking for, our customer support team can help you find what you need! Feel free to contact them here: support@studysoup.com

Recurring Subscriptions: If you have canceled your recurring subscription on the day of renewal and have not downloaded any documents, you may request a refund by submitting an email to support@studysoup.com

Satisfaction Guarantee: If you’re not satisfied with your subscription, you can contact us for further help. Contact must be made within 3 business days of your subscription purchase and your refund request will be subject for review.

Please Note: Refunds can never be provided more than 30 days after the initial purchase date regardless of your activity on the site.