May-27-22 - Friday Daily Update - Mohammad Elkady

 

Daily Update - Mohammad Elkady - (May-27-2022)

1.  What did I do today?

I have found more papers about ANN in Chemical EOR, I have completed SPE 195492-MS.

2.  What did I find interesting/Work on? 

SPE 195492-MS

·      Model #1: predicts pressure response, the optimum structure is one hidden layer of 112 neurons and a ‘logisig’ transfer function.

·      Model #2: predicts reservoir characteristics, ANN structure is two hidden layers of 98 and 50 neurons with logsig transfer funtions.

o   They constructed back-Check loop to validate the predicted parameters in model #2

 

 

·      Model #3: predicts design parameters, optimum structure is two hidden layers 60 and 90 neurons with tansig and logsig transfer functions.

o   They constructed a back-check model as the one before.

o   To improve the model, they introduced Functional Links to model #3 which are intended to narrow the solution and enhance the ANN by providing additional descriptions of the inputs or predicting additional parameters constructed from inputs and outputs.

·      Model #4: predicts the project design parameters for a desired cumulative oil and target project duration, Optimum structure two hidden layers with 65 and 20 neurons connected by tansig and logsig transfer functions.

·      Model #5: predicts the project design parameters AND the project duration, ANN Structure is two hidden layers with 65 and 20 neurons connected by tansig and logsig transfer functions.  

3.  What will I do next? 

I will continue searching for new papers about ANN utilization in Chemical EOR and I will continue reading more literature about the topic, I will get familiar with ANN models through an online course on DataCamp website. In addition, I will get familiar with CMG STARS through some online tutorials.