Podcast - Petroleum Engineering Data Science

Alfonso R Reyes
(7 May 2019)


The Energy Innovators podcast just released this evening, May 6, 2019, an interview I gave to Francis Norman, with NERA, during my recent trip to Perth, Australia.

Link: https://energyinnovatorspodcast.com/alfonso-reyes-petroleum-data-science/

I hope you enjoy it.

Where I talk about:

  • How I started my career
  • My first contact with data and digital stuff
  • How digital helped model our field fracs
  • Data was not really a love at first sight; digital controllers were
  • What was my Eureka moment in Data Science
  • How looking at datasets of many wells instead of one-well-at-a-time (standalone well) turns to increase oil in all the fields that data science touched
  • How computer science and data science intersect
  • Why analyzing datasets of wells brings you discoveries
  • More Eureka moments with statistical analysis
  • Why now is the moment of experimenting with new approaches in petroleum engineering
  • This is not stopping: as more fields are converted to digital oilfields more data will be flowing
  • Of course you cannot detect outliers with one well = one row
  • No matter how good is your well modeling software is, it becomes irrelevant when you input bad well data
  • Going from standalone wells to networks of wells
  • Applying statistical tools to create multiple scenarios in network models
  • Automating the selection the best oil production scenarios at the lower cost
  • The joy of sending network model programs to production operators
  • How I learned data science with real data, a real challenge: increasing oil rates through production optimization
  • How to listen your wells through data. Wells speak with data
  • Data Science or scratching where it itches
  • Data is generally pretty … pretty ugly: not in the format or shape you want; lots of missing data; bunch of outliers …
  • Data Science, ML and AI is to today’s world what the PC (personal computer) was to the world in the early 80’s
  • Data Science, contrary to the general belief, is not new. Data Science has been around for more than fifty years. It was called Data Analysis then.
  • The field of data science owes John Tukey a lot: a visionary statistician that saw data analysis transcending beyond statistics
  • How far back in time can we go in SPE papers for the terms of data science, machine learning and artificial intelligence
  • What the petroleum engineering industry should be doing on data science the next 25 years
  • Data Science means you have to apply the scientific method to data activities
  • The petroleum industry should be encouraging writing papers in the data science, reproducible way
  • Teaching the younger generation how to fish their own fish
  • How I found that statisticians are very good at teaching data science
  • Data Science, Machine Learning and Artificial Intelligence are complementary -an addon-, to physics based petroleum engineering
  • It is wrong to interpret Data Science as a confrontation between the Physics based world vs Data based solutions (DS, ML, AI)
  • Studying Data Science feels like a journey to our first two years in college or university: math, calculus, statistics, linear algebra … even physics