Selected work repository

Andrzej Tunkiel
2020

Abstract

This is a repository holding some of my work done at University of Stavanger. These are mostly related to the crossover between Machine Learning, Petroleum and Mechanical Engineering that I am focusing on in pursue of a PhD.

Volve Dataset

Real-time drilling data as CSV

Volve dataset was made public by Equinor in 2018. While it contains a lot of interesting and valuable data, the format is rarely plug and play. This is the case with real-time drilling data that is stored in WITSML format. The logs can be time-based or depth-based with varying degree of completeness inside. These are now parsed and stored at CSV files. Small page dedicated to the dataset is hosted here on the servers of University of Stavanger.

Attribute search

Simple tool was developed for quickly searching all available real-time drilling logs. Just type in a keyword and you will know in which logs attributes containing it reside. Tool is available here.

Interactive data explorer

To explore Volve dataset online a pototype data explorer was created. This is in very early stages in development, and currently it is on a proof of concept stage. It can be found here.


Python

Data analysis crash course

In an effort to promote Python as an alternative to MatLAB a short crash course was developed that contains all the basics needed to get one started. It exposes one to data analysis in Pandas, basic math in Numpy and plotting in Matplotlib and Seaborn. It is in a form of Jupyer Notebook and can be found here.


Publications

Data-driven sensitivity analysis of complex machine learning models: A case study of directional drilling

Classical sensitivity analysis of machine learning regression models is a topic sparse in literature. Most of data-driven models are complex black boxes with limited potential of extracting mathematical understanding of underlying model self-arranged through the training algorithm. Sensitivity analysis can uncover erratic behavior stemming from overfitting or insufficient size of the training dataset. It can also guide model evaluation and application. In this paper, our work on data-driven sensitivity analysis of complex machine learning models is presented. Rooted in one-at-a-time method it utilizes training, validation and testing datasets to cover the hyperspace of potential inputs. The method is highly scalable, it allows for sensitivity analysis of individual as well as groups of inputs. The method is not computationally expensive, scaling linearly both with the available data samples, and in relation to the quantity of inputs and outputs. Coupled with the fact that calculations are considered embarrassingly parallel, it makes the method attractive for big models. In the case study, a regression model to predict inclinations using recurrent neural network was employed to illustrate our proposed sensitivity analysis method and results.

Access via Journal of Petroleum Science and Engineering.

Download PDF

OMAE2020-18151: Drilling Dataset Exploration, Processing and Interpretation Using Volve Field Data

In 2018 Equinor made an unprecedented step for an energy company and made a multi-terabyte dataset from Volve field open. However, there is a long way from downloading data to executing meaningful analysis. With no way of quickly evaluating the data due to its size and unfamiliar file formats the use of Volve data was so far limited. This paper presents our exploratory work related to the real-time drilling part of the dataset. We provide description of common obstacles and approaches for overcoming them. We also describe specific contents of the dataset for others to gauge the potential for case studies. We hope that this will lower the bar for Volve field data accessibility, promote research, and become a catalyst for other data science projects.

Presented on OMAE2020 Virtual Conference, 3-7 August 2020

Access pendingArticle peer-reviewed and accepted, to be published in Q4 2020.

OMAE2020-18154: Continuous Drilling Sensor Data Reconstruction and Prediction via Recurrent Neural Networks

There is an ever-increasing amount of data being recorded in oilfield operations. During drilling a well a large number of parameters is being monitored and saved, often reaching several hundreds. We are seemingly monitoring everything, from basic parameters such as Weight on Bit, Torque, and Rate of Penetration (ROP), to the exhaust temperature of engine no. 3. Unfortunately, the quality of collected data does not match the quantity. Critical sensors, such as gamma and inclination, are often lagging many meters behind the bit. Despite best efforts, sensors stop working, hard drives corrupt files, and data mud pulse telemetry uplinks fail. Methods of infilling data spanning many meters or minutes are necessary. We present a novel approach that enables reliable prediction of data lagging behind the bit through deep neural networks by merging trend-based prediction with traditional neural network approach. We were able to predict continuous inclination data in a curved section of a well with an average absolute error of only 0.4 degrees up to 20 meters from last known value.

Presented on OMAE2020 Virtual Conference, 3-7 August 2020

Access pendingArticle peer-reviewed and accepted, to be published in Q4 2020.

In the oven

I authored multiple publications that are currently in the oven at different stage of completion. This section will be updates as the papers get published.

Patents

Here is a list of patents that I invented. These are always filed in the US and oftentimes in multiple other countries.

Autonomous flow control system and methodology (Granted)

Filed Aug 29, 2014; Patent number: US10145223B2

A technique that facilitates regulation of flow through a flow control device to improve a well operation, such as a production operation. The technique utilizes a flow control device which has a valve positioned in a housing for movement between flow positions. The different flow positions allow different levels of flow through a primary flow port. At least one flow regulation element is used in cooperation with and in series with the valve to establish a differential pressure acting on the valve. The differential pressure is a function of fluid properties and is used to autonomously actuate the flow control device to an improved flow position.

Autonomous flow control system and methodology (2) (Granted)

Filed Aug 28, 2015; Patent number: US9896906B2

A technique that facilitates regulation of flow through a flow control device to improve a well operation, such as a production operation. The technique utilizes a flow control device which has a valve positioned in a housing for movement between flow positions. The different flow positions allow different levels of flow through a primary flow port. At least one flow regulation element, e.g. a Venturi based flow regulation element, is used in a control line in communication with the valve to establish a differential pressure acting on the valve. The differential pressure is a function of fluid properties and is used to autonomously actuate the flow control device to an improved flow position.

System and method for dispersing fluid flow from high speed jet (Pending)

Filed Oct 30, 2015; Patent number: US20160123125

A technique that facilitates dispersion of injected fluid flow. A well string may be constructed with a screen assembly having a base pipe with a radial port, a filter media, and a housing positioned along an exterior of the base pipe. The housing is constructed and positioned to form a chamber which receives high-pressure fluid exiting from an interior of the base pipe through the base pipe port. In some applications, a separate nozzle may be mounted in cooperation with the base pipe port. The screen assembly further comprises a dispersion member having features positioned in a flow path of the injected fluid to disperse the flow and thus to reduce the erosive effects.

System and method for making a threaded connection (Pending)

Filed Nov 29, 2017; Patent number: US20190162031

A system for connecting a male component to a female component comprises a male component comprising a main body including a first compression face, a first tension face, and an outer surface that includes a plurality of first longitudinal splines; a middle element including an inner surface, an outer surface, and a second tension face configured to engage the first tension face, the inner surface including a plurality of second longitudinal splines corresponding to and engaging the first longitudinal splines so as to form a splined interface and the outer surface including a threaded section; and a female component defining a box, the box including a second compression face for engaging the first compression face and an inner wall that includes a wall threaded section corresponding to and engaging the middle element threaded section. The middle element may comprise a plurality of azimuthal segments.

Sand control system and methodology (Pending)

Filed Jul 25, 2013; Patent number: 20150027700

A system and methodology focused on implementing an Inflow Control Device technology in a Gravel Pack product with a specific dehydration system for an efficient system deployment. Previously it was impossible to deploy an ICD completion with a APS Gravel Pack.

Sand control system and methodology employing a tracer (Abandoned)

Filed Aug 29, 2014; Patent number: US20150060059

Techniques and methods allowing for field installation of a tracer into a Sand Screen product. A tracer system traditionally had to be installed in the factory that made a Sand Screen, this invention allows to postpone that until field deployment.

About me

My name is Andrzej Tunkiel. I hold a Masters degree in Mecahnical Engineering with specialty in Mechatronics from Poznan University of Technology, Poland. I have over ten years of experience mechanical design in automotive and petroleum industries. More recently I worked on data analysis as well as in project management positions. I am inventor on six patent applications. I am the founder of Additive Pens, 3D printed fountain pens as well as the editor of fsgk.pl.

Since summer of 2019 I am pursuing a PhD at the University of Stavanger in Data driven methods applied to drilling.

You can learn more about me on my LinkedIn page

Feel free to contact me at andrzej.t.tunkiel@uis.no or visit in person in KE E-331.