Determination of Potential Impact Radius for CO2 Pipelines using Machine Learning Approach

Overview

Main Objective

The main objective of this project will be to establish a computational fluid dynamics (CFD) model to simulate the release and dispersion of supercritical CO2 from full pipeline ruptures, then use the simulation results to construct a database comprising CO2 dispersion data under different scenarios. The researcher will use the resulting scenario data in a machine learning analysis for predicting dispersion ranges and health consequences. Using the analysis results and existing literature, the researcher will develop a rapid, universally applicable tool to assess the consequences of accidental CO2 dispersion from high-pressure pipelines.

Public Abstract

Transport of CO2 via pipelines in a supercritical phase is more economic for a larger amount of CO2 over long distances. However, if the CO2 pipeline is accidentally ruptured, it can release a considerable amount of CO2 into the air that could pose harm to humans, especially in those areas that are characterized by a high population density. Accurate source modeling and dispersion modeling form the basis for accurately predicting the consequence. CFD models are becoming more popular to evaluate the release and subsequent dispersion of CO2 because they can fully take into account pipeline characteristics, operation conditions, local geology/geography, and the weather. In this project, we will first establish a CFD model to simulate the release and dispersion of supercritical CO2 from full pipeline ruptures with reasonable accuracy and acceptable computational cost. Then we will construct a database of CO2 dispersion under different scenarios along with some experimental data. Afterward, we will use the scenario and physical properties of CO2 as property descriptors and independent variables in machine learning-based quantitative property consequence relationship (QPCR) analysis for dispersion ranges prediction and health consequence. Being coupled with existing literature about how PIR was addressed for natural gas pipelines and evacuation time evaluation, we will perform risk assessment and develop a rapid, universally applicable tool to assess the consequences of accidental CO2 dispersion and determine the PIR for CO2 pipelines. This tool can be applied to assess the risk of CO2 pipelines during the planning stage and emergency responses.

Anticipated Results: A well-established CFD model for the release and dispersion of supercritical CO2 from pipeline ruptures, a database of CO2 dispersion scenarios, and a reliable and user-friendly machine learning web-based tool to determine the PIR for CO2 pipelines.

Potential Impact on Safety: Using our tool to determine the PIR for CO2 pipelines quickly, the risk of full pipeline ruptures can be thoroughly assessed during the planning stage and responders can quickly determine the emergency plan.

Relevant Files & Links