Mohsen Kazemimanesh, Alireza Moallemi, A. Melina Jefferson, Darcy J. Corbin, Alireza Vali, Matthew R. Johnson, Larry W. Kostiuk, Jason S. Olfert
December 2016
Executive Summary
The objectives of this project were to i) complete a literature review of the composition of flowback of conventional and unconventional wells, ii) design, build, and test a small scale flame and sampling system to enable future work investigating particle formation mechanisms in flames with flowback droplets, and iii) large scale experiments to investigate the effect of flowback droplets on flare efficiency.
Extensive compositional data of flowback and produced water (water produced by the well after the production process has started) is available from thousands of conventional and unconventional well sites through a database maintained by the U.S. Geological Survey. Anions found in high concentrations in the flowback water include (listed in order of decreasing concentration) Cl-, CHO3-, and SO42-, with Cl- having orders of magnitude higher concentrations than the other anions. Common cations include: Na+, Ca2+, K+, and Mg2+. Data analysis on the USGS data showed that conventional hydrocarbon wells has the highest median concentration of cations and chlorine. This is likely due to the fact that the produced water from conventional hydrocarbon wells is mostly brine water and it is not diluted with fresh water from the fracturing process. Furthermore, for unconventional wells, the lowest concentrations of cations and chlorine are found at the beginning of flowback and they slowly increase in concentration with respect to time as flowback continues.
In the small-scale experiment, nanoparticles generated by a laminar methane jet diffusion flame were characterized by direct sampling through the pinhole of a probe and sized with a nanoscanning mobility particle sizer. Particular attention was paid to verify the validity of the probesampling technique. Results showed that the particle size distributions (PSD) were strongly affected by sample dilution immediately after extraction. Within the flame, the soot PSD did not become independent of dilution ratio until the sample was diluted several thousand times (depending on initial concentration) within a few milliseconds of extraction.
High spatially resolved experimental results at different heights (typically 1 mm apart) along the central axis of the flame showed the evolution of PSD for particles larger than 2 nm covering early nucleation and growth to oxidation regions. Moving upward from the fuel-side, prior to the visible flame, the PSD in the particle nucleation region was bimodal (with one of the modes being smaller than 2 nm) and gradually turned into a unimodal PSD with a median of 10 nm due to particle growth. Nearer to the visible flame, the PSD became bimodal again with the larger-diameter mode growing in magnitude and being made up of ever-larger particles, while the smaller-diameter mode gradually shrunk in size until it vanished. Eventually, as the flow was leaving the visible flame, all the particles gradually oxidized until at the tip of the flame they reached a median diameter of 9 nm and in very low concentrations. This work will enable future work in the FlareNet project investigating particle formation mechanisms in flames with flowback droplets.
In the lab-scale experiment, fuel gas comprised of methane, propane, and butane in a mixture representative of upstream oil and gas products in western Canada was burned at atmospheric pressure. At the base of the flame, liquid aerosols were injected into the flare gas stream using an atomizing system, which is capable of delivering consistently-sized droplets at flow rates of up to approximately 60 mL/min. The methodology to evaluate species yields and combustion efficiency is based on capturing the whole product plume and fully mixing it prior to sampling. The effects of injected liquid aerosols will be discussed in terms of prominent combustion product emission trends for various liquid loadings and solution strengths. Key results were: CO had a nonmonotonic response to increasing levels of salt water injection, and emission trends with hydrochloric acid solutions were remarkably similar to distilled water.
# 15-ARPC-02