Title

Nucleation Time Prediction of Selected Inorganic Salts during Reverse Osmosis Treatment of Produced Water in Kern County

Poster Number

9

Lead Author Affiliation

Department of Civil Engineering

Lead Author Status

Masters Student

Introduction

Produced water treatment in drought-stricken regions can serve to augment freshwater supplies for potable and non-potable use. Extensive hydraulic fracturing operations in California contributed to the production of approximately 10.5 billion gallons of natural oil and gas, making it the third-largest oil producing state in America. Thus, the reuse of the large amount of produced and flowback water from these operations serves to benefit drought stricken-regions if viable and cost-effective combinations of membrane treatment can be determined

Purpose

The outcome of the study was to gain general insight into how and when regional variations of produced water constituents can potentially affect reverse osmosis (RO) recovery.

Method

Southern Central Valley oilfields in Kern County, CA were deemed the focus of the current research effort. Data was obtained from a Division of Oil, Gas and Geothermal Resources (DOGGR) produced and flowback water database that contained water quality sampling data from well stimulation disclosures (WSTs) until March 2017. As stated previously, the transition from flowback to produced water has been observed to occur after approximately two to three weeks (Oetjen et al., 2018). Thus, the data set was filtered for samples taken after two weeks to obtain an accurate spread of constituent quantities across the Kern Oil Field.

The data was further filtered to minimize the inclusion of outlier constituent values that could potentially skew the true range of nucleation times in the bulk feed and at the membrane surface. Potential causes for outlier values include sampling and laboratory errors, as well as irregular changes in geological formation characteristics and/or HF additive selection. Thus, the identified outliers may not be representative of general trends in produced water quality on a daily basis. The mean and standard deviation for each constituent were determined and one standard deviation was added and subtracted from the mean values to represent worst-cases and favorable fouling scenarios, respectively. Thus, it is assumed that the data selection methodology is generally-representative of the inorganic fouling index spectrum that would potentially be observed in RO process units across the region.

The modeling process consisted of interfacing results from the Water Applications Value Engine (WAVE) (DOW Chemical) and PHREEQC software packages to develop nucleation models for selected salts of three representative produced water data sets.

PHREEQC was used to determine the SIs and aqueous inorganic salt concentrations in the bulk feed and at the membrane surface. A VMINTEQ (Visual Minteq, Version 3.1) database was used to account for complexation, ionic strength, and pH effects of real waters. Induction times in the bulk feed and at the membrane surface are dependent on the SI values and equilibrium speciation concentrations in their respective locations. Bulk feed induction times were assumed to begin in produced water storage tanks before entering the treatment train, allowing SI and equilibrium concentrations to be extrapolated from speciation of the input values (no reactions were simulated for the bulk feed values). Induction times at the membrane surface were selectively determined by identifying scaling-prone salts in each bulk feed speciation.

Correlations of nucleation induction times for barite, calcite, and silica were extrapolated from prior studies to develop contour maps for three representative source waters.

Results

Barite and calcite induction times were calculated. The onset of fouling in membranes of the following, stated recovery ratios ranges of each fouling scenario occur from 180 to 90 minutes, respectively. In the worst-case fouling scenario (+1 STD), barite fouling is expected at recovery ratios ranging from 55 to 65 percent, whereas calcite fouling is expected to occur at recovery ratios less than 40 percent. In the generally-representative fouling scenario (Median), significant barite fouling begins at recovery ratios ranging from 63 to 70 percent, while significant calcite fouling is expected to at recovery ratios ranging from 57 to 64 percent. In the favorable fouling scenario (-1 STD), significant barite fouling is expected to begin at recovery ratios ranging from 75 to 85 percent, while significant calcite fouling is expected to begin at recovery ratios ranging from 80 to 85 percent.

A contour map for silica was produced to illustrate the need for desilication prior to RO processing. As indicated in Figure 3, silica has incredibly quick induction times relative to barite and calcite. At recovery ratios greater than approximately 57 percent, silica induction occurs within seconds for all three source waters. Thus, lowering the hydraulic residence time to reduce silica nucleation is not possible.

It is indicated at recovery ratios ranging from 55 to 75 percent, barium sulfate has a quicker rate of accumulation at the membrane surface than calcium carbonate. Furthermore, in each representative source water, barium sulfate generally has a higher membrane surface concentration than calcium carbonate. It is also interesting to note that the maximum concentration of barium sulfate occurs at the generally-representative fouling scenario and not the worst-case fouling scenario

Significance

The study developed a model for predicting inorganic fouling in continuous RO systems by interfacing modeled results from PHREEQC and DOW WAVE into numerical estimation of onset fouling times and concentrations as function of RO pass recovery ratios. The contour maps serve as baseline references to predict the onset of barite and calcite fouling, while considering the regional constituent variances in Kern Oil Field source waters. Prediction of selected salt concentrations at the membrane surface provide insight into the impact of inorganic fouling on attainable recovery ratios and RO resistance to fouling of selected salts. The results of this study yielded the following conclusions:

· Wide, regional variances of produced water constituents require specific evaluations to determine which scaling-prone salts will dominate fouling.

· Calcite has a greater impact on RO fouling than barite for all representative source waters.

· The difference in nucleation induction times between the generally-representative and worst-case fouling scenarios of barite is less than calcite at lower recovery ratios, indicating the RO may be more resistant to increases in barium and sulfate ions than calcium and carbonate ions past a certain saturation index.

· The rate of barium sulfate accumulation at the RO membrane surface is quicker than that of calcite, given physical and hydrodynamic characteristics within the chosen DOW SW30HR-380 RO modules, in addition to the approximated molecular diffusion behavior of both salts.

Given that the modeling results suggest barite may induce a greater impact on RO fouling than calcite for certain regional source waters (and vice versa), fouling mitigation strategies should be developed on a regional basis.

Overall, for source water compositions similar to that of the worst-case fouling scenario (particularly, where high concentrations of calcium and carbonate exist), it is recommended that a form of lime softening be utilized as pretreatment. However, in source waters similar to the most favorable and generally-representative fouling scenarios, it is recommended that a pH increase (barite has a PHREEQC-determined saturation pH value of 9.98) and co-precipitation with barite seeding and other inorganic salts be employed. While lowering the hydraulic residence time of RO modules remains a valid strategy, it is recommended that comparative analyses be conducted to evaluate economic and practical operational trade-offs of emphasizing either pretreatment or decreased hydraulic residence time as viable fouling mitigation strategies.





Location

DeRosa University Center

Format

Poster Presentation

Poster Session

Morning

This document is currently not available here.

Share

COinS
 
Apr 27th, 10:00 AM Apr 27th, 12:00 PM

Nucleation Time Prediction of Selected Inorganic Salts during Reverse Osmosis Treatment of Produced Water in Kern County

DeRosa University Center

Produced water treatment in drought-stricken regions can serve to augment freshwater supplies for potable and non-potable use. Extensive hydraulic fracturing operations in California contributed to the production of approximately 10.5 billion gallons of natural oil and gas, making it the third-largest oil producing state in America. Thus, the reuse of the large amount of produced and flowback water from these operations serves to benefit drought stricken-regions if viable and cost-effective combinations of membrane treatment can be determined