Benefits of polymeric membranes in Oil and Gas produced water treatment

In oil and gas formations, the subsurface rocks are generally filled with fluids such as water, oil and/or gas, or some combination of these. Therefore, oil and gas reservoirs often contain both formation water as well as hydrocarbons. During production, this formation water is co-produced along with the oil and gas (Amyx et al. 1960) and is defined as Produced Water (PW). PW is the largest waste stream generated in oil and gas industries.

Although conventional oil production has peaked, the PW volumes are increasing because in mature fields, water is injected to displace oil from the pore spaces towards a producing well. This process is known as Enhanced Oil Recovery (EOR) and more specifically water flooding if only water is injected.

The cost of managing such a large volume of water is a key consideration to oil and gas producers. Historically, PW was disposed of in large evaporation ponds and/or in disposal wells. However, these disposal methods are becoming increasingly unacceptable from both environmental and social perspectives. PW is considered an industrial waste and producers are now required to employ beneficial reuse options.

In water-scarce regions, limited freshwater resources in conjunction with the high treatment cost for PW discharge (Clark & Veil 2009) makes PW reinjection an attractive opportunity. However, new extraction processes such as chemical/steam/low salinity EOR drive the need for improved treatment of dispersed oil and particles in PW. Conventional PW treatment technologies as shown in Figure 1, such as gravity separators for primary deoiling, induced gas or dissolved gas flotators for secondary deoiling and walnut shell filters for tertiary deoiling are often unable to meet these challenging injection specifications.

UltraFiltration (UF) and MicroFiltration (MF) membranes, as secondary and/or tertiary deoiling step, can address these needs. Indeed UF/MF membranes not only generate higher quality effluent then conventional technologies but are also cost-competitive and less chemical intensive (Padaki et al. 2015). Selecting the right membrane is a complicated task and requires a multi-criteria approach. There is a wide variety of manufacturers offering membranes of various materials, cut-offs, designs and shapes. While membrane filtration is used extensively in domestic water applications, only a few full scale membrane systems exist to treat PW. Due to their high structural robustness, ceramic membranes are often preferred over polymeric membranes for the PW treatment application (Drewes et al. 2009; He & Vidic 2016). Nevertheless, polymeric membranes offer the benefits of being less expensive and result in a lower footprint and weight, which fits the requirements of the offshore industry.

The objective of the study was to compare the performances of ten membranes to treat real PW and identify which of the structural and operational characteristics of the membranes are the success factors to ensure cost-effective, long-term and reliable operation. For that purpose, the membranes tested were selected to cover a wide variety of materials and designs: five membrane materials (PVDF, PAN, PTFE, SiC, ZrO₂), Pore sizes ranging from 0.01 to 1 μm, four membrane designs (multichannel, tubular, hollow fiber, spiral wound) and two types of filtration (crossflow, dead-end).

The ten membranes studied are presented in Table 1. The membranes were organized in three groups: dead-end polymeric membranes (group 1), crossflow polymeric membranes (group 2), and crossflow ceramic membranes (group 3).

The filtration tests were run with real PW collected from an onshore conventional oilfield. The oil and suspended solids content of the PW was adjusted during the tests using crude oil and silica sand to reach a target value of approximately 100 ppm to match the typical quality of a plate gravity separator outlet. The particle size distribution of the silica sand added was chosen to mimic the typical particle size distribution of PW after gravity separation in a plate interceptor with a typical 60 μm cut-off.

The particle size distribution of the Silica sand in the PW during the filtration tests is displayed in Figure 2.

The tests were run at constant feed pressure (Xiong et al. 2016) using lab scale modules of 0.01 to 0.3 m² surface area. The characterization of the oil rejection of the membranes was done by offline lab analyses of the PW and the permeate. Three reference methods for the analysis of oil in water were used to assess the filtration performances of the membranes: Hexane Extractable Matter (US-EPA 1,664 HEM), Silica Gel Treated Hexane Extractable Matter (US-EPA 1664 SGT-HEM) and Total Petroleum Hydrocarbons (ASTM-D7678). Naphthenic Acids and Chemical Oxygen Demand were also analyzed to further characterize the water soluble oil (WSO) content of the PW.

The hydraulic performances of the membranes were assessed based on the following indicators (Alpatova et al. 2014):

• - Permeability over time indicating the operational cost efficiency of the membrane.

• - Reversible fouling (A) indicating the resistance due to loosely attached and pore blocking foulants that are removed by the backwash sequence. A (10¹²/m²) = ΔRn/(V/S) over 1 filtration cycle between 2 backwash sequences.

• - Irreversible fouling (B) indicating the resistance due to adsorbed and pore blocking foulants that are not removed by the backwash sequence. B (10¹²/m²) = ΔRn/(V/S) over the 16–20 filtration cycles, with: Rn = 1/ (Lp × μ_{(20 °C)}) in m⁻¹

  • V = Volume filtered (m³)

  • S = Membrane active area (m²)

  • Lp = Membrane permeability (mL/m².s.kPa)

  • μ = viscosity (Pa,s)

• - Backwash recovery (C) characterizes the permeability recovery after the backwash sequence. C (%) = (1–B / A) × 100.

The hydraulic tests were run over 16–20 hours of filtration per membrane at a feed temperature of 40 °C. For each of the three membrane groups an average filtration flux was defined. To reach this average flux over the 16–20 hours of filtration, the initial filtration flux was set at 130% of the target flux. The feed pressure was kept constant by the means of a pressure controlled variable frequency drive on the membrane feed pump. The hydraulic tests were run based on the operating guidelines presented in Table 2.

The ageing tests were conducted on polymeric membranes in artificially accelerated exposure to aromatic hydrocarbons to assess and compare their lifetime and resistance to chemical ageing when used for PW treatment. For safety reasons, Toluene was chosen to represent the Benzene Toluene Ethylbenzene Xylene (BTEX) compounds for the tests because less toxic and volatile than Benzene (Van Kuppevelt 2015). An accelerating factor of 17.4 was chosen to simulate 4 years lifetime of the membranes within 12 weeks of testing. The accelerated ageing chemical bath was prepared with a synthetic solution of Toluene at 1.7 g/L and 35 °C. Membrane fibers were sampled at weeks 0, 1, 3, 6 and 12 of soaking to assess the physical and chemical changes of each membrane with a range of analyses. The tests were conducted according to the experimental plan presented in Table 3.

Tensile tests were conducted to assess the ageing effect on the mechanical properties of the membranes. Fourier Transform InfraRed (FTIR) spectroscopy analyses were carried out to assess the ageing effect on the chemical structure of the membranes. FTIR can detect if the hydrophilic additives of the membranes are stable or eluate over time resulting in a loss of the membrane fouling resistance. Filtration tests were run with real oilfield PW to assess the oil rejection over time of the membranes.

The results presented in this section aim to compare polymeric and ceramic membranes based on the performance indicators identified as key for reliable and cost-effective operations: contaminants rejection, permeability, resistance to fouling and life expectancy.

The oil rejection rates (by US-EPA1664 HEM method) of the membranes organized by material and pore size are presented in Figure 3 .

Both the polymeric and ceramic membranes achieved a very satisfactory removal of oil from PW at an average rate of 88.2% generating a permeate with 10–24 mg/L of oil residual. Both the water insoluble oil (WIO) and WSO fractions were analyzed. The WIO was removed at an average rate of 99.4% while the WSO was removed at 81.5% in average. The rejection rates of the overall contaminants assessed were fairly homogeneous among the ten membranes tested and no correlation with the cut-off or the membrane material was established, as shown in Table 4.

Permeability was evaluated over 16 to 20 hours of filtration for all ten membranes. The permeability value at 1,600 liters of filtrate per m² of membrane was set as benchmark indicator, as presented in Figure 4 and 5.

The ceramic membranes in SiC outperformed the other membranes with permeability values at 800 to 2,000 L/m².h.bar at 40 °C. All the other membranes showed fairly consistent permeability values in the range of 300 to 500 L/m².h.bar. Apart from the SiC membranes, no permeability difference is observed between polymeric and ceramic membranes. Once again, no correlation was established between the pore size and the permeability of the membrane. However the best permeability values are observed for the two membranes exhibiting a pore size in the range of 0.1 to 0.5 μm.

The irreversible fouling indicators of the ten membranes tested are displayed here below. In Figure 6 irreversible fouling values are organized by membrane material categories.

In Figure 7 irreversible fouling values are plotted against the membrane pore sizes.

In Figure 8 irreversible fouling values are organized by filtration types.

The membranes that showed the best resistance to fouling were the dead-end hollow fiber membrane in PTFE and the crossflow multichannel ceramic membranes in SiC. Once again, it is difficult to observe clear correlations of irreversible fouling with specific characteristics of the membranes tested. However low pore size seemed to slightly favour irreversible fouling in this study.

Resistance to ageing and life expectancy were assessed for the three polymeric membranes that achieved the best results in contaminants rejection, permeability and resistance to fouling: polymeric a (ref.: 1-HF-DE-0.08), polymeric b (ref.: 1-HF-DE-50kD) and polymeric c (ref.: 2-TUB-CF-0.03). Resistance to ageing is a critical performance indicator in the PW treatment application since the polymeric material is expected to be far less resistant chemically and mechanically than the ceramic material over time.

In this section, the effect of ageing on the filtration performance of the membranes was assessed. The oil rejection rates were selected as key performance indicators. The tests showed that accelerated ageing had no significant impact on the insoluble oil rejection rates. On the other hand, the WSO rejection rates were slightly affected as shown in Figure 9.

At year 2 of simulated age, the WSO rejection rate starts decreasing with an analogous trend for the three membranes. However, the minor ageing effect observed indicates a good stability of the filtration performance over time.

Tensile tests were conducted to assess the ageing effect on the mechanical properties for three replicates per membrane. Out of the three membranes tested, two showed minor evolution of their mechanical properties, indicating a good resistance to ageing. One membrane (polymeric a) showed signs of mechanical ageing as shown in Figure 10.

A significant evolution of the elongation at break can be observed between year 2 and 4 of simulated age indicating that the mechanical properties of the membrane are affected and that the fibers irreversibly elongate under a tensile load. In real operations, this phenomena leads to an increasing fiber slack in the module over time resulting in accelerated membrane wear due to fibers friction.

FTIR spectroscopy analyses were carried out to assess the ageing effect on the chemical structure of the three membranes. FTIR can detect if the hydrophilic additives of the membranes are stable over time. The tests showed that the chemical structure of two membranes out of three remained stable. One membrane (polymeric c) exhibited signs of chemical ageing, as shown in Figure 11.

At year 1 of simulated age, the loss of the hydrophilic agent of the polymeric membrane c was observed. This result can possibly lead to a loss of the membrane permeability and fouling resistance. The results of the ageing tests are summarized in Table 5.

This study allowed a multi-criteria assessment of ten membranes for use in oil and gas PW treatment. All membranes achieved satisfactory deoiling with 88.6% average removal rate. The membranes showed variable performances in terms of permeability and resistance to fouling. Two ceramic membranes in SiC outperformed the other membranes with high permeability and resistance to fouling. No correlations of the membrane material, configuration or pore size were established with the performances achieved by the membranes.

These results validate SiC membranes as a robust secondary deoiling option for challenging and high temperature effluent. However, due to their high cost of ownership, ceramic membranes are not competitive for large flow applications.

The resistance to ageing and the life expectancy were assessed for three polymeric membranes. An ageing effect was observed on the mechanical and chemical properties of two samples. However, the resulting life expectancies of two polymeric membranes out of three are satisfactory and give a competitive advantage to polymeric membranes as a cost-effective tertiary deoiling step over ceramic membranes.