** This Chapter is Divided into 3 web pages **

 

DESIGN AND DEVELOPMENT OF THE SYSTEM -- Continued

Commercially Available Condenser/Heat Exchanger Designs

VOC recovery using condensation in industrial applications is typically conducted using a four step process (Dunn and El-Halwagi, 1994): 1) dehumidification, 2) heat integration, 3) VOC condensation in a condenser, and 4) distillation to separate VOCs (Figure 3.4). Dehumidification removes water vapor from the gas stream, via condensation, typically by reducing gas temperature to 274 K. This is necessary if TVOC condensation is conducted at temperatures below 273 K to prevent fouling in the condenser from frozen water vapor. The system designed for this research uses LN₂ as a refrigerant and a carbon adsorber pre-concentrator. The vaporized LN₂ emitted from the condenser can be recycled to the inlet of the carbon adsorber. The TVOC is thus desorbed in a pure N₂ carrier gas, and minimal water vapor is present. This reduces the dehumidification load before TVOC condensation.

Heat integration exchanges heat between the entering TVOC laden gas stream and the exit gas stream. This process lowers the gas stream temperature entering the condenser and therefore reduces the amount of refrigerant required. The vast majority of condensation will occur in the condenser. The refrigerant is injected in either a direct or indirect contactor to cool the condenser to the desired temperature and thus the desired TVOC outlet concentration. Distillation of the condensate can be used to separate the TVOCs for re-use. The selection of a distillation process is dependent on the composition of the condensate and the desired degree of TVOC separation (USEPA^b, 1991).

A wide variety of heat exchange condenser designs are available for VOC condensation (Jacobs and Nadig, 1987; Wilbur, 1985; Perry and Green, 1984; Kern, 1950). The solid curtain condenser and the jet condenser are two common direct contact condenser designs that utilize a liquid refrigerant (Jacobs and Nadig, 1987). Most designs incorporate a falling refrigerant with the VOC laden gas stream flowing up the condenser counter-current to the refrigerant (Figure 3.5).

A common type of indirect-contact condenser is the shell-and-tube design (Wilbur, 1985). Condenser-surface areas from under 1 m² to 30,000 m² are available. Construction specifications for shell-and-tube designs can be obtained from the Tubular Exchangers Manufacturers Association (TEMA) Standards (TEMA, 1978). A multitude of design variations exist for these designs depending on the specific application criteria (e.g. minimize pressure drop, maximize surface area or maximize residence time). A simple 1-2 exchanger has one shell pass and two tube passes (Figure 3.6).

The selection of tube-side or shell-side refrigerant flow is dependent on the variables that are to be optimized. As a general rule, selection can be based on balancing the residence times of the flows as to maximize the heat transfer times (Kern, 1950). Other considerations include condensation surface area differences between the shell and tube, heat loss to the outer shell wall, condenser cleaning and extraction of the condensate (Wilbur, 1985). Selection of flow direction is also important in optimizing condenser performance (Perry and Green, 1984). Typical flow patterns include co-current, countercurrent, reversed, cross-flow and mixed flow orientations. When faced with the choice between co-current and countercurrent flow selection for condensation, there is a thermal disadvantage in selecting a co-current flow (Kern, 1950). The minimum attainable VOC temperature (for a condenser of infinite surface area) is the outlet temperature of the refrigerant. In countercurrent flow the VOC can attain temperatures closer to the inlet refrigerant temperature (Figure 3.7). For other flow selections, Perry and Green (1984) provide detailed design parameters to determine flow selection.

Refrigerants

Another condensation design criteria is the selection of an appropriate refrigerant. Refrigerant selection can be based on achievable temperature range, thermodynamic properties, cost, ease of use, handling and required auxiliary equipment (Dunn and El-Halwagi, 1994; USEPA^b, 1991; Wilbur, 1985; Perry and Green, 1984). The VOC removal efficiency and thus VOC recovery is based on the condenser temperature as outlined in section 3. Lower condenser temperatures will result in greater removal efficiencies and more product recovery. Temperature is often the primary refrigerant selection criteria for this reason (USEPA^b, 1991).

Selection of a refrigerant based on the desired outlet concentration is dependent on the refrigerant temperature range achievable. By knowing the desired effluent concentration of the vapor, the required condenser temperature can be determined from the Wagner equation. A refrigerant is selected that can achieve this temperature (Table 3.5).

A common refrigerant used in pollutant removal condensers is water (Wilbur, 1985). Water is inexpensive and easy to handle. However, because the condenser temperature is limited by the refrigerant temperature, cooling water results in low mass removal efficiencies for many TVOCs (Table 3.6). Mass removal efficiencies were determined for a 10% volume concentration TVOC stream to the equilibrium saturation vapor concentration. For applications where the process stream needs to be cooled below ambient temperatures, the use of cooling water typically requires auxiliary equipment to chill the water prior to use.

Ethylene glycol and water mixtures are also commonly used refrigerants (ASHRAE, 1983). Lower operating temperatures can be achieved with this mixture than for pure water, thereby lowering the effluent TVOC concentration and recovering more condensate (Table 3.6). Ethylene glycol water mixtures are exclusively used with indirect contact condensers to prevent ethylene glycol losses to the effluent gas stream (ASHRAE, 1983). Auxiliary equipment is needed to cool the mixture to temperatures below ambient temperature.

Using LN₂ as a refrigerant can provide a wide range of condenser temperatures by controlling the LN₂ flow rate delivered to the condenser. The ability to lower condenser temperature near the freezing point of TVOCs results in theoretical removal efficiencies as high as 99+% (Table 3.6). Furthermore, because LN₂ undergoes a phase change in the condenser, both the enthalpy of vaporization and sensible heat provide cooling capacity. The use of LN₂ as a refrigerant generally requires a vacuum jacketed storage vessel and well insulated or vacuum jacketed delivery lines. However, auxiliary cooling equipment is not necessary, as the refrigerant is available in liquid form from commercial sources. LN₂ can be used in direct or indirect contactors. After the LN₂ passes through the condenser, the gaseous N₂ refrigerant can be used as a blanket gas in process streams or as a purge stream during desorption. Using the gaseous N₂ refrigerant as a purge stream during the adsorber desorption cycle helps reduce moisture levels, normally present in air purge streams, that may foul the condenser and helps prevent explosive hazards.

Below 273 K, condenser fouling may occur from frozen water vapor. In high humidity process streams, two condensers can be used in series. The first condenser can operate at temperatures above the freezing point of water to remove water vapor from the stream. The second condenser can then be operated at temperatures below the freezing point of water to achieve greater removal efficiencies of the organic and reduce condenser fouling from frozen water vapor. The use of LN₂ can maintain low condenser temperatures and eliminates the need for a dehumidification process. Condensation is a recoverable type of control technology that provides a method to recover TVOCs for re-use. The amount of condensate recovered is dependent on the temperature of the condenser. Lower condenser temperatures result in higher removal efficiencies and more recovered TVOC. The Wagner equation provides a method to determine the TVOC equilibrium saturation concentration dependence on temperature. A variety of condenser designs and refrigerants are available for implementation for condensation of VOCs. The selection of an appropriate condensation system is dependent on many factors including the inlet gas characteristics and the desired outlet TVOC concentrations.

Design of the Laboratory-Scale ACFC System

Design steps of the laboratory-scale adsorption/regeneration/recovery system were identifying TVOC target adsorbates and concentration levels in typical effluent streams of full-scale facilities and selection of suitable adsorbent materials, adsorber type, regeneration method, condensation method, mechanical design of adsorber and condenser, and instrumentation design of the process. Development steps of the system were: fabrication of the adsorber and condenser, development of the gas generation unit, and assembly of the units and integration with analytical and measurement instruments.

The bench-scale ACFC adsorption-cryogenic vapor recovery system is presented in Figure 3.8. The system is composed of the units for sample gas generation, the ACFC fixed bed, cryogenic recovery, and analytical measurement. The calibrated gas stream passes through the fixed bed of ACFC, where the organic material is separated from the carrier gas by adsorption. The exhaust gas from the fixed-bed is analyzed continuously for the vapor concentration to make sure that breakthrough has not occurred. Then the vapor free carrier gas is vented under an exhaust hood.

Breakthrough for this project is defined as the condition when the concentration of the organic contaminant in the effluent reaches 5% of the influent concentration. After breakthrough of TVOC from the fixed bed, pure N₂ gas is passed through the adsorption bed and electrical power is supplied to the ACFC. Electrothermal energy regenerates the adsorption capacity of the ACFC and provides a N₂ gas stream containing concentrated desorbed TVOC. The TVOC concentration in the N₂ carrier gas is controlled by the supplied electrical power and the flow rate of the carrier gas. The concentrated vapor stream is then directed to the custom-designed shell-and-tube cryogenic condenser where the TVOC is condensed on the condensers internal cold surfaces. The condensed TVOC is transferred from the bottom of the condenser into an Erlenmeyer flask.

Sample Gas Generation Unit

The gas generation unit is shown in Figure 3.7. Ultra-high purity (UHP) N₂ gas passes through a purifier desiccant (Drierite, model L68GP) where any trace water and organic vapors are removed. Then the purified nitrogen is branched to two separate streams controlled by two mass flow controllers (Tylan models FC-260 and FC-280). One of the N₂ gas streams passes through two fritted glass bubblers connected in series and immersed in a temperature controlled water bath. The bubblers contain the liquid TVOC, causing the N₂ gas stream to become saturated in TVOC vapor. The saturated gas stream is then mixed with the second pure N₂ gas stream to produce a gas mixture with a desired TVOC concentration and total flow rate.

The gas generation system was calibrated using draw samples and a gas chromatograph/mass spectrometer (GC/MS, Hewlett Packard GC 5890 and MS 5971). In the calibration steps the bed was by-passed. The first calibration point was for UHP N₂ and the second for a Matheson calibrated standard TVOC gas mixture. A linear relationship was assumed between the MS counts and the related vapor concentration values. The calibration procedure was checked using another GC (Hewlett Packard Model 5880A) equipped with a flame ionization detector (FID). The FID signal was calibrated, by passing a continuous stream of the Matheson calibrated standard gas mixture through the GC/FID. A linear relationship was assumed between the FID zero signal and the signal of the standard gas. The concentration of generated gas was stable: at a near isothermal lab condition (+1EC), during 3 hr, a linear concentration decrease of less than 2.5% was observed for MEK and acetone streams of 5,000 and 10,000 ppmv, respectively, at a flow rate of 5 slpm.

Analytical Gas Measurement

As mentioned above, a GC/MS (Hewlett Packard GC 5890 and MS 5971) and a GC/FID (Hewlett Packard 5880A) were used for gas detection and measurement purposes. The concentration of gas mixture in the ACFC system was measured at several sampling ports. Locations of these sampling ports and the continuous sampling system of GC/FID are shown in Figure 3.7. A 30 ml/min sample gas stream, controlled by a factory calibrated Tylan mass flow controller, was drawn by a Mini Dia-Vac gas pump and directly forced through the GC/FID. Since the FID signal was dependent on the gas flow rate, any fluctuations in the flow rate could produce noise on the FID signal. To reduce the signal noise generated by the pump, the GC was equipped with a 0.4% Carbowax stainless steel packed column. The column was 1/8" in diameter and 12 feet in length. The sample effluent from the column was mixed with hydrogen and air and the mixture was burned in the FID. In a FID, the extent of ionization depends on the nature of the compound and the temperature of the flame. In our set-up with a 30 ml/min gas sample, the maximum FID sensitivity was obtained by mixing the sample with 30 ml/min of hydrogen. The balance of the gas entering the FID consists of 400 ml/min clean, dry air. These gas flows were checked using a Bios Dry-Cal instrument and regulated by mass flow controllers located within the GC. The FID ionization signal was amplified and plotted as a function of time on an attached GC console terminal. As with the gas generation system, the concentration measurements using the GC/MS and GC/FID involved the calibration procedures described in the previous section.

Design of Adsorber

Factors that are important in the design of the adsorber include choosing the type of adsorber, regeneration, TVOC and adsorbent, and the concentration and flow rate of TVOC. A fixed-bed was selected as the type of adsorber for this research. Target adsorbates were selected according to industrial use, governmental regulations and cost effectiveness of adsorption technology as discussed in the previous sections. ACFC was selected as the adsorbent material for this project. Electrothermal regeneration was selected as the type of regeneration. The following subsections describe the procedure for selection of ACFC as the adsorbent type and its utilization in the design of fixed-bed.

Selection of Adsorbent

Selection of the right adsorbent plays an important role on the performance of the adsorption separation process. ACFC was selected as a superior adsorbent for TVOC recovery. ACFC is made of woven activated carbon fibers (ACFs), and is an efficient adsorbent to remove TVOCs from gas streams. ACFs are made by carbonizing and activating fibers of phenolic resin, polyacrylonitrile (PAN), viscose, pitch, or rayon. Compared with activated carbon pellets (ACPs), granular activated carbons (GACs) or powder activated carbons (PACs), ACFCs have a higher specific surface area for a higher specific micropore volume which results in a higher contact efficiency. ACFCs have faster adsorption and desorption rates due to their microporous structure (there are no macropores, no or little mesopore volume, and uniform distribution of micropores from the external surface to the core of fiber) and small fiber size (Ermolenko et al., 1992; Fuji, 1994). ACFCs rapid adsorption and desorption rates can reduce the required process cycle time and increase the bed adsorption capacity by reducing the length of mass transfer zone (MTZ). ACFC can be installed in different configurational forms inside fixed beds providing desired adsorption and desorption performance as well as specific pressure drop (Pa/g adsorption). Suitability for in-situ electrothermal regeneration and easy handling are other advantages of ACFCs. Examples of commercially available ACFCs and ACFs are listed in Table 3.7.

ACFC samples for this project were obtained from American Kynol, a subsidiary of Nippon Kynol, located in Pleasantville, NY. Kynols ACFCs are made of novoloid fibers (polymerized cross-linked phenolic-aldehyde fibers). Novoloid fibers have an amorphous network structure containing 76% carbon, 18% oxygen, and 6% hydrogen (Hayes, 1981). These fibers are woven by conventional textile techniques to produce a novoloid cloth. The cloth is then carbonized and activated in an O₂ free atmosphere using steam or CO₂ at 900 EC in a one step process to produce ACFC. Pore volume and pore size of the ACFC increase with increasing duration of activation.

Kynol ACC-5092 (150 g/m² areal density) and ACC-519 (250 g/m² areal density) samples were identified as suitable ACFC adsorbents for the experiments. ACC-5092 showed much better structural stability than ACC-519. Furthermore, ACC-5092 was selected due to its high surface density (150 g/m²) and its structure stability. Specific surface area, total micropore volume, average micropore size, and elemental analysis of ACC-5092 are given in Table 3.8.

Adsorption capacities of ACC-5092-15, ACC-5092-20 and ACC-5092-25 were determined from Freundlich isotherms using the coefficients given by Cal (1993 and 1995). Modeled adsorption capacity of these ACFCs for 10,000 ppmv acetone were 442, 589 and 364 mg/g, respectively. ACC-5092-20 was thus selected as the preferred adsorbent due to its higher adsorption capacity. The higher adsorption capacity of ACC-5092-20 is mainly due to the existence of a high specific pore volume and high volume ratio for pore widths in the supermicropore and transitional ranges between supermicropores and mesopores. SEM (Scanning electron microscopy) micrographs of the ACFC sample are provided in Figure 3.9. The cumulative pore size distribution of the ACFC sample in comparison with some other adsorbents is given in Figure 3.10.

The adsorption characteristics of the ACC-5092-20 sample were measured by the Brunauer-Emmett-Teller (BET) adsorption method using a Micromeritics ASAP 2400 surface area analyzer. The surface areas were determined from the calculation of a monolayer adsorption using the BET equation at the relative pressure of 0.01 to 0.25. These measurements were necessary to insure homogeneity of the ACFC samples within the same lot and reproducibility of experimental results. For example, the BET surface area of two ACC-5092-20 samples from the same lot were measured to be 1592 and 1603 m²/g (Kynols reported nominal value was 1600 " 100 m²/g). The micropore volume was determined to be 0.69 cc/g and 0.72 cc/g. Quantified physical characteristics of ACC-5092-20 are listed in Table 3.9.

Click Here To Continue